Screen touch optimization method, device and equipment applied to dexterous hand, medium and product

Abstract

The embodiment of the application provides a screen touch optimization method applied to a dexterous hand, which can be applied to the field of artificial intelligence technology. The screen touch optimization method applied to the dexterous hand comprises the following steps: in a target touch operation process of the dexterous hand, current force sensing data transmitted in a second transmission period is received based on a preset bidirectional communication rule; the current working state of a switch state machine or the down pressure amount of the dexterous hand is adjusted to a back position state according to the current force sensing data and a preset force control threshold range; and the screen touch optimization method applied to the dexterous hand is realized based on at least one of the down pressure amount and the back position state. The embodiment of the application further provides a screen touch optimization device applied to the dexterous hand, equipment, a storage medium and a program product.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, specifically to the field of dexterous hand control technology and multimodal perception fusion technology, and more specifically to a screen touch optimization method, device, equipment, medium and product applied to dexterous hands. Background Technology

[0002] With the development of human-computer interaction technology, the application of humanoid dexterous hands in service robots, remote operation, and special operations is becoming increasingly widespread, and they are able to reproduce increasingly complex human hand movements. For example, in the operation of touch screen displays on electronic devices, typing and touch screen operation (such as watching short videos) are among the most delicate and complex movements of the human hand. They not only require the fingers to have high-frequency independent movement capabilities, but also require precise force control (to avoid damaging or scratching the screen).

[0003] However, existing dexterous hand control schemes for complex hand motion reproduction mostly rely on inverse kinematics calculation or pure position teaching reproduction, which have the following obvious defects: (1) Lack of tactile feedback: Traditional schemes only control joint angles and cannot sense contact force, resulting in excessive force damaging the equipment or insufficient force failing to trigger the touch screen, and stiff and unresponsive touch screen sliding; (2) Limited communication bandwidth: After the introduction of force sensors, the amount of data surges, and traditional communication protocols are difficult to complete real-time synchronization of all states (position, speed, torque, force, temperature) under limited bandwidth. Therefore, it will further cause a series of problems such as stiff interaction and high risk, poor continuous operation capability, low communication efficiency and weak scene adaptability. Summary of the Invention

[0004] In view of at least one of the above problems, embodiments of the present invention aim to provide a screen touch optimization method, device, equipment, medium and product for dexterous hands, thereby constructing an efficient, real-time dexterous hand control scheme with multimodal perception and fusion capabilities and the ability to accurately simulate human strength and thermal adaptability, and constructing a general state machine framework that supports multiple scenarios such as typing and browsing TikTok, so as to realize human-like, safe and efficient fine operation of dexterous hands in the screen touch process.

[0005] One aspect of the present invention provides a screen touch optimization method for a dexterous hand, comprising: during a target touch operation of the dexterous hand, receiving current force sensing data transmitted in a second transmission cycle based on a preset bidirectional communication rule; adjusting the pressure amount of the dexterous hand or switching the current working state of the state machine to a return state based on the current force sensing data and a preset force control threshold range; and implementing screen touch optimization for the dexterous hand based on at least one of the pressure amount and the return state.

[0006] According to an embodiment of the present invention, during the target touch operation of a dexterous hand, the current force sensing data transmitted in the second transmission cycle is received based on a preset bidirectional communication rule, including: receiving the current force sensing data transmitted in the second transmission cycle according to a preset force sensing frame format conforming to the preset bidirectional communication rule and a preset dual timing mechanism.

[0007] According to an embodiment of the present invention, a preset force sensing frame format is used to define the data transmission format of the frame header, thumb pressure value, index finger pressure value, middle finger pressure value, ring finger pressure value, little finger pressure value, and check bit.

[0008] According to an embodiment of the present invention, in adjusting the downward pressure of the dexterous hand or switching the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range, the method includes: determining the downward pressure of the dexterous hand based on the force sensing deviation determined by the current force sensing data and the preset force control threshold range, combined with a preset force sensing deviation coefficient, for adjusting the downward pressure of the dexterous hand fingers on the target screen based on the downward pressure and a first preset limit value while maintaining the current working state of the state machine.

[0009] According to an embodiment of the present invention, in adjusting the downward pressure of the dexterous hand or switching the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range, the method includes: determining the downward pressure of the dexterous hand based on the force sensing deviation determined by the current force sensing data and the preset force control threshold range, combined with the contact stiffness coefficient, for adjusting the downward pressure of the dexterous hand fingers on the target screen based on the downward pressure and a second preset limit value while maintaining the current working state of the state machine.

[0010] According to an embodiment of the present invention, in adjusting the downward pressure of the dexterous hand or switching the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range, the method further includes: when switching the current working state of the state machine to the return state, calling a preset return state execution sequence based on the current force sensing data and the preset force control threshold range to perform return control of the dexterous hand to the starting position state.

[0011] According to an embodiment of the present invention, in adjusting the dexterity hand's downward pressure or switching the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range, the method further includes: responding to the received current joint temperature data, current preliminary temperature data, and preset warning temperature range, adjusting the dexterity hand's downward pressure or switching the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range, and then switching the current working state of the state machine to a deceleration state or a heat dissipation state based on the downward pressure or the return state.

[0012] Another aspect of the present invention provides a screen touch optimization device for a dexterous hand, comprising a second data receiving module, a second state switching module, and a second state execution module. The second data receiving module is used to receive current force sensing data transmitted in a second transmission cycle based on a preset bidirectional communication rule during a target touch operation of the dexterous hand. The second state switching module is used to adjust the downward pressure of the dexterous hand or switch the current working state of the state machine to a return state based on the current force sensing data and a preset force control threshold range. The second state execution module is used to implement screen touch optimization for the dexterous hand based on at least one of the downward pressure and the return state.

[0013] Another aspect of the present invention provides an electronic device including one or more processors and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors perform the above-described screen touch optimization method for dexterous hands.

[0014] Another aspect of the present invention provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, cause the processor to perform the above-described screen touch optimization method for dexterous hands.

[0015] Another aspect of the present invention provides a computer program product including a computer program that, when executed by a processor, implements the above-described screen touch optimization method for dexterous hands.

[0016] The screen touch optimization method for dexterous hands provided in this invention can at least partially solve the problems of lack of haptic feedback, neglect of thermal management, and limited communication bandwidth in screen touch processes in dexterous hand control-related technologies, and thus can achieve at least one of the following technical effects:

[0017] (1) Fine-grained interaction strategy of force-position hybrid control: A closed-loop control algorithm based on thin-film pressure sensor / six-dimensional force sensor is proposed. For example, at the moment of contact (such as touch screen control), the system automatically switches from "position control mode" (normal working state) to "force-position hybrid control mode" (returning state), and finely adjusts the joint angle (downward pressure) according to the preset threshold (preset force control threshold range, such as F_min and F_max), which ensures effective triggering and avoids hard collision damage to the screen. That is, it can effectively solve the problem of stiff and high risk of screen touch interaction, and effectively avoid screen damage or scratches in the case of "blind typing" and "blind swiping".

[0018] (2) Multi-degree-of-freedom decoupling and multi-dimensional mapping control strategy: For 16-degree-of-freedom dexterous hands, the joint mapping algorithm is optimized. It not only supports position mapping, but also adds the inverse mapping of torque command and sensor data (the upward movement of the current preliminary temperature data and the downward movement of the joint motor torque command), realizing the logical decoupling and coordination of the thumb Pitch / Yaw / Roll / Tip with the other four fingers.

[0019] (3) Dual-channel CAN bus concurrent communication architecture (retained and expanded): Based on the original 20ms dual timer mechanism, the CAN protocol frame definition is expanded. Frame 0x06 (current force sensing data uplink) and frame 0x07 (current joint temperature data) are added. While ensuring low-latency control, high-frequency backhaul of multimodal sensing data (10ms cycle) is achieved, constructing a complete sensing-control closed loop. Therefore, this can effectively solve the problem of low communication efficiency. Even when facing high-dimensional data of "16 degrees of freedom + 5 force sensors + 5 temperature sensors", it can effectively complete efficient and high-precision bidirectional synchronization within a very short cycle (20ms).

[0020] Therefore, the screen touch optimization method for dexterous hands described above in this embodiment of the invention, while maintaining the original high real-time performance of 20ms, endows dexterous hands with "tactile sensation" and "body temperature sensation." In typing scenarios, the touch success rate is increased to over 99% with no risk of mechanical damage; in new scenarios such as browsing TikTok, it can perfectly reproduce the smoothness and pressure changes of human swipes; at the same time, the system has self-health monitoring capabilities, increasing continuous working time by more than 50%, significantly enhancing the robot's safety and human-like interactive experience.

[0021] It should be understood that the above general description and the following specific embodiments are merely exemplary and illustrative, and do not limit the scope of the invention. Attached Figure Description

[0022] The above-described features, other objects, and advantages of the present invention will become clearer from the following description of embodiments of the invention with reference to the accompanying drawings, in which:

[0023] Figure 1 The illustration schematically depicts an application scenario of a screen touch optimization method, apparatus, device, medium, and program product for dexterous hands according to embodiments of the present invention;

[0024] Figure 2A A flowchart illustrating a screen touch optimization method for dexterous hands according to an embodiment of the present invention is shown schematically.

[0025] Figure 2B A flowchart illustrating a screen touch optimization method for dexterous hands according to another embodiment of the present invention is shown schematically;

[0026] Figure 3 This illustration schematically depicts an application scenario of a screen touch optimization method for dexterous hands according to an embodiment of the present invention.

[0027] Figure 4A This schematically illustrates a structural block diagram of a screen touch optimization device for a dexterous hand according to an embodiment of the present invention;

[0028] Figure 4B A schematic diagram illustrates a structural block diagram of a screen touch optimization device for a dexterous hand according to another embodiment of the present invention; and

[0029] Figure 5 A block diagram of an electronic device adapted to implement a screen touch optimization method for dexterous hands, according to an embodiment of the present invention, is illustrated.

[0030] The accompanying drawings mentioned above are part of the specification of embodiments of the present invention, illustrating exemplary embodiments of the invention. The drawings, together with the description in the specification, serve to illustrate the principles of the embodiments of the present invention. It should be understood that the above general description with reference to the drawings and the following detailed description are merely exemplary and illustrative, and do not limit the scope of the invention. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the spirit of the contents disclosed in the present invention will be clearly explained below with reference to the accompanying drawings and detailed description. After understanding the embodiments of the present invention, any person skilled in the art can make changes and modifications based on the technology taught in the present invention without departing from the spirit and scope of the present invention.

[0032] The illustrative embodiments and descriptions of the present invention are used to explain the invention, but are not intended to limit the invention. Furthermore, elements / components using the same or similar reference numerals in the drawings and embodiments are used to represent the same or similar parts.

[0033] The terms "first," "second," etc., used in this invention do not specifically refer to any order or sequence, nor are they intended to limit the invention; they are merely used to distinguish elements or operations described using the same technical terms.

[0034] The directional terms used in this invention, such as up, down, left, right, front, or back, are merely for reference to the accompanying drawings. Therefore, the directional terms used are for illustrative purposes and not for limiting the scope of this invention.

[0035] The terms “comprising,” “including,” “having,” “containing,” etc., used in this invention are all open-ended terms, meaning that they include but are not limited to.

[0036] The term "and / or" as used in this invention includes any or all combinations of the things mentioned.

[0037] In this invention, "multiple" includes "two" and "more than two"; in this invention, "multiple groups" includes "two groups" and "more than two groups".

[0038] The terms "approximately," "about," etc., used in this invention are intended to modify any quantity or error that may vary slightly, but these slight variations or errors do not change the essence of the quantity or error. Generally, the range of slight variations or errors modified by such terms may be 20% in some embodiments, 10% in some embodiments, 5% in some embodiments, or other values. Those skilled in the art should understand that the aforementioned values ​​can be adjusted according to actual needs and are not limited thereto.

[0039] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0040] When expressions such as "at least one of A, B, and C" are used, they should generally be interpreted in accordance with the meaning commonly understood by a person skilled in the art (e.g., "a system having at least one of A, B, and C" should include, but is not limited to, systems having A alone, having B alone, having C alone, having A and B, having A and C, having B and C, and / or having A, B, and C, etc.). When expressions such as "at least one of A, B, or C" are used, they should generally be interpreted in accordance with the meaning commonly understood by a person skilled in the art (e.g., "a system having at least one of A, B, or C" should include, but is not limited to, systems having A alone, having B alone, having C alone, having A and B, having A and C, having B and C, and / or having A, B, and C, etc.). A person skilled in the art should also understand that any conjunction and / or phrase that substantially arbitrarily indicates two or more optional items, whether in the specification, claims, or drawings, should be understood to indicate the possibility of including one of these items, either of these items, or both items. For example, the phrase “A or B” should be understood as including the possibility of “A” or “B”, or “A and B”.

[0041] In existing technologies, dexterous hands typically employ methods such as pure position closed-loop control, single communication link, open-loop thermal protection, and hard-coded action sequences for screen touch control.

[0042] (1) Pure position closed-loop control: The host computer directly sends the target angle of each joint, lacking the perception and adjustment of contact force. (2) Single communication link: Most systems use serial communication or a single CAN frame to transmit all data, which limits the data bandwidth and makes it difficult to support high-frequency transmission of multi-dimensional data such as "position + force + temperature". (3) Open-loop thermal protection: Hard overcurrent protection is only set at the bottom layer of the driver board, lacking system-level temperature trend prediction and dynamic speed adjustment strategies. (4) Hard-coded action sequence: The action logic is hard-coded in the bottom firmware, and the action parameters cannot be dynamically adjusted according to sensor feedback (such as insufficient pressing force).

[0043] Therefore, the existing technology for touch operation on the screen has at least the following problems: (1) Stiff interaction and high risk: lack of force control leads to "blind typing" and "blind swiping", which can easily damage or scratch the phone screen. (2) Poor continuous operation capability: without temperature adaptive mechanism, long-term operation (such as continuously watching videos for 10 minutes) is very likely to trigger overheat protection and cause task interruption. (3) Low communication efficiency: when facing high-dimensional data of "16 degrees of freedom + 5 force sensors + 5 temperature sensors", the existing protocol often cannot complete bidirectional synchronization within a 20ms cycle. (4) Weak scene adaptability: it is difficult to smoothly switch from "typing" scene to "watching Douyin" and other scenes that require complex force changes.

[0044] Existing technologies do propose a control method for a bionic robotic hand controller. This method primarily involves acquiring finger position, temperature, and pressure information from a position feedback module and an end effector module when a robotic hand finger is detected reaching a designated position. If these parameters meet preset thresholds, the robotic hand finger maintains its current action; otherwise, it refuses to execute control commands (such as finger retraction), thus performing human-like behaviors and operations to improve its adaptability. However, because this method does not involve joint temperature control, pressure adjustment during screen touch operation, or data communication, and its control precision falls far short of the scenario requirements of this invention, it still fails to address the problems of lack of tactile feedback, neglect of thermal management, and limited communication bandwidth in dexterity hand control technologies during screen touch operation.

[0045] In view of at least one of the above problems, embodiments of the present invention aim to provide a screen touch optimization method, apparatus, device, medium and product for dexterous hands, thereby constructing a dexterous hand control scheme that is efficient, real-time, multimodal sensing, and can accurately simulate human strength and thermal adaptability.

[0046] One aspect of an embodiment of the present invention provides a screen touch optimization method for a dexterous hand, comprising: responding to a received work operation command, performing a target touch operation on a target screen based on the starting position state of the dexterous hand and the current working state of the state machine; during the target touch operation of the dexterous hand, receiving current joint temperature data transmitted in a first transmission cycle and current preliminary temperature data transmitted in a third transmission cycle based on a preset bidirectional communication rule; switching the current working state of the state machine to a deceleration state or a heat dissipation state according to the current joint temperature data, the current preliminary temperature data and a preset warning temperature range; and performing screen touch optimization based on at least one of the deceleration state and the heat dissipation state.

[0047] Another aspect of the present invention provides a screen touch optimization method for a dexterous hand, comprising: during a target touch operation of the dexterous hand, receiving current force sensing data transmitted in a second transmission cycle based on a preset bidirectional communication rule; adjusting the pressure amount of the dexterous hand or switching the current working state of the state machine to a return state based on the current force sensing data and a preset force control threshold range; and implementing screen touch optimization for the dexterous hand based on at least one of the pressure amount and the return state.

[0048] Figure 1 The illustration schematically depicts an application scenario of a screen touch optimization method, apparatus, device, medium, and program product for dexterous hands according to embodiments of the present invention.

[0049] like Figure 1 As shown, application scenario 100 according to this embodiment may include terminal devices 101, 102, and 103, network 104, and server 105. Network 104 is used as a medium to provide a communication link between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables, etc.

[0050] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social media platform software, etc. (for example only).

[0051] Terminal devices 101, 102, and 103 can be various electronic devices with displays and web browsing capabilities, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0052] Server 105 can be a server that provides various services, such as a backend management server that supports websites browsed by users using terminal devices 101, 102, and 103 (for example only). The backend management server can analyze and process data such as received user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0053] It should be noted that the screen touch optimization method for dexterous hands provided in this embodiment of the invention can generally be executed by server 105. Correspondingly, the screen touch optimization device for dexterous hands provided in this embodiment of the invention can generally be located in server 105. The screen touch optimization method for dexterous hands provided in this embodiment of the invention can also be executed by a server or server cluster that is different from server 105 and capable of communicating with terminal devices 101, 102, 103 and / or server 105. Correspondingly, the screen touch optimization device for dexterous hands provided in this embodiment of the invention can also be located in a server or server cluster that is different from server 105 and capable of communicating with terminal devices 101, 102, 103 and / or server 105.

[0054] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0055] The following will be based on Figure 1 The described scene, through Figures 2A to 4B The screen touch optimization method for dexterous hands according to the disclosed embodiments is described in detail.

[0056] like Figure 2A As shown, one aspect of an embodiment of the present invention provides a screen touch optimization method for dexterous hands, which includes operations S211-S214.

[0057] In operation S211, in response to the received work operation command, a target touch operation is performed on the target screen based on the starting position state of the dexterous hand and the current working state of the state machine;

[0058] In operation S212, during the target touch operation of the dexterous hand, based on the preset two-way communication rules, the current joint temperature data transmitted in the first transmission cycle and the current preliminary temperature data transmitted in the third transmission cycle are received.

[0059] In operation S213, based on the current joint temperature data, the current preliminary temperature data, and the preset warning temperature range, the current working state of the state machine is switched to the deceleration state or the heat dissipation state.

[0060] In operation S214, screen touch optimization is performed based on at least one of the deceleration state and the heat dissipation state.

[0061] like Figure 2B As shown, another aspect of the present invention provides a screen touch optimization method for dexterous hands, which includes operations S221-S223.

[0062] In operation S221, during the target touch operation of the dexterous hand, current force sensing data transmitted in a second transmission cycle is received based on a preset bidirectional communication rule. In another embodiment of the invention, during the target touch operation on the target screen in operation S211, operations S221 to S223 can be further executed synchronously or separately. Therefore, the aforementioned operations S221-S223 and operations S211-S214 of this embodiment can also be executed separately or synchronously, and specific details will not be elaborated further.

[0063] In operation S222, the dexterity hand's downward pressure is adjusted or the current working state of the state machine is switched to the return state based on the current force sensing data and the preset force control threshold range.

[0064] In operation S223, screen touch optimization for dexterous hands is implemented based on at least one of the pressure amount and return state.

[0065] In this embodiment of the invention, the dexterous hand can be a robotic end effector designed to mimic the structure and function of the human hand, and is generally considered one of the key components of embodied intelligence. The target screen can be a touch-controlled display screen of various electronic devices (including smartphones, tablets, laptops, and desktop computers). In this embodiment of the invention, the screen touch operation of the dexterous hand can be a single dexterous hand (such as the left or right dexterous hand) touching the screen (e.g., holding a mobile phone with one hand to watch videos), or it can be a two-handed dexterous hand (such as the left and right dexterous hands) touching the screen (e.g., the left dexterous hand holding the mobile phone while the right dexterous hand's index finger collects screen touches). Figure 3 As shown, the right-hand dexterous hand 301 holds the mobile phone 302 with its thumb and the other four fingers, and performs touch operations on the mobile phone screen 302a through the movable part of its thumb, which can realize complex human-like fine operations such as watching videos.

[0066] Furthermore, the screen touch optimization method for dexterous hands provided in this embodiment of the invention can be based on the Robot Operating System (ROS) as a node architecture to achieve environment dependency. The ROS system can run on a Linux embedded platform and communicate through topics and services to achieve complex robot functions, such as fine motion control of dexterous hands, which will not be elaborated here.

[0067] Furthermore, in embodiments of the present invention, each fingertip of the dexterous hand (such as the pad of the thumb and the pad of the index finger, etc.) can be embedded with a micro-thin-film pressure sensor, a six-dimensional force sensor, and / or a six-dimensional torque sensor, through a method based on similar I... 2 Data communication protocols using C or SPI protocols are used to acquire these pressure sensing data, forming raw pressure sensing data. Thin-film pressure sensors can output pressure values, which, after quantization, range from 0 to 4095, corresponding to an actual contact force of 0 to 15 Newtons. Six-dimensional force / torque sensors can simultaneously acquire force components in three directions and torque components in three directions, providing more accurate force information.

[0068] Furthermore, in this embodiment of the invention, the motor windings of each joint of each dexterous hand or the key areas of the drive circuit board can be integrated with devices such as NTC thermistors (Negative Temperature Coefficient Sensors) or digital temperature sensors to acquire the working temperature of each joint and each key area, and output raw temperature data in degrees Celsius. Moreover, by acquiring joint position data through encoders built into each joint of the dexterous hand, the actual working angle value of each joint is output, constituting the raw joint position data of the dexterous hand.

[0069] Specifically, the raw pressure sensing data, raw temperature data, and raw joint position data can be further mapped using a data transmission format conforming to preset bidirectional communication rules (such as the bidirectional data communication rules defined by the CAN bus protocol). This allows for the generation of uplink current force sensing data, current joint temperature data, and current joint position data based on the preset bidirectional communication rules. In this embodiment, "mapping" can be understood as converting the raw data collected by the sensors into a format that can be transmitted via the CAN bus and placing it into a specified data frame according to a custom protocol. The entire mapping process consists of two stages: hardware acquisition (corresponding data transmission protocol interfaces, such as I2C or SPI interfaces) and software packaging (such as data packaging functions).

[0070] Taking the current force sensing data as an example, the system can fill the pressure values ​​of the five fingertips into the corresponding member variables of the sending structure, and then call the sending function of a data transmission protocol (such as CAN protocol) based on preset bidirectional communication rules to assemble these data into corresponding protocol frames. The first byte of these protocol frames can be a frame header of 0x06 to identify that this is a force sensing data frame, and the following five bytes store the quantized pressure values ​​of the five fingertips in sequence. After packaging, this frame of data is sent to the transmission bus through a preset controller. Similarly, the current joint temperature data and the current joint position data can also be identified and packaged in the same way, using frame headers of 0x07 and 0x01, respectively. Through the above two steps of hardware acquisition and software packaging, the sensor data is completely "mapped" to the data transmission bus that conforms to the preset bidirectional communication rules, realizing the real-time transmission of multi-source sensor data on the same bus.

[0071] The current force sensing data primarily serves three purposes: first, detecting whether a finger is in contact with an object; for example, the system determines contact when the pressure exceeds 0.5 Newtons. Second, controlling the pressing pressure within a certain range (e.g., 8 ± 0.1 Newtons) during touch operation, ensuring button activation while preventing keyboard damage. Third, using a light touch (e.g., 0.3 Newtons) for smooth swiping on the touchscreen to prevent screen scratches. Current joint temperature data can be primarily used to establish an active thermal protection mechanism. For example, when the joint temperature exceeds 45 degrees Celsius, a first-level warning is triggered, and the system automatically halves the movement speed to reduce heat generation; when the temperature exceeds 60 degrees Celsius, a second-level protection is triggered, forcing the system into a cooling state and cutting off motor power to prevent hardware damage. Current joint position data can be used to achieve closed-loop joint position control, ensuring that finger movements accurately reach the target position, while simultaneously feeding back the actual joint position to the host computer for status monitoring.

[0072] Those skilled in the art should understand that I²C (Inter-Integrated Circuit) is a half-duplex, synchronous two-wire serial communication protocol; SPI (Serial Peripheral Interface) is a full-duplex, synchronous high-speed serial communication protocol; and CAN bus (Controller Area Network) is a highly reliable, multi-master controlled asynchronous serial communication bus, which will not be elaborated further.

[0073] The operational instructions can be generated by the embodied intelligence (or the control "brain" for the dexterous hand) based on the actual interaction scenario, to control the dexterous hand to perform target action tasks. The target action task can be a touch operation on the screen, such as tapping the screen to unlock it, or continuously swiping the screen to switch display pages. The actual interaction scenario can include voice interaction between the user and the embodied intelligence (such as dialogue), or scenarios in which the embodied intelligence self-recognizes based on its existing knowledge and its own environment (such as detecting an incoming call and grabbing the phone to answer it).

[0074] The initial position can be the hand posture position of the dexterous hand just before it is controlled to perform the target touch operation, and can be used as the initial posture of the target touch operation. For example... Figure 3 As shown, the dexterous hand's posture when its thumb and the other four fingers hold the phone 302, with the movable part of the thumb positioned above the phone screen 302a, can be considered the starting position. Based on this starting position, when performing a target touch operation on the target screen, the other four fingers can remain stationary, while the movable part of the thumb (such as the thumb tip and adjacent joints) can press down to touch the target screen and perform a swipe operation, thus achieving a human-like screen-scrolling behavior. The target touch operation can be understood as the dexterous hand performing touch operations such as pressing down (e.g., single click or continuous double click, like typing) or swiping (short or long distance swiping in any direction, like watching videos) on the target screen in the starting position to achieve the target action task.

[0075] A state machine is a mathematical model used to describe the transitions and behavioral changes of a system between different states. In this embodiment of the invention, it can be applied to the aforementioned ROS system to switch the stationary operating states of a dexterous hand, enabling the dexterous hand to perform different target action tasks based on these different states. Therefore, as a finite state automaton, a state machine can be used to describe the transitions and behavioral changes of a system between different states. In the dexterous hand control process of this embodiment of the invention, the state machine can be used to manage various stages of various movements including the wrist, palm, and fingers. Each stage corresponds to a state, and transitions between states can be achieved based on the aforementioned force feedback, temperature feedback, and time conditions.

[0076] The preset bidirectional communication rules can be data communication preset rules that enable uplink and downlink parallel data transmission at the same time or within the same time period. For example, they can be implemented using improvements to related communication protocols such as CAN, and specific improvements can be referred to below. In this embodiment of the invention, uplink or downlink transmission of different data can be achieved through different transmission cycles.

[0077] By using a data transmission method that conforms to preset bidirectional communication rules, current joint temperature data, current force sensing data, and current preliminary temperature data can be received in real time, all conforming to the preset bidirectional communication rules. These three different types of data can be transmitted uplinked based on a first transmission cycle (e.g., 100 milliseconds), a second transmission cycle (e.g., 10 milliseconds), and a third transmission cycle (e.g., 20 milliseconds), respectively. The transmission cycle can be understood as the interval between the transmission of preceding data and adjacent subsequent data, and can be used to define the transmission frequency of specific data types. Therefore, a layered and periodically differentiated data transmission strategy can be implemented. For critical control commands or data (such as initial temperature data), a stable transmission cycle of 20 milliseconds or even less can be used to ensure continuous action execution or provide effective temperature feedback to key parts of the dexterous hand. For current force sensing data, a high-frequency uplink transmission cycle of 10 milliseconds or even less can be used, enabling millisecond-level response speeds to force feedback, ensuring more precise force control, and effectively preventing screen scratches or damage. Furthermore, for current joint temperature data, a low-frequency uplink transmission cycle of 100 milliseconds or even longer can be used, satisfying the temperature feedback requirements for thermal protection without excessively consuming bus bandwidth. Therefore, by employing a layered data transmission strategy with different cycles, actual data feedback needs can be effectively met while avoiding prolonged and excessive bus bandwidth consumption.

[0078] In this embodiment of the invention, the current force sensing data can be understood as the real-time feedback data of pressure detection at the fingertips of at least each finger of the dexterous hand, conforming to the data transmission format of the preset two-way communication rules at the current moment; the current joint temperature data can be understood as the real-time feedback data of temperature detection at least each joint of the dexterous hand, conforming to the data transmission format of the preset two-way communication rules at the current moment; and the current preliminary temperature data can be understood as the real-time feedback data of temperature detection in the key area or key position of the dexterous hand, conforming to the data transmission format of the preset two-way communication rules at the current moment.

[0079] In this embodiment of the invention, the current working state can be the working status or operating conditions of the dexterous hand at the current moment. The current working state can include at least one of the following: normal working state (i.e., normal state), pressure state, force control state, return state, deceleration state, and heat dissipation state.

[0080] The normal state can be understood as the working conditions under which a dexterous hand can be controlled to safely and reliably perform predetermined target actions. At a minimum, fingertip pressure normally triggers touch to achieve the target action without damaging the screen; joint temperature is normal, allowing for smooth execution of the target action without damaging the dexterous hand joints; and critical areas or locations of the dexterous hand (such as the motor windings or critical temperature control parts of the drive circuit board) are considered normal. Under normal conditions, multiple normal operating conditions, including temperature, force control, and communication, are met without any abnormalities. Specifically, in this normal state, the temperature of all joints is below 45 degrees Celsius, and no level one warning is triggered, indicating that the motor and driver are operating within a safe temperature range and will not be damaged by overheating. Additionally, when the fingers are in a suspended state, the pressure values ​​of all fingertips are close to zero, with no abnormal contact; or when the fingers are in a gripping state, the gripping force is stable within the target range of 2 to 3 Newtons, without over- or under-pressure. Regarding communication, CAN bus communication is normal, receiving one frame of sensor data every 20 milliseconds without frame loss or timeout. If no data is received for three consecutive cycles, it can be determined as a communication anomaly. Furthermore, in terms of motion execution, the deviation between the actual joint position and the target position is within the allowable range, the encoder feedback is normal, and there are no jamming or step loss phenomena. Only when all the above conditions are met is the system considered to be in a normal state and can perform fine operations such as typing and sliding.

[0081] The downward-pressing state can be understood as the state in which the dexterous hand's fingers move downwards from a hovering position towards the target screen surface. In this state, the dexterous hand control system can drive joint movement according to preset speed and angle commands, while continuously monitoring force sensor data. The function of this mode is to enable the finger to approach the target action more quickly. For example, when typing, the index finger starts moving downwards from 5 millimeters above the screen; during this process, the system is in the downward-pressing state. The downward-pressing state embodies the position control stage in "force-position hybrid control," creating conditions for subsequent force control intervention.

[0082] Force control state can be understood as the state in which the system adjusts the joint angle in real time (to maintain the target force) based on force feedback after the fingertip has touched or is about to approach the screen. In this state, the system no longer moves according to fixed angle commands, but instead performs closed-loop control based on the PID (Proportion Integral Differential) algorithm with the goal of maintaining the target contact force. The function of this mode is to achieve fine force control and prevent damage or scratches to the screen. The current force sensing data plays a role in this state. For example, after the fingertip touches the screen, the system continuously adjusts the pressing pressure to keep it around 8 Newtons, which is the force control state. During the target touch operation, the dexterous hand can remain in this force control state. In one embodiment of the present invention, the normal state may include at least one of the executable states such as the pressing state and the force control state.

[0083] The return state can be understood as the state where the finger is lifted from the contact position and returns to the starting position. In this state, the system quickly reverses its movement, causing the finger to detach from the screen. The function of this mode is to complete the completion of one action and prepare for the next action. The return state can be understood as the closing stage of the force-controlled adaptive logic. It is also triggered when the pressure exceeds the maximum safe force F_max, reflecting a safety protection mechanism. For example, after the screen is triggered, the finger is quickly lifted back to the hovering position; this lifting process is in the return state.

[0084] The deceleration state can be understood as the automatic slowdown of all movements (e.g., joint speed halved) when the temperature exceeds a certain threshold (such as the first temperature warning threshold). In this state, all movements are executed at a slower pace, and the rhythm of the movements is reduced. The function of this mode is to reduce heat generation and achieve thermal balance without stopping operation. The deceleration state is the core manifestation of the active temperature protection mechanism, realizing a leap from passive overheating shutdown to active deceleration thermal management. For example, if continuous typing causes the joint temperature to rise to 48 degrees Celsius (exceeding the first temperature warning threshold of 45 degrees Celsius), the system automatically reduces the typing speed from 90 degrees per second to 45 degrees per second, which is the deceleration state.

[0085] The cooling state can be understood as a state forcibly entered after the temperature exceeds a certain threshold (usually greater than the first temperature warning threshold). All joints slowly return to a cooling posture (such as fingers naturally spread, palm facing up or to the side), and the motor power supply is cut off. In this state, all joints slowly return to the cooling posture, and the motor enable can be further cut off. The function of this mode is to maximize heat dissipation and protect the hardware, preventing motor burnout or driver damage. The cooling state temperature active protection mechanism is the last line of defense, embodying the design concept of "active safety," and can be understood as a safe mode. For example, when the temperature reaches 60 degrees Celsius, the fingers automatically spread and stop moving, waiting for cooling down; this is the cooling state.

[0086] Therefore, the normal state, the pressure state, the force control state, and the return state can constitute the basic action sequence of touch screen interaction and serve as the carrier of force control adaptation; while the deceleration state and the heat dissipation state can be the specific manifestation of the active temperature protection mechanism, realizing the leap from passive overheating or even shutdown to active thermal management.

[0087] The preset warning temperature range can reflect the one-to-one correspondence between the detected feedback temperature value and the dexterity state and heat dissipation state when the dexterity hand is operating under abnormal temperature conditions. It can also define the one-to-one correspondence between the detected feedback temperature value and the dexterity-related data in a certain state (such as the dexterity state). Therefore, the preset warning temperature range can include specific state switching thresholds, or it can define a state execution change area based on two adjacent state switching thresholds. When a specific state switching threshold is met, the switching from one execution state to another can be directly realized. In the corresponding specific state execution change area, the execution parameters can be adjusted according to the change of the detected feedback temperature value.

[0088] For example, when the joint temperature exceeds 45 degrees Celsius (such as the first temperature warning threshold), the state machine switches the current working state to the deceleration state. When the joint temperature exceeds 60 degrees Celsius (such as the second temperature warning threshold), the state machine switches the deceleration state to the heat dissipation state. When the joint temperature is 50 degrees Celsius, it is in a specific state execution change region between the first and second temperature warning thresholds. In this state, the joint speed can be further reduced relative to the joint speed corresponding to 45 degrees Celsius. That is, the closer the temperature is to the second temperature warning threshold, the slower the joint speed (even until the joint speed is close to or equal to 0). The closer the temperature is to the first temperature warning threshold, the faster the joint speed. When the temperature value is less than the first temperature warning threshold, the joint speed can be restored to the normal working state.

[0089] Since the current joint temperature data is the temperature detection value for each joint of the dexterous hand, when the temperature value exceeds the first temperature warning threshold defined by the preset warning temperature range, the state machine can switch from the current normal working state to the deceleration state, performing the target action task at a lower joint speed. Correspondingly, since the current preliminary temperature data is the temperature detection value for the critical area of ​​the dexterous hand, when this temperature detection value exceeds the second temperature warning threshold defined by the preset warning temperature range, the state machine can quickly switch from the deceleration state to the heat dissipation state. Therefore, unlike traditional technologies or conventional design concepts, the determination of the heat dissipation state is usually based on the current preliminary temperature data and the preset warning temperature range to determine the switching timing, while the determination of the deceleration state is usually based on the current joint temperature data and the preset warning temperature range to determine the switching timing. In the deceleration state, when the joint speed drops to 0, the state machine can switch from the deceleration state to the heat dissipation state. Therefore, parallel state switching judgments can be achieved, thus maintaining the most accurate and efficient state switching without excessively consuming data transmission bandwidth. This ensures both effective heat dissipation to reduce risks and the ability to immediately stop in the face of real risks to prevent damage.

[0090] During the target touch operation of the dexterous hand, the current working state of the dexterous hand can be a force control state. The preset force control threshold range can be understood as a one-to-one correspondence between the detected feedback pressure value, the amount of pressure applied, and the return state under abnormal pressure working conditions. It can also define a one-to-one correspondence between the detected feedback pressure value and the amount of pressure applied in a certain state (such as the force control state). Therefore, the preset force control threshold range can include specific state switching thresholds or adjustment thresholds for the amount of pressure applied. In this embodiment of the invention, the touch force of each finger of the dexterous hand on the screen can be within a normal force control threshold range, ensuring touch operation while preventing situations where the screen is touched by a finger but receives no response. The normal force threshold range can be defined by the minimum trigger force F_min and the maximum safety force F_max. The force control threshold range between the minimum trigger force F_min and the maximum safety force F_max can be considered the normal force control threshold range. Pressure values ​​outside this normal force control threshold range fall into the preset force control threshold range, such as the force control threshold range below the minimum trigger force F_min and above the maximum safety force F_max. Within the normal force control threshold range (greater than or equal to the minimum trigger force F_min and less than or equal to the maximum safety force F_max), a target holding force F_target can be further set for each finger. This ensures that, in force control mode, the pressure of the corresponding finger on the target screen fluctuates around this pressure threshold, maintaining stable contact between the finger and the screen. The force control threshold range for each finger can differ, and its corresponding minimum trigger force F_min, maximum safety force F_max, and target holding force F_target can also be different.

[0091] In this embodiment of the invention, when the dexterous hand compares the current force sensing data of each finger with a preset force control threshold range, if the detected feedback pressure value of a certain finger is less than the minimum trigger force F_min of the preset force control threshold range for that finger, the downward pressure of that finger can be adjusted accordingly to increase the downward pressure, thereby ensuring that the finger triggers the screen. Simultaneously, the degree of adjustment of the downward pressure can be controlled according to the difference between the detected feedback pressure value of that finger and the minimum trigger force F_min (e.g., increasing the downward pressure). Specifically, a matching effect can be achieved through a one-to-one mapping relationship between the degree of downward pressure adjustment and this difference. During this downward pressure adjustment process, the state machine can maintain the execution of the force control state without switching.

[0092] In this embodiment of the invention, when the dexterous hand compares the current force sensing data of each finger with a preset force control threshold range, if the detected feedback pressure value of a certain finger is greater than the maximum safe force F_max of the preset force control threshold range for that finger, the state machine can directly switch from the force control state to the return state. In another embodiment, while keeping the force control state unchanged, the adjustment of the pressing amount can be directly controlled based on the difference between the detected feedback pressure value of the finger and the maximum safe force F_max, ensuring that the finger can immediately reduce the pressing amount. Therefore, it can effectively ensure that the finger's trigger operation avoids invalid touch, and can also immediately stop the pressing trigger return action to prevent damage or scratches to the screen.

[0093] Therefore, force and temperature sensors can be introduced as key criteria for state transitions on the basis of traditional state machine control. This allows for real-time adjustment of the pressure depth based on finger pressure feedback, achieving a human-like "touch-and-return" feel. Furthermore, temperature feedback can be used to dynamically adjust the typing frequency to prevent motor overheating. In scenarios like browsing TikTok, force control can provide a smooth "acceleration-stop" feel for screen swiping, while temperature control ensures stability during prolonged swiping. Moreover, this effectively solves the problem of stiff and risky touchscreen interaction, preventing screen damage or scratches during "blind typing" and "blind swiping." Furthermore, a highly adaptive temperature protection mechanism can be established, effectively improving the continuous operation capability of dexterous hands and directly preventing task interruptions caused by overheating protection during prolonged operation (such as watching videos continuously for 10 minutes). Finally, while ensuring low-latency control, high-frequency backhaul of multimodal sensing data (10ms cycle) is achieved, and a complete sensing-control closed loop is constructed. Therefore, the problem of low communication efficiency can be effectively solved. Even when facing high-dimensional data of "16 degrees of freedom + 5 force sensors + 5 temperature sensors", efficient and high-precision bidirectional synchronization can be effectively completed within an extremely short cycle (20ms).

[0094] like Figures 2A-3 As shown, according to an embodiment of the present invention, the process of receiving current joint temperature data transmitted in a first transmission cycle and current preliminary temperature data transmitted in a third transmission cycle based on a preset bidirectional communication rule includes:

[0095] Based on a preset temperature frame format that conforms to the preset bidirectional communication rules, the current joint temperature data transmitted in the first transmission cycle is received according to a preset dual timing mechanism.

[0096] Based on the preset torque frame format that conforms to the preset bidirectional communication rules, when transmitting the detected joint motor torque command downlink, the current preliminary temperature data transmitted uplink in the third transmission cycle is received according to the preset dual timing mechanism.

[0097] The first transmission cycle is longer than the third transmission cycle.

[0098] like Figures 2A-3 As shown, according to another embodiment of the present invention, during the target touch operation of a dexterous hand, based on a preset bidirectional communication rule, the current force sensing data transmitted in a second transmission cycle is received, including:

[0099] Based on a preset force sensing frame format that conforms to the preset bidirectional communication rules, the current force sensing data transmitted in the second transmission cycle is received according to a preset dual timing mechanism.

[0100] The first transmission cycle is longer than the second transmission cycle.

[0101] like Figures 2A-3 As shown, according to an embodiment of the present invention, a preset temperature frame format is used to define the data transmission format of the frame header, the temperature value of each joint of the dexterous hand, and the check bit, wherein the temperature value of each joint is implemented by offset encoding; the current preliminary temperature data includes the temperature value of at least one preset key position of the dexterous hand for emergency overheating judgment.

[0102] like Figures 2A-3 As shown, according to another embodiment of the present invention, a preset force sensing frame format is used to define the data transmission format of the frame header, thumb pressure value, index finger pressure value, middle finger pressure value, ring finger pressure value, little finger pressure value, and check bit.

[0103] In this embodiment of the invention, during the target touch operation process of the dexterous hand (such as the pressure state or force control state), configuration parameters can be read from the parameter server of the ROS system through periodic real-time dynamic loading. These parameters may include force control-related parameters such as the force threshold ForceThreshold which meets the preset force control threshold range, and temperature control-related parameters such as the first temperature warning line TempLimit_L1 (as mentioned above, the first temperature warning threshold) and the second temperature warning line TempLimit_L2 (as mentioned above, the second temperature warning threshold) which meet the preset warning temperature range.

[0104] Specifically, force control parameters include: force threshold ForceThreshold, minimum trigger force F_min, maximum safety force F_max, target holding force F_target, and preset force control parameters (such as various proportional coefficients Kp, integral coefficients Ki, and / or derivative coefficients Kd). In addition, temperature control parameters include: a primary temperature warning threshold TempLimit_L1 (such as the aforementioned first temperature warning threshold, e.g., 45 degrees Celsius) and a secondary temperature warning threshold TempLimit_L2 (such as the aforementioned second temperature warning threshold, e.g., 60 degrees Celsius), a primary deceleration coefficient TempSpeedRatio_L1 (e.g., can be set to 0.5 times), and a temperature sampling period (e.g., the first transmission period is set to 100 milliseconds, and the second transmission period is set to 20 milliseconds).

[0105] Regarding the configuration of the aforementioned force control and temperature control parameters, a "manual preset plus read-only during runtime" mode can be adopted. Specifically, all parameters can be pre-set through a configuration file and read once from the ROS system's parameter server upon system startup. During operation, the ROS system typically only reads and uses these parameters without automatically modifying or updating them. In other words, these parameters are usually statically configured; if a change in parameter value is needed, it can be achieved by modifying the configuration file and restarting the system. As mentioned earlier, temperature and force detection (such as the aforementioned current force sensing data and current joint temperature data) can be performed automatically in real time. In some embodiments of this invention, the ROS system can also periodically and repeatedly read parameter values ​​during operation to determine parameter validity and update parameters in a timely manner.

[0106] The ForceThreshold parameter sets the trigger threshold for when a dexterous hand's finger touches the screen. When the detected pressure exceeds this threshold, it determines that the finger has touched the target screen, and the state machine transitions from the pressure state to the force control state. Without this threshold, the system cannot accurately determine whether a finger has pressed the screen.

[0107] The minimum trigger force F_min can be used to ensure that touch screen operations can be reliably triggered. Different types of screens have different trigger forces. By setting an appropriate minimum trigger force F_min value, you can ensure that your finger can trigger the operation with minimal force, which saves effort and avoids invalid presses.

[0108] The maximum safety force F_max can be used to protect the target screen being operated on. When the detected pressure exceeds this upper limit, the system will immediately control the dexterous hand fingers to stop pressing down and trigger a return action to prevent scratching, damaging or crushing the phone screen.

[0109] The target holding force F_target is used to maintain a stable contact force between the finger and the target screen in force-controlled mode. The system can fine-tune the joint angle in real time through algorithms such as PID, so that the actual contact force always fluctuates around the target holding force F_target, thereby achieving human-like force control.

[0110] The first-level temperature warning threshold, TempLimit_L1, can be used to trigger preventative slowdown. When the current joint temperature of a certain joint exceeds the first-level temperature warning threshold, TempLimit_L1, the system can control the corresponding joint of the dexterous hand to slow down its movement speed, thereby reducing heat generation and preventing the temperature from rising further.

[0111] The secondary temperature protection threshold, TempLimit_L2 (i.e., the second temperature warning threshold), can be used to trigger emergency protection. When the initial temperature of a corresponding joint exceeds this secondary temperature warning threshold, TempLimit_L2, the system can control the corresponding joint of the dexterous hand to forcibly switch to the heat dissipation mode. At least the corresponding joint will slowly reset to the heat dissipation mode and cut off the motor power supply to prevent hardware damage. Typically, all joints can be controlled to reset to the heat dissipation mode and cut off the power supply.

[0112] Based on the data transmission protocol with preset bidirectional communication rules in the embodiments of the present invention, multiple types of frame data conforming to the preset data transmission format, including frame headers 0x01, 0x04, 0x05, 0x02, 0x06, and 0x07, can be provided, thereby more effectively supporting multimodal data.

[0113] The 0x01 frame can be used to transmit joint angle control commands for the first six joints of a dexterous hand. Specifically, it can include target angle values ​​for the six joints: thumb pitch, thumb yaw, index finger pitch, middle finger pitch, ring finger pitch, and little finger pitch. Each angle value can be represented by one byte, ranging from 0 to 255, corresponding to a joint's range of motion from 0 to 270 degrees.

[0114] The 0x04 frame can be used to transmit joint angle control commands for the last three joints and rotation angle control commands for the thumb. Specifically, it can include the joint angles of the index finger yaw, middle finger yaw, ring finger yaw, little finger yaw, and thumb roll. Due to the different number of degrees of freedom, for a 9-DOF dexterous hand, the 0x04 frame can carry only three joint angle data; for a 10-DOF dexterous hand, the 0x04 frame carries four joint angle data.

[0115] The 0x05 frame can be used to transmit joint movement speed commands for the five fingers, with each finger corresponding to a joint speed value. This allows control over the speed of movement of all joints within that finger. A higher speed value results in faster finger movements.

[0116] Frame 0x02 enables the transmission of torque command data in downlink and the transmission of preliminary temperature data in uplink. The torque command can be understood as the output torque limit value for each joint motor. Setting the torque upper limit prevents motor overload and also serves as a preliminary force limit. Furthermore, the uplink transmission of the preliminary temperature data can be understood as a new backhaul function added to frame 0x02, designed to simultaneously carry and transmit the preliminary temperature data of key locations or areas at the current moment in the uplink direction while transmitting the torque command in the downlink direction.

[0117] The so-called backhaul can be understood as using the 0x02 frame, originally used only for downlink transmission, to carry the current preliminary temperature data back to the host computer in the uplink direction. Specifically, the original function of the 0x02 frame is for the host computer to send torque commands downlink. The bidirectional data transmission characteristics defined by the preset bidirectional communication rules allow the same frame ID to be used for both downlink and uplink transmission. Therefore, the 0x02 frame still transmits torque commands in the downlink direction, but when the dexterous hand sends data to the host computer, it can use the same 0x02 frame ID for uplink data transmission, but the data content is changed to the current preliminary temperature data. Thus, without adding a new frame type, an additional channel for uplink transmission of temperature data is obtained, simultaneously achieving the transmission of two different types of data under the same frame type. In addition to the current preliminary temperature data, the 0x02 frame return field can also contain the following: the first byte is the frame header 0x02, used to identify that this is a return frame; the second to sixth bytes can store the temperature values ​​at five key locations as the current preliminary temperature value, typically the detection feedback temperature values ​​from the base drive circuits of the thumb, index finger, middle finger, ring finger, and little finger; the seventh byte can store the status flag bit to indicate whether there is an anomaly; and the eighth byte serves as a check bit. Therefore, the current preliminary temperature value can be used for preliminary overheating judgment, the status flag bit is used for rapid reporting of motor overload or sensor failure, and the check bit is used to ensure the reliability of data transmission.

[0118] The 0x06 frame can be dedicated to transmitting real-time force sensing data. Each frame can carry five fingertip pressure values ​​(quantized values ​​from 0 to 4095), and the second transmission cycle can be 10ms. Regarding the frame format, the preset force sensing frame format, taking the 0x06 force sensing data frame as an example, defines the following format: the first byte is fixed at 0x06 to represent the frame header; the second byte stores the thumb pressure value; the third byte stores the index finger pressure value; the fourth byte stores the middle finger pressure value; the fifth byte stores the ring finger pressure value; the sixth byte stores the little finger pressure value; and the seventh and eighth bytes are reserved or used as check bits. The real-time force sensing data transmitted in the 0x06 frame can be the contact pressure value collected by the thin-film pressure sensor installed on each fingertip. When a finger contacts the screen, the sensor detects the pressure and outputs an electrical signal proportional to the pressure magnitude. After analog-to-digital conversion, a quantized value between 0 and 4095 is obtained, corresponding to an actual contact force of 0 to 15 Newtons. Therefore, the current force sensing data can provide more precise force feedback, enabling the system to adjust the joint angle in real time based on the difference between the current contact force and the target force, so that the contact force is always kept within the preset range, thereby achieving human-like force control, specifically as mentioned above in the preset force control threshold range.

[0119] The 0x07 frame can be dedicated to transmitting high-precision current joint temperature data. Each frame can carry the temperature values ​​(in degrees Celsius) of 5-10 joints, with feedback detected according to a first transmission cycle of 100ms (i.e., a slower temperature detection frequency) or triggered on demand. Similarly, the format of the 0x07 temperature data frame (preset temperature frame format) can be defined as follows: the first byte is fixed at 0x07 to represent the frame header, and subsequent bytes sequentially store the temperature values ​​of each joint. The temperature values ​​use offset encoding, that is, adding 40 to the actual temperature, so that it falls within the range of 0 to 100, making it easy to represent with a single byte. The high-precision current joint temperature data transmitted by the 0x07 frame can be the temperature value inside the motor windings of each joint (or the critical area of ​​the drive circuit board), in degrees Celsius. Specifically, the temperature sensor can be installed inside the joint motor or near the power devices on the drive board to directly measure the temperature of the heat source as the current joint temperature data. The current joint temperature data can also be used to achieve precise thermal management. Unlike the preliminary temperature data transmitted in frame 0x02, the current joint temperature data in frame 0x07 has higher accuracy, specifically ±0.5 degrees Celsius, and covers a larger number of joints, typically five to ten joints whose temperatures can be uploaded in real time. The current joint temperature data function mainly includes two aspects: first, it is used for first-level early warning judgment, triggering a speed reduction state switch when the temperature of any joint exceeds the first-level temperature warning threshold TempLimit_L1 (i.e., the first temperature warning threshold); second, it is used for temperature trend analysis, predicting overheating risks by observing the rate of temperature rise in the current joint temperature data, and taking remedial measures such as speed reduction or heat dissipation in advance. In another embodiment of the invention, the current joint temperature data may also include the preliminary temperature data, which can also be used for second-level protection judgment, forcing entry into a heat dissipation state mode when the temperature exceeds the second-level temperature warning threshold TempLimit_L2 (i.e., the second temperature warning threshold).

[0120] In this embodiment of the invention, the downlink channels of frames 0x01, 0x04, 0x05, and 0x02 based on preset bidirectional communication rules are usually common frame functions. The extended protocol can use multiple frame types plus layered periods to realize the uplink and downlink transmission of more different types of data. For example, the frame data of frame headers 0x01, 0x04, and 0x05 can usually transmit control data such as joint angles and joint speeds, and can be transmitted based on a transmission period of 20 milliseconds. Frame 0x02 can realize the uplink transmission of the current preliminary temperature data during the downlink process of torque command data. Frame 0x06 can realize the uplink and downlink transmission of the current force sensing data, and frame 0x07 can realize the uplink and downlink transmission of the current joint temperature data.

[0121] Therefore, in this embodiment of the invention, while adding the uplink current preliminary temperature data in frame 0x02, dedicated force control sensing data frames (frame 0x06) and temperature control sensing data frames (frame 0x07) are also added. Specifically, frames 0x01, 0x04, and 0x05 can be used to transmit joint angle and joint speed commands, with a transmission period of 20 milliseconds. Frame 0x02 transmits torque commands downlink and uplinks preliminary temperature data back uplink, with a third transmission period of 20 milliseconds. Frame 0x06 can be specifically used to transmit real-time force sensing data (current force sensing data), carrying pressure values ​​from five fingertips per frame, with a high-frequency period of 10 milliseconds for the second transmission period. Frame 0x07 can be specifically used to transmit high-precision temperature data (current joint temperature data), carrying temperature values ​​from five to ten joints per frame, with a low-frequency period of 100 milliseconds for the first transmission period due to the slow temperature change.

[0122] In summary, based on the preset bidirectional communication rules of this invention, the transmission extension of current joint temperature data, current force sensing data, and current preliminary temperature data can be effectively realized. This allows at least four dimensions of data, such as joint position data, torque data, force-related parameters, and temperature-related parameters, to be transmitted in an orderly manner through the same data transmission channel, with each component performing its function without interfering with the others. The specific technical effects can be manifested as follows:

[0123] (1) Effectively expands from a single data type to a multimodal data type. Compared with the existing technology that can only transmit data of torque, position and velocity, the above method of this embodiment can expand the frame header type to five types, adding 0x06 force sensing data frame and 0x07 temperature data frame, realizing the leap from "single-dimensional control" to "multi-dimensional perception". This expansion is not a simple increase in quantity, but introduces force and temperature sensing into the control system of the dexterous hand, enabling the state machine to make comprehensive decisions based on force, temperature and position data.

[0124] (2) A differentiated periodicity hierarchical transmission strategy is introduced. Compared with the traditional approach in existing technologies where all types of data use the same transmission period during data transmission, this innovative approach utilizes a differentiated transmission period strategy based on frame headers to achieve different transmission frequencies for different types of frame data. For example, "position data maintains a 20ms control period to ensure continuous action, force data uses a 10ms high-frequency period to ensure force control response speed, and temperature data uses a 100ms low-frequency period." Therefore, it can effectively avoid the serious bandwidth waste in the traditional method (for example, temperature data changes slowly and does not require high-frequency transmission, but it occupies valuable high-frequency bandwidth), greatly reducing bandwidth usage. The priority and transmission frequency of data can be determined according to the frame header type. For example, frame 0x06 has the highest priority (the second transmission period is the smallest, i.e., the transmission frequency is the largest), and frame 0x07 has the lowest priority (the second transmission period is the largest, i.e., the transmission frequency is the smallest). Among them, the transmission priority of frame 0x02 can be between that of frame 0x06 and frame 0x07.

[0125] (3) Bidirectional multiplexing of frame IDs is achieved. Compared with the traditional method in the prior art where each frame data is usually transmitted in one direction (uplink or downlink) or bidirectional transmission of a single type of data (limited to only one type of data), the method of this embodiment can effectively realize bidirectional multiplexing design, that is, torque command is transmitted in the downlink direction and the current preliminary temperature data is transmitted back in the uplink direction. In this way, an additional temperature feedback channel can be obtained without increasing the number of frame IDs, which greatly improves the bus utilization efficiency. This bidirectional multiplexing can be achieved by presetting a transmission protocol based on preset bidirectional communication rules. For example, the receiving end can distinguish the meaning of the frame content according to the data source (whether it is sent by the host computer or by the dexterous hand).

[0126] (4) Data compression and multiplexing are achieved. Compared with the traditional method in the prior art where each frame usually carries only the data of one joint (16 frames are needed to transmit 16 degrees of freedom), the method described in this embodiment of the invention can compress the data of multiple joints into one frame. For example, frame 0x01 carries the angles of 6 joints at the same time, frame 0x06 carries the pressure values ​​of 5 fingertips at the same time, and frame 0x07 carries the temperature values ​​of 5 to 10 joints at the same time. This multiplexing greatly reduces the number of frames, so that 16 degrees of freedom full state synchronization can be completed in parallel even in a very short period of 20 milliseconds.

[0127] It should be further explained that, in this embodiment of the invention, the frame header of the aforementioned data frame can achieve at least three functions:

[0128] On the one hand, data types can be distinguished through frame headers, avoiding bus conflicts. For example, different data types use different frame headers, and the receiving end can identify the data type and perform corresponding parsing based on the frame header, without the need for additional handshakes or negotiations. Specifically, frames 0x01 and 0x04 are dedicated to joint angles, frame 0x05 is dedicated to speed commands, frame 0x02 is dedicated to torque commands and temperature feedback, frame 0x06 is dedicated to force sensing data, and frame 0x07 is dedicated to high-precision temperature data. This clear frame header definition allows multiple data types to coexist orderly on the same bus without interfering with each other.

[0129] On the other hand, priority scheduling can be implemented through frame headers to ensure the real-time performance of critical data. The system prioritizes data frames based on their header type. Force data frames (0x06) have the highest priority because force feedback requires millisecond-level response speeds to prevent damage to objects. Position and velocity frames (0x01, 0x04, 0x05) have the next highest priority to ensure timely issuance of control commands. Temperature data frames (0x07) have the lowest priority because temperature changes are relatively slow and can tolerate slightly larger delays. When the bus becomes congested, high-priority data frames are sent first, while low-priority data frames wait. This priority scheduling based on frame headers ensures that force data can be reliably uploaded within a 10ms cycle.

[0130] On the other hand, periodic configuration can be achieved through frame headers, supporting differentiated transmission frequencies. Specifically, the system can configure different transmission timers based on the frame header type. Frame 0x06 is sent every 10ms, frames 0x01, 0x04, 0x05, and 0x02 are sent every 20ms, and frame 0x07 is sent every 100ms. This frame header-based periodic configuration ensures that different types of data are transmitted at their most suitable frequencies, satisfying the real-time requirements of force control and temperature control while avoiding unnecessary bandwidth consumption.

[0131] The effect of "significantly reducing bandwidth usage" of the aforementioned data frame extension can be demonstrated through bandwidth calculations in the following example: Based on CAN bus 2.0 at a baud rate of 1 Mbps, the transmission time per frame is approximately 100 to 130 microseconds. Approximately 150 to 200 frames can be sent within 20 ms. Specifically, for the data transmission in this embodiment, the frames to be sent every 20 ms include: frame 0x01 (7 bytes), frame 0x04 (4 bytes), frame 0x05 (6 bytes), frame 0x02 (6 bytes), frame 0x06 (6 bytes, sent twice every 20 ms), and frame 0x07 (8 bytes, sent 0.2 times every 20 ms). A total of approximately 8.2 frames are sent every 20 ms, resulting in a bandwidth usage rate of less than 10%. Therefore, based on the above frame header definition and design, 16-DOF full-state synchronization can be achieved within a 20 ms period.

[0132] It should be further explained that the torque command transmitted in frame 0x02 refers to the output torque limit value of the joint motor, that is, the maximum torque that the motor of each finger joint can output. Specifically, the torque command is a control parameter sent by the host computer to the dexterous hand actuator to limit the magnitude of the force generated by the motor during movement. This torque limit value can be expressed as a percentage or a current value. For example, setting it to 100 means that the motor can output the rated maximum torque, and setting it to 50 means that the motor output torque is limited to half of the rated value. The purpose of the torque command is to prevent motor overload, protect the mechanical structure, and also to play a preliminary role in limiting force. In this embodiment of the invention, the torque command can be stored and sent through variables such as RightHand_Send.pressure_1 to pressure_5, with one torque limit value corresponding to each finger. As can be seen from the variable naming in the code, although the variable is named pressure, it actually stores the torque command value. For details, please refer to the JointEffort parameter in the ROS parameters. Furthermore, in the force control system of the dexterous hand, the torque command sets the safety upper limit of force control, and the current force sensing data can be finely adjusted within this upper limit.

[0133] It's important to further clarify that there are differences in accuracy, coverage, transmission frequency, and functionality between the current preliminary temperature data and the high-precision current joint temperature data. First, the current preliminary temperature data is transmitted via frame 0x02. Due to limited frame space and the fact that it's not a dedicated temperature channel, its accuracy is relatively low, approximately ±2 degrees Celsius. The current joint temperature data, transmitted via a dedicated frame 0x07, achieves an accuracy of ±0.5 degrees Celsius, which is significantly higher. Second, the current preliminary temperature data only covers a few key locations and typically reflects the temperature of the drive circuit board. The current joint temperature data can cover five to ten joints, each with an independent temperature value, offering a wider coverage and potentially including the content of the preliminary temperature data. Furthermore, the current preliminary temperature data can be transmitted every 20 milliseconds with frame 0x02, a high transmission frequency but insufficient accuracy. The current joint temperature data can be transmitted every 100 milliseconds, a relatively low transmission frequency but more accurate data. Finally, the current preliminary temperature data is mainly used for emergency overheat detection, such as immediately shutting down the machine when the temperature exceeds the secondary temperature protection line TempLimit_L2 (i.e., the second temperature warning threshold, 60 degrees Celsius). High precision is not required, but a fast response is essential. The current joint temperature data is used for finer thermal management, such as reducing the deceleration rate to 0.8 times at 45 degrees Celsius and to 0.5 times at 50 degrees Celsius. More precise temperature values ​​are needed to determine the deceleration factor.

[0134] In summary, the method described above based on the embodiments of the present invention can achieve different transmission speeds for different types of data, ensuring faster transmission of critical data. For example, force control data can be transmitted every 10ms, ensuring rapid response in the force control process and preventing damage to the screen; in addition, position / speed control commands can be transmitted every 20ms, ensuring transmission stability; and temperature control data can be transmitted every 100ms, preventing overly frequent temperature transmissions. This effectively ensures that the data transmission process is "unblocked," while the force control effect is more sensitive, fast, and efficient, allowing a dexterous hand to react immediately upon touching something, while ensuring that excessive force is not applied.

[0135] like Figures 2A-3 As shown, according to an embodiment of the present invention, the process of switching the current operating state of the state machine to a deceleration state or a heat dissipation state based on the current joint temperature data, the current preliminary temperature data, and the preset warning temperature range includes:

[0136] When the current working state of the state machine is switched to the deceleration state, the joint speed variable is dynamically modified according to the current joint temperature data and the preset warning temperature range, in order to reduce the joint motion angular velocity of at least one joint of the dexterous hand.

[0137] For the deceleration state transition process of the state machine, a global temperature monitoring thread can be used to analyze the current joint temperature data transmitted based on frame 0x07 in real time, and compare it with a preset warning temperature range to determine the deceleration state transition. In this embodiment of the invention, the joint velocity variable can be the joint velocity variable (JointVelocity) of each joint of the dexterous hand, which is modified by the global temperature monitoring thread. It can be understood as the baseline execution speed of the joint at the current moment (before the deceleration state transition). The global temperature monitoring thread can be a background task that runs independently of the state machine. It can continuously analyze the high-precision current joint temperature data returned by frame 0x07 and determine whether to modify the joint velocity variable of a specific joint in real time based on the corresponding temperature value, so that the joint speed execution is faster or slower.

[0138] Modifications to joint velocity variables can be achieved using a coefficient multiplication method. Specifically, a baseline joint velocity value can be preset through the ROS system. When the current joint temperature of a detected joint exceeds the first-level temperature protection threshold TempLimit_L1 (the first temperature warning threshold, 45 degrees Celsius), the actual execution speed of the current joint is updated. This is achieved by multiplying the joint velocity variable (baseline joint execution speed) by a decay coefficient α, where α can be less than or equal to 1. This allows for direct assignment of values ​​to the joint velocity variable, thus adjusting the joint execution speed. Specifically, after the temperature monitoring thread detects over-temperature, the state machine switches to a deceleration state. Furthermore, the temperature monitoring thread executes instructions such as JointVelocity = JointVelocity × α, ensuring that the actual joint execution speed value is used by the state machine in the next control cycle. This allows for slower dexterity hand movements, such as slower typing or sliding.

[0139] Therefore, after triggering a temperature-based threshold, the state machine implements a deceleration state switch. The execution of the deceleration state depends on the update of the joint speed variables by the temperature monitoring thread. In some embodiments of this invention, the global temperature monitoring thread can continuously monitor the current temperature value of each joint of the dexterous hand using the current joint temperature data. When the temperature of any joint exceeds the first-level temperature protection line TempLimit_L1 (the first temperature warning threshold), a first-level warning is triggered. Afterward, the state machine immediately executes a deceleration state switch. For example, the state machine can jump directly from the force control state to the deceleration state. The system joint execution speed is reduced according to the attenuation coefficient α, and the current joint execution speed is proportionally reduced according to the attenuation coefficient. Therefore, its core logic is "deceleration occurs when the temperature exceeds the limit," with temperature being the sole criterion for judgment, no longer relying on time, position, or other sensor data. Thus, the implementation of the deceleration state greatly simplifies the complex judgment criteria of the prior art.

[0140] The attenuation coefficient is designed to modify the joint velocity variable. In deceleration mode, different attenuation coefficients can achieve different degrees of attenuation of the joint's baseline execution speed. For example, when the attenuation coefficient satisfies α=0.5, the joint's baseline execution speed is halved for deceleration execution to reduce heat generation. Therefore, the joint velocity variable JointVelocity can be a variable that stores the current joint execution speed value. The ROS system reads this joint velocity variable when executing each action to determine the speed of finger movement. When the temperature monitoring thread detects overheating, it multiplies the current value of the joint velocity variable by the attenuation coefficient α, and then stores the calculation result back into the joint velocity variable, thus updating the joint velocity variable. From this moment on, all subsequent actions executed by the state machine will use this new joint velocity variable for the joint velocity execution value, thereby achieving the effect of speed reduction. The joint speed attenuation can be understood as the angle of rotation of the corresponding joint per unit time being reduced proportionally according to the attenuation coefficient. For example, α=0.5 means that the angle of rotation of the finger joint per unit time is reduced to half of the original value, thus halving the joint speed. Specifically, the angular velocity of the index finger joint pressing down during normal typing may be 90 degrees per second. After the speed is halved, it becomes 45 degrees per second. This means that the time required for the finger to move from the hovering position to the keyboard will be doubled. The pressing action that originally took 200 milliseconds will take 400 milliseconds to complete after the speed is halved.

[0141] Furthermore, joint velocity decay affects the angular velocity of all moving joints (i.e., global speed regulation), including the speed during downward pressure, upward pressure, sliding, and any movement requiring joint rotation. In other embodiments, joint velocity decay affects the angular velocity of specific moving joints, primarily those whose joint temperature exceeds the first-level temperature protection threshold (TempLimit_L1). This means that not all moving joints experience angular velocity decay (i.e., localized speed regulation) to conserve energy and achieve targeted heat dissipation. Joint velocity commands can be sent to each finger via frame 0x05, and each finger can have an independent velocity value. When this joint velocity variable is modified, the system uses this new velocity value when packaging frame 0x05, thereby reducing or decaying the joint movement velocity of at least one finger.

[0142] To achieve a more coordinated and human-like operating performance, a globally unified speed adjustment strategy can be adopted in this embodiment of the invention. That is, when the temperature monitoring thread detects that any joint exceeds the first-level temperature protection threshold (TempLimit_L1), the system will uniformly multiply the speed variable of that joint by a decay coefficient. This variable is globally shared; all fingers and all joints read the same joint speed baseline execution value when performing actions. Therefore, the speed decay effect will simultaneously apply to all 16 degrees of freedom, including every joint of the thumb, index finger, middle finger, ring finger, and little finger.

[0143] By using global, unified speed control instead of localized speed control, several advantages are achieved. First, it prevents other joints from overheating by uniformly reducing speed, as the joint that overheats is often the one that experiences the most severe overheating, while other joints may also be at relatively high temperatures. Second, it ensures more coordinated execution of the dexterous hand's target movements, resulting in better operational efficiency. When performing actions like typing or swiping, the movements of each finger are coordinated. If some fingers move faster than others, it can lead to incoordination and even affect operational efficiency. For example, if the index and middle fingers move at different speeds when typing, it may cause keystroke timing errors. Therefore, global speed control ensures the effectiveness of thermal protection while maintaining the coordination of finger movements.

[0144] like Figures 2A-3 As shown, according to an embodiment of the present invention, reducing the joint motion angular velocity of at least one joint of a dexterous hand includes: determining a corresponding target deceleration dexterous hand joint that does not meet the preset warning temperature range based on current joint temperature data; and reducing the joint motion angular velocity of the target deceleration dexterous hand joint.

[0145] The current joint temperature data can include the actual current operating temperature of each joint in the dexterous hand. Therefore, by comparing the actual current operating temperature of each joint with the temperature threshold of a preset warning temperature range (such as a primary temperature warning line and a secondary temperature warning line), the corresponding joints that do not meet the preset warning temperature range are marked, forming multiple joint data points exceeding the threshold temperature as target dexterous hand joints for speed reduction. Based on these target dexterous hand joint data, the joint motion angular velocity of these marked joints is reduced specifically. In other words, the target dexterous hand joints for speed reduction include joints whose actual operating temperature exceeds the preset warning temperature range. Therefore, by selectively slowing down only the target dexterous hand joints, the normal operation of other joints can be ensured, effectively preventing the slowing down of all hand joints, thus achieving significant energy savings. Simultaneously, it ensures the overall continuity of movement, minimizes deviations in hand and finger trajectories, and improves operational efficiency.

[0146] like Figures 2A-3As shown, according to an embodiment of the present invention, in switching the current operating state of the state machine to a deceleration state or a heat dissipation state based on the current joint temperature data, the current preliminary temperature data, and the preset warning temperature range, the method further includes:

[0147] When the current working state of the state machine is switched to the heat dissipation state, the preset heat dissipation state execution sequence is called according to the current preliminary temperature data and the preset warning temperature range to perform heat dissipation posture control on the fingers and palm of the dexterous hand.

[0148] In this embodiment of the invention, under the control of the global temperature monitoring thread, the state machine can be forced to switch to the heat dissipation state. Specifically, it can directly switch from any working state such as the force control state or the deceleration state to the heat dissipation state. Therefore, as a background task independent of the main loop of the state machine, the global temperature monitoring thread can also continuously monitor the temperature values ​​of each joint (current joint temperature data) and actively intervene in the operation of the state machine when an anomaly is detected.

[0149] Specifically, the global temperature monitoring thread obtains the current temperature value of each joint in real time by parsing the high-precision current joint temperature data returned in frame 0x07 and the coarse preliminary current temperature data returned in frame 0x02. When it detects that the temperature of any joint reaches or exceeds the secondary temperature protection line TempLimit_L2 (the second temperature warning threshold, such as 60 degrees Celsius), it can directly control the execution of a jump operation without waiting for the state machine to react automatically, forcing the current working state of the state machine to the heat dissipation safe protection mode (Safe_Mode). This design ensures that even if the state machine is performing an action, the temperature monitoring thread can interrupt the current action and immediately switch to the protection mode.

[0150] The forced jump mechanism uses a flag and state overriding. Specifically, the ROS system can set the `thermal_emergency_flag` flag globally, initially setting it to false. When the temperature monitoring thread detects that the joint temperature, such as the current preliminary temperature data, exceeds the secondary temperature protection threshold `TempLimit_L2`, it can first set this flag to true and then directly modify the current state variable of the state machine, assigning it the state value corresponding to the heat dissipation state.

[0151] Furthermore, the state machine's main loop checks the current state value at the beginning of each control cycle. When the state machine detects that the current state has been modified to a heat dissipation state, it executes the corresponding processing function for the heat dissipation state and stops executing the original action. Simultaneously, the state machine also checks the `thermal_emergency_flag` flag; if it is true, it prioritizes the safety protection logic, performs the heat dissipation state switch, and ignores all other instructions.

[0152] Therefore, the temperature monitoring thread can be given the highest priority, interrupting the normal operation of the state machine at any time and forcibly switching the system to a safe mode for heat dissipation. Similar to hardware interrupts in a computer system, exceeding the temperature limit is the condition that triggers the interrupt, and the heat dissipation state Safe_Mode can be understood as an interrupt handler.

[0153] State transitions in the heat dissipation state can be achieved by directly assigning values ​​to modify state variables. In the code implementation, the state machine typically uses an enumerated variable, such as `current_state`, to record the current state. During normal operation, the state machine can automatically modify the values ​​of these variables based on the execution of actions, such as transitioning from the pressure state to the force control state, then to the return state, or the heat dissipation state. When the temperature monitoring thread detects over-temperature (secondary temperature protection line TempLimit_L2), it can directly execute an assignment update, setting the current `current_state` variable to the Safe_Mode of the heat dissipation state, for example, `current_state = Safe_Mode`. Since the temperature monitoring thread and the state machine's main loop share the same state variable, modifications made by the temperature monitoring thread are immediately seen by the state machine's main loop. At the beginning of each cycle, the state machine's main loop reads the value of `current_state`. If it finds that it is already in Safe_Mode, it stops executing the original state logic and instead executes the corresponding Safe_Mode processing function, thus achieving the state transition.

[0154] Therefore, this state transition method allows for quick and direct state switching without complex condition checks or function calls; a single assignment statement can complete the transition. Furthermore, mutexes or atomic operations can be used to address thread safety issues, preventing conflicts caused by the temperature monitoring thread and the state machine's main loop simultaneously modifying state variables.

[0155] The preset heat dissipation state execution sequence can be the sequence of joint actions performed by each joint during the controlled execution of the heat dissipation state, from the initial working state (such as the force-controlled state) to the final heat dissipation posture (i.e., the heat dissipation state). This sequence includes the joint execution time, angle, speed, and execution order of each joint. In this embodiment of the invention, when the dexterous hand switches from the current working state to the heat dissipation state, the reset and adjustment to the heat dissipation posture can be gradually achieved based on the slow movement of the joints.

[0156] During the cooling-down phase, the slow joint movement can reduce the joint's angular velocity to 10% to 20% of the normal baseline execution speed, a significant decrease. For example, in normal typing, the joint's angular velocity is approximately 90 degrees per second, but in the slow cooling-down phase, it drops to 10 to 18 degrees per second, about 1 / 5 to 1 / 9 of the normal speed. Similarly, in normal gliding, the finger moves at approximately 200 millimeters per second, but this can decrease to 20 to 40 millimeters per second in the slow cooling-down phase. Therefore, while a normal downward press takes about 200 milliseconds to go from hover to contact, the entire process of reaching the cooling-down posture in the slow cooling-down phase can take 1 to 2 seconds.

[0157] By utilizing the slow reset process of the cooling posture, minimal frictional heat is generated through low-speed movement, avoiding additional frictional heat from joint movement. Simultaneously, the slow opening of the fingers avoids impact from rapid movement and facilitates user observation and intervention. Furthermore, it prevents excessively slow joint movement, which could lead to prolonged cooling posture execution and keep hot joints at high temperatures for extended periods, thus maximizing heat dissipation requirements. For example, in a preferred embodiment of this invention, during normal typing, the joint angular velocity decreases from approximately 90 degrees per second to 10 to 18 degrees per second after entering the slow cooling state. From a heat dissipation perspective, the frictional heat generated by low-speed movement is minimal, almost negligible in terms of additional heat generation. From a protection perspective, the impact force generated by low-speed movement is small, preventing damage even if the fingers encounter obstacles. At the same time, the speed of 10 to 18 degrees per second is not too slow, ensuring that the cooling posture reset action is completed within an acceptable timeframe.

[0158] It should be noted that resetting the heat dissipation posture can be understood as moving each joint of the dexterous hand from its current position to a preset heat dissipation posture position, rather than returning to the mechanical zero point or the initial working position. That is, moving each joint of the dexterous hand from its current position to a preset target joint position that can achieve heat dissipation. This target position is not the mechanical zero point, but a heat dissipation posture that can be defined specifically for heat dissipation design.

[0159] The heat dissipation posture is a hand posture specifically designed for heat dissipation optimization. Its specific form can be: the five fingers are naturally straightened and spread out to increase the heat dissipation surface area; there are gaps between the fingers to facilitate air convection; the palm faces upward or to the side to allow heat to dissipate upward; the joints are in a relaxed state, the motor load is minimal, and no additional heat is generated. Specifically, the heat dissipation posture is a hand posture specifically optimized for heat dissipation that the dexterous hand automatically enters when the joint temperature of the current preliminary temperature data exceeds the secondary temperature protection line TempLimit_L2 (the second temperature warning threshold, such as 60 degrees Celsius). The specific forms are as follows: (1) Finger posture: the five fingers are naturally straightened and spread out, with gaps between the fingers, no longer bent or clenched; the angle of each joint is set to a position close to the middle of the mechanical stroke, neither the dead point of complete extension nor the extreme position of bending. (2) Palm posture: the palm faces upward or to the side to allow heat to dissipate upward naturally; if the palm faces downward, the heat will be trapped between the back of the hand and the palm, resulting in poor heat dissipation. (3) Joint state: All joints are in an extended state, the motor is basically unloaded, the output current is close to zero, and no new heat is generated. Simply put, the heat dissipation posture is like the way a human hand naturally opens and rests after an injury, rather than a working posture of clenching a fist, grasping, or pointing in a certain direction.

[0160] Therefore, by adopting a cooling posture—extending and spreading the fingers—the heat dissipation area of ​​each joint is maximized, allowing air to flow through the gaps between the fingers. This cooling efficiency is far superior to that of a clenched fist or a naturally bent hand. Furthermore, in this cooling posture, the motor experiences virtually no load, with the output current approaching zero, thus generating no new heat. Moreover, the cooling posture is stable and easily observable, allowing users or maintenance personnel to visually determine whether the system is in a protection state. Therefore, the cooling posture achieved after execution offers at least the following technical benefits:

[0161] First, the optimal cooling posture maximizes heat dissipation efficiency. When fingers are naturally extended and spread out, the heat dissipation area of ​​each joint is at its maximum, allowing air to flow through the gaps between the fingers, resulting in optimal convective heat transfer. In contrast, when the fingers are clenched into a fist, heat is not easily dissipated. Therefore, when the fingers are extended and spread out, the surface area of ​​each joint and finger is fully exposed to the air without any obstruction. The gaps between the fingers allow air to flow freely, maximizing convective heat transfer. Experiments show that the cooling rate in the optimal cooling posture is approximately 30% to 50% faster than in the clenched fist posture.

[0162] Secondly, the motor load is minimized in the cooling posture. With fingers extended and spread, the motor only needs to overcome very little elastic resistance to maintain the posture, with output current close to zero, and it generates no new heat. If the system returns to a stressed posture, the motor still needs to output torque to maintain the posture, which will actually generate more heat. In the cooling posture, the motor is essentially unloaded, requiring only a small current to overcome the static friction of the joint to maintain the posture. This means that the system itself does not generate new heat during the cooling process and can focus on dissipating existing heat.

[0163] Third, the cooling posture provides a clear visual indication. The cooling posture is a distinctive hand shape that clearly distinguishes it from the normal working posture. When a user sees a dexterous hand suddenly spread its fingers, palm facing upwards, and cease all movement, they can intuitively determine that the system is in overheat protection mode, allowing them to promptly take measures such as checking the cooling fan, reducing the workload, or suspending use. If the hand is reset to the zero position or another working posture, the user cannot distinguish between normal standby and overheat protection.

[0164] Fourth, the cooling posture facilitates a quick return to work. Once the temperature drops to a safe range, the system can switch directly from the cooling posture to the working posture much faster than from a fully reset state, reducing the number of steps and recovery time. Because the fingers are in a naturally open position in the cooling posture, rather than in an extreme curled or twisted position, switching to working postures such as typing or swiping requires the fewest steps and has the shortest recovery time.

[0165] Fifth, protect hardware safety. When the temperature reaches the secondary temperature protection threshold TempLimit_L2, the motor winding insulation may begin to be damaged, and the power devices on the drive board may burn out due to overheating. The heat dissipation mechanism immediately stops the current output by cutting off the motor enable, preventing further heat accumulation and thus protecting the motor and driver from damage.

[0166] Therefore, in the implementation of this invention, the heat dissipation posture is an optimized posture that integrates heat dissipation efficiency, low power consumption, status indication and fast recovery, rather than simply "returning to the original position" or "stopping at will".

[0167] In summary, in this embodiment of the invention, when the joint temperature value of a specific joint in the current preliminary temperature data exceeds the second-level temperature protection line TempLimit_L2 (the second temperature warning threshold, such as 60 degrees Celsius) defined by the preset warning temperature range, the state machine is forced to jump from the current working state to the heat dissipation state. The function of this heat dissipation state can be reflected in the passive heat dissipation achieved by changing the hand posture and stopping the power output when the system temperature is too high, thereby protecting the hardware and maintaining the availability of the system. Specifically, it involves the following four aspects: (1) Protecting the hardware: When the temperature exceeds the second-level temperature protection line TempLimit_L2, the insulation layer of the motor winding may be damaged, and the power devices on the drive circuit board may also burn out due to overheating. The heat dissipation state immediately stops the current output by cutting off the motor enable, preventing the heat from continuing to accumulate, thereby protecting the motor and the driver. (2) Accelerating cooling: By straightening and spreading the fingers with the palm facing up, the heat dissipation area is maximized, the air convection is smoothest, and the cooling speed is about 30% to 50% faster than maintaining the working posture. (3) Maintaining system availability: Compared to direct power outages and shutdowns, where the system is completely unavailable and requires manual intervention to recover, the traditional overheat protection scheme, in the heat dissipation state of this embodiment, the system remains in standby mode during cooling. Once the temperature drops to a safe range, it can automatically resume operation without manual restart. (4) Providing status indication: The heat dissipation posture is a special hand shape that is clearly different from the working posture. When users or maintenance personnel see the fingers naturally spread and stop moving, they can intuitively judge that the system is in an overheat protection state, which facilitates timely measures such as checking the cooling fan and reducing the workload.

[0168] The process of switching from the current working state to the heat dissipation state, and then resetting to the heat dissipation posture to complete the disabling of the disabled state, can be described as follows: First, when the system detects that the temperature of any joint in the current preliminary temperature data exceeds the secondary temperature protection threshold TempLimit_L2, the secondary protection mechanism is activated. TempLimit_L2 is a critical safety threshold or range, because exceeding this temperature setting may damage the motor winding insulation and burn out the power devices on the drive board due to overheating. Then, leveraging the highest priority of the temperature monitoring thread, the normal operation of the state machine can be interrupted at any time, directly setting the current working state to the heat dissipation safe mode (Safe_Mode). At this point, regardless of the dexterous hand's current state or mode, whether it is typing, swiping, or performing other actions, it will immediately stop and execute the protection logic, initiating the switch to the heat dissipation state. This differs from normal state machine transitions, where transitions occur sequentially after actions are completed, while forced transitions are immediate interruptions in emergency situations. Furthermore, during the slow reset to the cooling posture after switching to the cooling state, the system does not immediately stop the movement of all joints. Instead, it slowly moves each joint from its current position to the preset cooling posture. The reason for the "slow" reset is that rapid movement generates additional frictional heat and impact force, which is detrimental to heat dissipation and may cause secondary damage. By using a hand posture specifically optimized for heat dissipation, such as five fingers naturally extended and spread, palm facing up or to the side, the heat dissipation area is maximized, air convection is smoothest, and motor load is minimized, achieving a rapid heat dissipation effect while also taking into account various benefits such as hardware protection. Finally, the enable is cut off, that is, the output of the motor driver is turned off, so that the motor no longer generates torque. At this time, the dexterous hand enters a zero-output state and relies entirely on passive cooling to cool down, without generating new heat by maintaining the posture. In addition, cutting off the enable is also a safety measure to prevent accidental injury caused by motor runaway at high temperatures.

[0169] Therefore, in the method of the above embodiments of the present invention, a complete "perception-decision-execution" chain can be formed with the help of a two-level protection mechanism. First, the high-bandwidth multi-frame fusion communication protocol is responsible for the "perception" link—uploading the current joint temperature data and the current preliminary temperature data in real time through frames 0x07 and 0x02, providing the system with a basis for judgment. Second, the dynamic state machine driven by multimodal sensing is responsible for the "decision" link—the temperature monitoring thread parses the temperature data, and when the temperature exceeds the two-level temperature protection line, it forcibly jumps the state machine to the heat dissipation state Safe_Mode. In addition, the active frequency reduction mechanism based on temperature trend is responsible for the "execution" link—the system executes a slow reset to the heat dissipation posture and cuts off the enable, completing the protection action. It can be seen that in the above "perception-decision-execution" heat dissipation state switching execution chain, all three are indispensable, so as to realize the complete two-level protection function of "automatically and slowly resetting to the heat dissipation posture and cutting off the enable when the temperature exceeds the standard".

[0170] like Figures 2A-3 As shown, according to an embodiment of the present invention, in responding to a received work operation command, and based on the starting position state of the dexterous hand and the current working state of the state machine, a target touch operation is performed on the target screen, including: responding to the received work operation command, determining the current working state of the state machine based on the starting position state of the dexterous hand; and performing the target touch operation on the target screen based on the current working state of the state machine.

[0171] As mentioned above, the work operation command can be a control response command used to control the dexterous hand to perform target touch operations on the target screen, aiming to enable the dexterous hand to complete the specified target action task. The starting position state, as the starting posture of the dexterous hand to perform the target touch operation, can define the spatial position of the dexterous hand at the start of the work (including the specific spatial positions of each joint, finger, wrist, etc.) as well as the joint movement angle, joint movement speed, etc., thereby determining the working state of the dexterous hand at that moment (such as the downward pressure state during the execution of other action tasks).

[0172] Based on the current working state of the dexterous hand, target touch operations can be performed on the target screen. Specifically, at the moment of response to the work operation command, when it is confirmed that the current working state of the dexterous hand is the pressing state of another action task, the pressing state of that action task can be terminated immediately, and the pressing state can be directly used as the starting position state to perform the next target action task. Adaptive adjustments can be made based on the joint angle, execution position, etc. of the previous pressing state to achieve target touch operation on the target screen.

[0173] Therefore, any gesture or posture of the dexterous hand at any time can be used as the starting position state to execute the target touch action specified by the target action task at the next moment. This avoids the situation in traditional technology where the starting posture is fixed and the default is reset to the starting posture before each task is executed. This greatly improves the execution continuity of dexterous hand action tasks, has higher execution efficiency, and is more in line with the execution habits of real human actions.

[0174] like Figures 2A-3 As shown, according to an embodiment of the present invention, in addition to switching the current operating state of the state machine to a deceleration state or a heat dissipation state based on the current joint temperature data, the current preliminary temperature data, and the preset warning temperature range, the method further includes: switching the current operating state of the state machine to a halt state based on the current joint temperature data, the current preliminary temperature data, and the preset warning temperature range; halting the movement of at least one joint of the dexterous hand in response to a switching command of the halt state; and switching the current operating state of the state machine to a deceleration state or a heat dissipation state based on updated joint temperature data determined during the halting of the movement of at least one joint.

[0175] In one embodiment of the present invention, the aforementioned global temperature monitoring thread can be used to force the state machine to jump to the halt state. The halt state, as a working state, can be understood as the state in which all joints of the dexterous hand (including the wrist joint) suspend all current movements, immediately entering a state where the action is halted and the current posture remains unchanged. That is, when entering the halt state, all joint movements of the dexterous hand are suspended, and at that moment, the dexterous hand maintains its original posture.

[0176] In some embodiments of the present invention, when the current joint temperature data or the current preliminary temperature data of the dexterous hand exceeds a specified threshold or threshold range of a preset warning temperature range, a stop state can be switched before switching between the deceleration state and the heat dissipation state. This causes all joints of the dexterous hand to be in a state of complete or partial suspension of movement at this time, allowing it to directly utilize the motion posture of the stop state for heat dissipation, thereby avoiding further switching to the deceleration state or the heat dissipation state. Especially when the current joint temperature data or the current preliminary temperature data significantly exceeds the preset warning temperature range, the current motion posture can be directly used for on-site heat dissipation, preventing further movement from causing a more serious or uncontrollable temperature rise.

[0177] Moreover, this operation can also avoid the situation where the dexterous hand immediately enters a slow-down or heat dissipation state once it overheats, so that it can quickly continue to complete the target action task after cooling down, thereby greatly reducing energy consumption and improving task execution efficiency.

[0178] In another embodiment of the invention, further, in the suspended state of the dexterous hand, although the current joint temperature data and the current preliminary temperature data still exceed the preset warning temperature range, they can be gradually cooled down, causing the temperature value of the corresponding structure to gradually decrease. At this time, periodic temperature transmission detection can be used to obtain the real-time temperature data of each joint after the update detection as the updated joint temperature data, thereby determining the relationship between it and a specified threshold or threshold range of the preset warning temperature range. When it gradually approaches the preset warning temperature range, for example, when it exceeds a certain preset temperature value (greater than the secondary temperature warning line), the current working state of the state machine can be switched to a deceleration state or a heat dissipation state.

[0179] In this way, a more stable deceleration or heat dissipation mode can be achieved, providing a safer heat dissipation method without damaging the dexterous hand due to high temperature.

[0180] like Figures 2A-3 As shown, according to another embodiment of the present invention, adjusting the downward pressure of the dexterous hand or switching the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range includes:

[0181] Based on the force sensing deviation determined by the current force sensing data and the preset force control threshold range, and combined with the preset force sensing deviation coefficient, the downward pressure Δθ of the dexterous hand is determined. This is used to adjust the downward pressure of the dexterous hand's fingers on the target screen based on the downward pressure and the first preset limit value while maintaining the current working state of the state machine.

[0182] In this invention, the current force sensing data returned via frame 0x06 is received. By comparing this data with force control parameters defined within a preset force control threshold range, such as the force threshold ForceThreshold, minimum trigger force F_min, maximum safety force F_max, and target holding force F_target, the corresponding force sensing deviation and state machine switching timing can be determined. The force sensing deviation can be understood as the difference between the current actual contact force F_curr of a finger in the current force sensing data and the corresponding target holding force F_target, F_target - F_curr. Furthermore, when it is determined that the current actual contact force F_curr of a finger in the current force sensing data satisfies the following condition: At this time, the state machine will maintain its current working state and will not force a jump to achieve state switching.

[0183] In this context, "no transition" in the state machine can be understood as the state machine remaining in its current mode. For example, when a dexterous hand is in a force-controlled state, if the detected actual contact force F_curr is less than the minimum trigger force F_min, it means that although the finger has made contact with the object, the pressure applied is insufficient, and a button press or click operation may not be reliably triggered. In this case, the system will not switch to another state but will remain in the force-controlled state, increasing the downward pressure by fine-tuning the joint angle until the contact force reaches the target range. This contrasts with normal state transitions. Normally, when the contact force reaches the target value and is maintained for a sufficient time, the state machine will transition from the force-controlled state to the return state, indicating that a button press action has been completed. However, when the force is insufficient, the state machine chooses not to transition and continues to adjust within the force-controlled state. Therefore, "no transition" means that the state machine remains in its current state without switching states, but the internal adjustment process continues. It can be understood as being in a state of in-situ adjustment without switching to other states.

[0184] Among them, the proportional control mechanism based on the PID algorithm, under the condition that the current actual contact force F_curr is less than the minimum trigger force F_min, the downward pressure Δθ of the dexterous hand can be determined based on the force sensing deviation between the corresponding current actual contact force F_curr and the target holding force F_target, specifically satisfying the following formula 1:

[0185] Δθ= Kp × (F_target - F_curr) (1)

[0186] The preset force sensing deviation coefficient Kp can be a proportionality coefficient calibrated through experiments. Furthermore, integral and differential terms can be added to achieve steady-state error elimination and oscillation suppression. The downward pressure Δθ of the dexterous hand can be understood as the adjustment amount (i.e., the degree of pressure, in degrees) of each joint angle of the corresponding finger. The aim is to adjust the finger's contact based on this adjustment amount, ensuring that the actual contact force between the finger and the screen satisfies the minimum trigger force F_min and the maximum safety force F_max, and is close to or equal to the target holding force F_target.

[0187] Using the formula (1) above, fine adjustments to the pressure of the fingers in a dexterous hand can be achieved. The range of a single adjustment can be limited to ±2 degrees to ±5 degrees, depending on the magnitude of the current force sensing deviation. If the force sensing deviation is very small, such as only 0.1 Newtons, the single adjustment may only be 0.5 degrees; if the force sensing deviation is large, such as not contacting the object at all, the single adjustment can reach 5 degrees. However, in any case, the fine adjustment of the pressure amount usually will not exceed the first preset limit value. The first preset limit value can define the maximum adjustment range of the finger pressure amount, which can effectively prevent force control oscillation caused by excessive adjustment in a single adjustment. For example, the single adjustment will not exceed the first preset limit value (such as 5 degrees). For cases that exceed the first preset limit value, the first preset limit value can be directly taken. In another embodiment of the present invention, if the total amount to be adjusted exceeds the first preset limit value, the system can also complete the adjustment step by step in multiple control cycles, with each cycle adjusting only one small step. Each control cycle is 10 milliseconds, which is consistent with the sampling cycle of the force sensing data. In this way, the adjustment of the downward pressure can be effectively achieved without force-controlled oscillation.

[0188] like Figures 2A-3 As shown, according to another embodiment of the present invention, adjusting the downward pressure of the dexterous hand or switching the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range includes:

[0189] Based on the force sensing deviation determined by the current force sensing data and the preset force control threshold range, and combined with the contact stiffness coefficient, the downward pressure of the dexterous hand is determined. This is used to adjust the downward pressure of the dexterous hand's fingers on the target screen based on the downward pressure and the second preset limit value while maintaining the current working state of the state machine.

[0190] In another embodiment of the present invention, when it is determined that the current actual contact force F_curr of a certain finger in the current force sensing data satisfies the following condition: At this time, the state machine will maintain its current working state and will not force a jump to achieve a state switch. Specifically, based on the proportional control mechanism of the PID algorithm, the downward pressure Δθ of the dexterous hand can be determined based on the force sensing deviation between the corresponding current actual contact force F_curr and the target maintaining force F_target, specifically satisfying the following formula 2:

[0191] Δθ = α × (F_target - F_curr) / K_contact (2)

[0192] Where: Δθ is the joint angle adjustment amount, in degrees; α is an empirical coefficient, the value range can be α=0.5 to α=1.5, and can be determined according to the joint type and movement speed; F_target is the target contact force, in Newtons; F_curr is the current actual contact force, in Newtons; K_contact is the contact stiffness coefficient, in Newtons per millimeter, which can be determined according to the screen being operated, for example, 10 to 20 Newtons per millimeter for mobile phone screens. At the same time, it is also necessary to give the limit condition for the single adjustment amount: |Δθ| ≤ Δθ_max, where Δθ_max is the second preset limit value, specifically 2 to 5 degrees. If the calculated Δθ exceeds the second preset limit value, then the second preset limit value is taken, and the residual force sensing deviation continues to be adjusted in the subsequent control cycle. Therefore, formula (2) not only includes the well-known PID proportional control idea, but also reflects the special parameter design for the force control scenario of dexterous hand, and can better reflect its technical contribution.

[0193] Regarding the technical content of the above formulas (1) and (2), a proportional-integral / derivative control algorithm can be used to calculate the control quantity based on the deviation. Among them, the controlled object for calculating the downward pressure Δθ is the mapping relationship between the joint angle of the dexterous hand and the fingertip contact force. This mapping has nonlinear and coupling characteristics. The contribution of the movement of different joints to the fingertip force is different. Therefore, the above method of the present invention is significantly different from the case in the existing known PID algorithm that requires targeted parameter tuning. Furthermore, the adjustment range of the downward pressure is constrained by the mechanical structure. The movement range of different fingers or even different joints is different, and it is necessary to set the limit value separately. The first preset limit value and the second preset limit value, etc., can be determined experimentally and have specificity. Thirdly, the calculation of the downward pressure also needs to consider the stiffness difference of the contact object. Different screens have different elastic coefficients, and the contact force generated by the same downward pressure is also different. Therefore, the application scenarios of different types of screens can be better adapted by the pre-set contact stiffness coefficient.

[0194] like Figures 2A-3 As shown, according to another embodiment of the present invention, in adjusting the downward pressure of the dexterous hand or switching the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range, the method further includes:

[0195] When switching the current working state of the state machine to the return state, based on the current force sensing data and the preset force control threshold range, the preset return state execution sequence is called to perform return control of the dexterous hand in the starting position state.

[0196] In this invention, the current force sensing data returned via frame 0x06 is received. By comparing this data with force control parameters defined within a preset force control threshold range, such as the force threshold ForceThreshold, minimum trigger force F_min, maximum safety force F_max, and target holding force F_target, the corresponding force sensing deviation and state machine switching timing can be determined. Furthermore, when it is determined that the current actual contact force F_curr of a certain finger in the current force sensing data satisfies the following condition: At this time, the dexterous hand will be controlled to immediately stop pressing down on the screen, and the state machine will be forced to trigger a return action. That is, the state machine will be forced to switch from the current working state to the return state. By triggering the return state execution process, the screen will be prevented from being damaged or scratched by pressing down.

[0197] Therefore, the state machine transitions when the current actual contact force F_curr is greater than the maximum trigger force F_max. For example, if At this time, the system is in force control mode, maintaining the pressure of the finger on the keyboard or screen. When it detects that the actual contact force F_curr exceeds the maximum safe force F_max, it usually indicates that the finger is pressing too hard on the screen and may break it. In this case, the system must immediately stop the finger from pressing down and release it from contact. The return state transition is an emergency jump, which differs from a normal state transition: normal state transitions occur sequentially after the action is completed, such as from force control mode to return state, while an emergency jump immediately interrupts the current action, regardless of whether the current action is completed, and directly switches to the target state. The target of the state transition is the return state. After entering the return state, the system will immediately begin executing the return action according to the preset return state execution sequence, that is, the finger quickly lifts away from the contact object and returns to the starting position before the start of this action. Therefore, the preset return-to-position execution sequence can be the sequence of joint actions performed by each joint from its initial working state (such as force-controlled state) to its initial position state during the controlled return-to-position process of a dexterous hand. This includes the timing, angle, speed, and execution order of each joint. In this way, the fingers can quickly detach from the object, avoiding the application of excessive pressure.

[0198] The return motion refers to the rapid lifting of the finger from its contact position back to its initial position before the action began. The return motion is executed much faster than the normal pressing speed, typically 1.5 times faster. For example, the joint angular velocity during normal pressing is 90 degrees per second, while the return motion can reach 135 degrees per second. In other words, the joint angular velocity controlled during return is greater than the joint angular velocity in the current working state. Therefore, even if excessive pressing pressure is detected, the finger can be released from contact in the shortest possible time, preventing damage to the screen. Furthermore, the return motion requires a crisp and clean movement, without hesitation or pauses; the upward lift of the finger is usually greater than the upward lift during the cooling state.

[0199] As mentioned earlier, the initial position can be the hand posture position just before the dexterous hand is controlled to perform the target touch operation, and can serve as the starting posture for that target touch operation. The initial position varies depending on the application scenario. For example, in typing, the initial position is when the finger hovers about 2 to 5 millimeters above the screen. This position prevents accidental touches and allows for quick downward pressure when touch is needed. The finger is in a naturally bent, ready position to execute the next screen touch. In touchscreen swiping scenarios, the initial position is when the finger hovers about 1 to 3 millimeters above the screen. Since touchscreens are generally more sensitive than mechanical keyboards, the hovering height can be lower for quick contact with the screen to begin swiping. In holding scenarios, the initial position is slightly different, such as... Figure 3 As shown, when the dexterous hand 301 grips the phone 302, the initial state may not be a hovering state, but rather a holding posture with a gripping force of 2 to 3 Newtons. In this case, the return motion does not completely remove the phone 302, but rather reduces the gripping force to the minimum level necessary to maintain contact. It should be noted that the initial position state may not be equal to the system's initial zero point or heat dissipation posture. The initial zero point can be the mechanical zero position of each joint when the system is powered on; the heat dissipation posture can be the specific posture of the dexterous hand for heat dissipation needs during overheat protection, as described above; and the initial position state can be the preparatory position of the fingers before each action begins, changing depending on the scenario and dynamically updated with repeated execution of the action.

[0200] like Figures 2A-3 As shown, according to another embodiment of the present invention, in adjusting the dexterity hand's downward pressure or switching the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range, the method further includes: responding to the received current joint temperature data, current preliminary temperature data, and preset warning temperature range, adjusting the dexterity hand's downward pressure or switching the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range, and then switching the current working state of the state machine to a deceleration state or a heat dissipation state based on the downward pressure or the return state.

[0201] In another embodiment of the present invention, if the current joint temperature data or the current preliminary temperature data exceeds the temperature threshold or threshold range defined by the preset warning temperature range, and if the current force sensing data exceeds the preset force control threshold range, the downward pressure of the dexterous hand in its current working state can be adjusted according to the current force sensing data and the preset force control threshold range, or the hand can be directly switched to the return state and executed.

[0202] After the adjustment of the downward pressure or the return state is completed, the system further switches to the deceleration state or the heat dissipation state based on the current joint temperature data and the current preliminary temperature data. For details regarding the adjustment of the downward pressure, the return state execution, the deceleration state execution, and the heat dissipation state execution, please refer to the previous text; they will not be repeated here.

[0203] Therefore, when both temperature data and force control data do not meet the corresponding preset warning range, by first performing pressure adjustment or return state switching, the protection effect on the target screen can be prioritized to prevent further damage to the screen. At the same time, by using the execution of pressure adjustment or return state, the influence of temperature can be further taken into account, and rapid heat dissipation can be achieved without damaging the screen.

[0204] In this embodiment of the invention, for data transmission based on preset bidirectional communication rules, a corresponding real-time communication closed-loop effect can be achieved based on a dual timing mechanism.

[0205] The dual-timer mechanism can be understood as equipping the dexterous hand control system with two timers, m_handSend_timer and m_sensorRecv_timer, each with its own function. m_handSend_timer is responsible for sending commands, while m_sensorRecv_timer is responsible for receiving feedback; both operate independently. For example, m_handSend_timer continuously sends commands (such as sending a "bend finger to 90 degrees" command every 20 milliseconds), while m_sensorRecv_timer continuously receives external feedback (such as receiving a "currently bent to 88 degrees" detection data from the finger every 20 milliseconds). m_handSend_timer and m_sensorRecv_timer do not interfere with each other, each working at its own pace, but they share the same information board. m_handSend_timer writes the commands to be sent onto the board, and m_sensorRecv_timer writes the received feedback information onto the board.

[0206] In this embodiment of the invention, `m_handSend_timer` can be triggered every 20 milliseconds. Its task is to package the current control commands and send them to the dexterous hand hardware through a preset bidirectional communication rule (such as the CAN bus protocol), telling each joint what angle to rotate to, how fast to rotate, and how much torque to output. `m_sensorRecv_timer` can also be triggered every 20 milliseconds. Its task is to receive data transmitted back from the dexterous hand via the CAN bus, including the actual joint angles, fingertip pressure, joint temperature, etc., and then parse this raw data into a form that the program can understand and store it in shared memory. Both timers have a period of 20 milliseconds, but they run independently, not sequentially. That is, within the same 20-millisecond window, the system is both sending commands and receiving data, achieving full-duplex real-time communication.

[0207] Specifically, in this embodiment of the invention, the timer m_handSend_timer can send control-related commands such as position, speed, and torque based on a first sending cycle (e.g., 20ms). Position refers to the target angle and actual feedback angle of each joint of the dexterous hand, which can be understood as the aforementioned joint position data. The target angle is the command value calculated by the host computer, reflecting the position the joint should rotate to; the actual feedback angle is the true value measured by the encoder, reflecting the actual position the system joint rotates to. As mentioned above, position detection can be achieved through an encoder built into each joint. The encoder is an angle sensor that measures the actual angle of joint rotation. When the motor drives the joint to rotate, the encoder generates a pulse signal proportional to the rotation angle, and the driver can obtain the current actual angle by counting the pulses. Furthermore, speed is detected by calculating the position difference from the encoder. The system records the position difference between two adjacent control cycles, divides it by the time interval, and obtains the average speed. For example, if the joint angle in the previous cycle was 80 degrees, and this cycle is 90 degrees, with a time interval of 20 milliseconds, then the speed is 10 degrees divided by 0.02 seconds, which equals 500 degrees per second. Speed ​​commands are directly issued by the host computer and can be sent to the dexterous hand via frame 0x05. Torque is detected by detecting motor current. The motor output torque is directly proportional to the drive current. The driver has an internal current sampling circuit that can measure the current flowing through the motor windings in real time, and then calculate the current output torque. Torque commands are sent by the host computer via frame 0x02 to limit the maximum output torque of the motor, serving as overload protection.

[0208] Accordingly, in this embodiment of the invention, the timer m_sensorRecv_timer(20ms) can be specifically used to handle the reception and parsing of data such as force and temperature sensor data based on the second transmission cycle (e.g., 20 milliseconds), updating the global shared memory for real-time calls by the state machine. Parsing refers to converting raw byte data received through preset bidirectional communication rules, such as the CAN bus, into a high-level data structure that the program can understand and use. For example, the data transmitted on the CAN bus is essentially just a string of bytes. For instance, receiving a frame of data from the bus, the content of which is "01 50 80 60 70 90 00 00" (eight bytes). The parsing process is to provide the meaning of each number byte by byte according to the predetermined protocol. For example, the parsing steps are: first read the first byte and determine the frame header type. If the first byte is 01, it indicates that this is a position data frame. According to the protocol, the second byte is the angle value of the first joint, the third byte is the angle value of the second joint, and so on. At this point, it can be determined that 50 represents 80 degrees for the first joint, 80 represents 128 degrees for the second joint, 60 represents 96 degrees for the third joint, and so on. The final parsed value can be a C language structure variable. For example, after parsing, RightHand_Recv.joint_angle_1 is assigned the value 80, RightHand_Recv.joint_angle_2 is assigned the value 128, and RightHand_Recv.pressure_1 is assigned a pressure value. Therefore, other parts of the program can directly read these variable names to obtain sensor data without considering the underlying byte stream. Updating can be understood as writing the parsed sensor data into global shared memory, overwriting the old values. Each time a new CAN data frame is received and parsed, the system performs an assignment operation, storing the new value in the corresponding variable, and the old value is replaced by the new value. Global shared memory can be understood as member variables of a class, such as the RightHand_Recv and LeftHand_Recv structure variables. Global sharing can be understood as allowing all member functions within a class to access variables declared as members of that class. Specifically, the CAN receive timer function writes to these variables, the state machine control function reads them, and the ROS publish function also reads them. Multiple functions share the same memory region, much like multiple departments share the same information board; hence the term "globally shared memory."

[0209] The connection between global shared memory and the real-time communication closed loop is very close; it serves as a bridge between the communication layer and the control layer. Specifically, the CAN receive timer parses and writes sensor data into the shared memory, completing the "sensing" phase of the closed loop. The state machine control function reads force and temperature data from the shared memory and makes decisions accordingly, completing the "decision-making" phase of the closed loop. The decision result is written to the transmit data buffer, and the CAN transmit timer reads the transmit buffer and issues commands, completing the "execution" phase of the closed loop. Throughout this process, shared memory acts as a relay station for sensed data and decision commands, enabling the communication layer and control layer to be decoupled yet efficiently collaborate. Without shared memory, received data cannot be transmitted to the state machine, and the state machine's decision results cannot be transmitted to the transmit module, making it impossible to form a real-time communication closed loop.

[0210] The purpose of calling the dual timers is to achieve real-time closed-loop control of "perception-decision-execution", ensuring that the dexterous hand can complete a complete control cycle within 20 milliseconds.

[0211] Specifically, after the `m_handSend_timer` timer is invoked, the `Hand_CanSendData_TimerEvent` function is executed. This function packages the control commands currently stored in the transmission buffer (including target angles for the 16 joints, speed commands for the 5 fingers, and torque limits for the 5 fingers) into frames according to preset bidirectional communication rules, and sends them to the dexterous hand hardware via a bus protocol based on these preset bidirectional communication rules. This completes the execution phase, providing feedback on what actions the fingers should perform.

[0212] After the `m_sensorRecv_timer` timer is invoked, the `Hand_CanRecvData_TimerEvent` function is executed. This function receives sensor data frames transmitted from the dexterous hand via a bus protocol with preset bidirectional communication rules. It determines the data type (position feedback, pressure feedback, or temperature feedback) based on the frame header, then parses out the specific values ​​and stores them in global shared memory. This completes the sensing phase, reflecting what the finger is actually doing and its current state.

[0213] The state machine control function runs in a separate thread, periodically reading sensing data from shared memory and updating the instructions in the send buffer after processing the decision logic. Since the two timers trigger every 20 milliseconds, this ensures that a complete loop of "send instruction - execute action - sense state - update decision" is completed every 20 milliseconds, achieving true real-time closed-loop control.

[0214] In one embodiment of the present invention, the receiving end parses the sensor data according to the new frame header 0x06 and 0x07. If an anomaly is detected (such as a sudden change in pressure of a finger or a sudden rise in temperature), an alarm message can be immediately issued through ROS Topic. The alarm is not just issuing a warning signal. Its function is reflected in triggering subsequent protection actions and state switching. Specifically, it has the following four aspects: (1) The alarm is used to trigger emergency protection actions: When a sudden change in pressure of a finger is detected, the system not only issues an alarm message, but also immediately stops the movement of the joint and triggers a return action to prevent damage to the screen. When a sudden rise in temperature is detected, the system will slow down or enter a heat dissipation state in advance, instead of waiting for the temperature to reach the threshold before taking action. (2) The alarm is used to notify the upper-level decision-making system: After the alarm message is issued through ROS Topic, the upper-level monitoring node can receive this information and make corresponding decisions according to the alarm level, such as pausing the current task, switching to safe mode, or notifying the user to check. (3) Alarms are used for fault diagnosis and log recording: The system records alarm information to the log file. Maintenance personnel can diagnose problems by viewing the alarm history, such as which finger is frequently overloaded or which joint has the fastest temperature rise, so as to carry out targeted maintenance. (4) Alarms are used for human-machine interaction prompts: Alarm information can be transmitted to the user interface to inform the user of the current status of the dexterous hand through screen display or sound prompts, such as "abnormal finger pressure, please check the keyboard" or "joint temperature is too high, speed is being reduced".

[0215] In one embodiment of the present invention, the external control interface can be fully utilized to subscribe to the right_hand_control and left_hand_control topics based on the ROS system, allowing external nodes to directly issue 16-dimensional joint angle commands, covering the preset typing action sequence, and realizing remote control or AI real-time control.

[0216] In ROS, a topic can be a publish-subscribe communication mechanism. Think of a topic as a "radio channel." Publishers are like radio hosts, broadcasting information on a specific channel; subscribers are like listeners, tuning in to hear the broadcast. Publishers and subscribers don't need to know about each other's existence, only the channel name. Subscription is done using the `subscribe` function provided by ROS. For example, the subscription statement is as follows:

[0217] sub_right_hand=m_hand_can_nh.subscribe(“ / right_hand_control”,10,&Can_Communication::RightHandCtrl_subCallback, this)

[0218] The above subscription statement can be understood as follows: This node will subscribe to the topic named " / right_hand_control". When a message is published to this topic, the system will automatically call the RightHandCtrl_subCallback function to handle the message. The parameter 10 represents the length of the message queue; when messages arrive too quickly to process in time, a maximum of 10 messages will be cached.

[0219] In summary, the entire system's workflow can be described as a cycle of three stages: "perception-decision-execution." In the perception stage, sensors on the dexterous hand collect data. Fingertip pressure sensors detect contact force, joint encoders detect angle position, and temperature sensors detect motor temperature. This data is uploaded to the dexterous hand's CAN node via frames 0x06 and 0x07 of the CAN bus. After parsing, the node stores the data in global shared memory. In the decision-making stage, the state machine makes decisions based on the perceived data. When the detected pressure exceeds 0.5 Newtons, the state machine transitions from a pressure-down state to a force-control state; when the detected pressure is below the minimum threshold, the joint angle is fine-tuned to increase the pressure; when the detected temperature exceeds the first-level temperature warning line (in degrees Celsius), the speed is reduced; when the temperature exceeds the second-level temperature warning line, a forced transition to a cooling state is initiated. In the execution stage, the decision results are converted into control commands. The state machine writes the determined target angle, speed, and torque into a transmission buffer. Every 20 milliseconds, the CAN transmission timer sends these commands to the dexterous hand hardware via frames 0x01, 0x04, and 0x05, conforming to preset bidirectional communication rules, driving the motor to rotate. The entire process cycles once every 20 milliseconds, achieving real-time closed-loop control.

[0220] The screen touch optimization method for dexterous hands described in the above embodiments of the present invention can be further illustrated by the following two implementation scenarios.

[0221] For implementation scenario one, the anthropomorphic typing (force adaptive + thermal protection) action flow is as follows: steps S11-S15:

[0222] In step S11, initialization: hand reset, temperature self-check normal. In step S12, downward pressure phase: state machine enters Key_Press step, index finger joint moves downward. In step S13, force control intervention: real-time monitoring of fingertip pressure. When (Upon contact with the screen), the system automatically switches to force control mode, maintaining pressure between 8N ± 0.1N for 50ms (simulated screen trigger time). In step S14, during the return phase: pressure is released, and the joint quickly rises. In step S15, during thermal adaptation: if continuous typing causes the joint temperature to rise to 48℃, the system automatically extends the "press-return" cycle from 200ms to 400ms to reduce heat generation. High-speed typing resumes only after the temperature drops.

[0223] For implementation scenario two, the action flow of browsing Douyin (touch screen swipe + double tap to like + long press to save) is as follows: sub-scenario AC.

[0224] Using the same state machine framework as Scenario 1, complex mobile phone interactions can be achieved simply by switching parameter configurations.

[0225] Sub-scene A: The video swipe action flow is as follows: steps S211-S215:

[0226] In step S211, holding: the thumb and four fingers encircle the phone, maintaining a clamping force of 2.5N (anti-slip and screen-shatterproof). In step S212, contact: the index fingertip contacts the lower center of the screen, and the force sensor confirms the contact pressure. (Touch). In step S213, swipe: The state machine triggers the Swipe_Up sequence. The index finger joint moves upward 200mm rapidly within 150ms (simulating a "flicking" motion), with a speed curve that is trapezoidal (acceleration - constant speed - sudden stop). In step S214, release: After reaching the top, force feedback is used to confirm that the screen's reaction force has disappeared, and the finger is quickly lifted away to avoid accidental touches. In step S215, temperature control: After 50 consecutive swipes, if a motor temperature rise is detected, the swipe speed is automatically reduced to extend the video viewing interval.

[0227] Sub-scene B: The Double Tap action flow is as follows: Step S221:

[0228] In step S221, timing control: The time interval is precisely controlled using a state machine counter. First click: Quickly press down to 0.6N, hold for 30ms, then release. Wait: The state machine pauses for Δ=150ms. Second click: Repeat the above actions.

[0229] In sub-scenario B, force feedback confirmation occurs when the phone has haptic feedback enabled. The force sensor detects a slight screen vibration as a confirmation signal for a successful "like" (optional). The force feedback confirmation signal refers to the weak vibration generated by the vibration motor when the phone detects a valid click or touch operation. When a dexterous hand clicks the phone screen, the linear or rotor motor inside the phone generates a brief vibration, which is transmitted through the screen to the fingertip. The thin-film pressure sensor installed on the fingertip has sufficient sensitivity and response speed to detect this weak vibration signal. Specifically, the generation and detection process of the force feedback confirmation signal is as follows: For example, when a dexterous hand clicks the screen with a force of 0.6 Newtons, the phone's touch controller detects a change in capacitance, confirming a valid click event. Subsequently, the phone system calls the vibration motor drive interface, and the motor generates a vibration with a frequency of approximately 150 to 200 Hz and a duration of approximately 20 to 50 milliseconds within milliseconds. This vibration is transmitted through the phone casing to the screen glass, and then through the contact surface of the fingertip to the sensitive unit of the pressure sensor. The electrical signal output by the pressure sensor will contain a fluctuating component that matches the vibration frequency. The system extracts this characteristic frequency fluctuation by bandpass filtering the force signal, thus confirming that the phone has responded to the tap and the "like" was successful. This feature is optional because not all phones have haptic feedback enabled, and not all scenarios require confirmation. However, in scenarios requiring high reliability, such as remote operation or automated testing, this feature can serve as a secondary confirmation mechanism to improve operational reliability.

[0230] It should be noted that in this embodiment of the invention, the technical means of sub-scene A and sub-scene B can be completely integrated, and they are essentially applications of the same set of technical means under different parameter configurations. Specifically, firstly, they can share the same force-position hybrid control framework: whether sliding or double-clicking, both rely on force sensors to detect contact, force control mode to maintain the target force, and state machine to manage the action timing. The only difference between the two is the parameter configuration: the target force in the sliding scene is 0.3 Newtons, the duration is 150 milliseconds, and the speed curve is trapezoidal; the target force in the double-click scene is 0.6 Newtons, each click lasts for 30 milliseconds, the interval between two clicks is 150 milliseconds, and the speed curve is pulse-shaped. However, the underlying execution is the same force-position hybrid control algorithm and the same state machine framework. Secondly, they can share the same state machine framework: the state sequence of the sliding scene is: idle state → press down state → force control state → sliding dynamic → disengagement state → return state. The state sequence of the double-click scene is: idle state → press down state → force control state → return state → waiting state → press down state → force control state → return state. As can be seen, the double-tap scene simply inserts a "waiting state" and a second repetitive action into the state sequence of the swipe scene; the core pressing state, force control state, and return state can be completely identical. Finally, the two can be dynamically switched and combined: in actual Douyin interactions, a user might first swipe to switch videos, then double-tap to like, and then long-press to favorite. The state machine provided in this embodiment of the invention can support such continuous switching, requiring only reconfiguration of parameters based on the target of the next action after each action is completed, without needing to restart the state machine or reload the code.

[0231] Therefore, the technical connection between the two technical scenarios A and B can be understood: the sliding scenario verifies the effectiveness of force-position hybrid control in continuous motion scenarios, and the double-click scenario verifies the effectiveness of force-position hybrid control in discrete motion scenarios. Together, they demonstrate the universal adaptability of the same technical framework to different interaction scenarios. This fully verifies the "construction of a universal state machine framework supporting multiple scenarios such as typing and browsing Douyin" in the embodiments of this invention.

[0232] Sub-scene C: The action flow for long-press to favorite / not interested is as follows: steps S231-S233:

[0233] In step S231, the action is: the index finger presses a specific area of ​​the screen and remains stationary. In step S232, the time is maintained: the state machine counter show_count is incremented, maintaining the pressing posture for more than 1.5 seconds. In step S233, the force is maintained: if pressure fluctuations occur due to mechanical vibration during this period, the PID algorithm fine-tunes the joint angle in real time to ensure that the pressure is always maintained above 0.7N, guaranteeing the menu pops up.

[0234] Therefore, based on the description of the aforementioned technical solutions, the three technical scenarios of swiping to switch videos, double-tapping to like, and long-pressing to favorite are not three independent technical solutions, but rather three application instances of the same technical framework under different parameter configurations. The specific connections are reflected in the following four aspects:

[0235] (1) Sharing the same force-position hybrid control core: All three scenarios rely on force sensors to detect contact, force control mode to maintain the target force, and state machine to manage the timing of actions. The sliding scenario requires maintaining a contact force of 0.3 Newtons to perform position sliding, the double-click scenario requires two short clicks with a force of 0.6 Newtons, and the long-press scenario requires maintaining a force of more than 0.7 Newtons for 1.5 seconds. Although the target force and duration are different, the underlying execution is the same force-position hybrid control algorithm, which uses PID to adjust the joint angle to maintain the target force and uses a state machine to manage the stage transitions of the action.

[0236] (2) They share the same state machine framework. The state sequences of the three scenarios have the same underlying structure. The state sequence of the sliding scenario is: idle state → press down state → force control state → slide dynamic → disengagement state → return state. The state sequence of the double-tap scenario is: idle state → press down state → force control state → return state → wait state → press down state → force control state → return state. The state sequence of the long press scenario is: idle state → press down state → force control state → long press maintenance state → return state. It can be seen that the core of each scenario is the basic action unit of "press down state + force control state + return state". The difference between the different scenarios is only that the subsequent states after the force control state are different - after sliding there is a slide dynamic, after double-tap there is a wait state and a repeating action, and after long press there is a long press maintenance state.

[0237] (3) Force sensing data is the common driving force for all three scenarios. In all three scenarios, each critical transition of the state machine depends on force sensing data. The transition condition from the pressure state to the force control state is that the contact force exceeds the threshold (0.3N for sliding, 0.6N for double-clicking, and 0.7N for long-pressing). The transition conditions from the force control state to subsequent states include force maintenance time meeting the standard, force release detection, and force overload protection. Without force sensing data, precise force control cannot be achieved in any of the three scenarios.

[0238] (4) The three scenarios together demonstrate the universality of the technical framework. The success of a single scenario only proves that the technical solution is effective in a specific scenario, while the fact that three significantly different scenarios can be implemented under the same framework proves that the technical framework of this invention has universality and scalability. This is precisely the embodiment of the invention's purpose in the disclosure: "to build a universal state machine framework that supports multiple scenarios such as typing and browsing Douyin." From the perspective of technical improvement, this invention does not optimize for a specific scenario, but rather establishes a unified framework that can adapt to multiple interaction scenarios. It can switch between different scenarios through parameter configuration without modifying the underlying code. The connection between the three technical scenarios is essentially the different performances of this unified framework under different parameter configurations.

[0239] First, all three sub-scenes of the Douyin (TikTok) browsing scenario employ a refined control strategy based on force feedback, which is the core embodiment of the "force-position hybrid touch screen interaction algorithm":

[0240] During the up-and-down swiping to switch videos, the system employs a "contact-accelerated swiping-inertial release" action sequence. In the contact phase, force detection confirms the finger has touched the screen; in the swiping phase, a trapezoidal velocity curve controls position movement; and in the release phase, force feedback confirms the disappearance of the screen's reaction force before the finger is lifted. This hybrid strategy of "force detection confirming contact, position control executing swiping, and force detection confirming release" solves the problem that traditional pure position control cannot simulate the feel of human swiping. During the double-tap to like event, the target force for both clicks is set to 0.6 Newtons, lasting 30 milliseconds. This parameter was determined experimentally, ensuring reliable touchscreen response without damaging the screen due to excessive force. The 150-millisecond interval between the two clicks is also precisely controlled by a state machine counter, demonstrating the combination of force control and timing control. During the long-press to favorite event, the system needs to maintain a contact force of at least 0.7 Newtons for 1.5 seconds. During this time, a PID algorithm fine-tunes the joint angle in real time to compensate for pressure fluctuations caused by mechanical vibration. This is a typical application of force-position hybrid control in long-term maintenance scenarios—position fine-tuning serves the goal of force maintenance.

[0241] Secondly, all three sub-scenes in the Douyin (TikTok) browsing scenario rely on force sensing data as the condition for state machine transitions. This is a typical application of directly using force sensing data as the condition for state machine transitions, which can fully demonstrate the dynamic state machine driven by multimodal sensing.

[0242] During the up-and-down swiping to switch videos, force control intervention is triggered when the contact pressure exceeds 0.3 Newtons. When the system detects this pressure value, the state machine transitions from a hovering state to a swiping state, initiating the upward swiping sequence. During the swiping process, when the screen's reaction force disappears, the state machine transitions to a release state, and the finger is quickly lifted off the screen. Both transitions rely on force sensor data for judgment, rather than fixed time or position. During the double-tap to like process, the state machine waits 150 milliseconds after the first tap before executing the second tap. During this wait, the state machine is in a paused state, but the force sensor remains active. If unexpected pressure fluctuations are detected within these 150 milliseconds, the state machine readjusts. After the double-tap, the force sensor detects pressure release, and the state machine returns to the idle state. During the long-press to favorite process, the state machine needs to maintain the pressing posture for more than 1.5 seconds. If mechanical vibration causes pressure fluctuations below 0.7 Newtons, the state machine will not transition but will instead use a PID algorithm to fine-tune the joint angle to maintain pressure. This demonstrates that "force sensing data serves as a condition for the state machine not to jump"—when the pressure is insufficient, the state machine remains in the current state and continues to adjust, rather than exiting prematurely.

[0243] Furthermore, the aforementioned implementation scenario one and real-time scenario two can also demonstrate an active frequency reduction mechanism based on temperature trends:

[0244] For example, the temperature control feature in the Douyin (TikTok) browsing scenario demonstrates an active temperature protection mechanism. The description of swiping up and down to switch videos states: "After 50 consecutive swipes, if a motor temperature rise is detected, the swiping speed will be automatically reduced, extending the video viewing interval." This is a specific application of the active speed reduction mechanism in the Douyin browsing scenario. Specifically, the system monitors the temperature change trend during continuous swiping. If a sustained temperature rise is detected, even before reaching the first-level warning threshold of 45 degrees Celsius, the system will actively reduce the swiping speed, for example, extending the swiping time from 150 milliseconds to 200 milliseconds, while also increasing the interval between swipes. This preventative speed reduction aims to control heat generation and achieve thermal balance before the temperature reaches a dangerous level, rather than waiting until the temperature exceeds the limit before taking emergency measures. This is fundamentally different from the passive protection method of "reducing speed only when the temperature exceeds the limit" in traditional solutions.

[0245] Furthermore, the two implementation scenarios mentioned above (including three sub-scenarios) can fully demonstrate the functional implementation of the high-bandwidth multi-frame fusion communication protocol:

[0246] The force and temperature control functions implemented in Douyin (TikTok) rely on a real-time data transmission channel provided by the communication protocol. During swiping and tapping, fingertip pressure sensor data is uploaded via frame 0x06 at a high frequency of 10 milliseconds, ensuring the system can promptly detect a light touch of 0.3 Newtons and a tap of 0.6 Newtons. Without this high-frequency channel, the force feedback delay would be too large, and the finger might have already applied too much force before the contact is detected. During temperature control, joint temperature data is uploaded via frame 0x07. Although the period is 100 milliseconds, it is sufficient for the relatively slow temperature changes. The system determines whether to reduce the swiping speed based on the temperature data, demonstrating the synergy between temperature and force control.

[0247] Finally, the synergistic relationship between the above-mentioned technical means in the relevant implementation scenarios of the present invention is reflected in the following:

[0248] In the Douyin (TikTok) browsing scenario, a complete "perception-decision-execution" chain is formed. The communication protocol handles the perception stage, uploading force data in real-time at frame 0x06 and temperature data at frame 0x07. The dynamic state machine handles the decision-making stage, determining when to switch to a sliding state and when to switch to a detached state based on force data, and whether to reduce speed based on temperature data. The force-position hybrid algorithm handles the execution stage, finely controlling the force and position during swiping, clicking, and long-pressing. An active frequency reduction mechanism is implemented throughout the entire process to ensure that prolonged Douyin browsing does not lead to overheating and shutdown.

[0249] In summary, the screen touch optimization method for dexterous hands provided in this embodiment of the invention can at least partially solve the problems of lack of haptic feedback, neglect of thermal management, and limited communication bandwidth in screen touch processes in dexterous hand control-related technologies, and therefore can achieve at least one of the following technical effects:

[0250] First, a dynamic state machine driven by multimodal sensing.

[0251] The triggering of the secondary protection mechanism relies on temperature sensor data, which is a typical manifestation of "using temperature sensor data directly as the state machine transition condition." Compared to the traditional method in existing technologies where state machine transitions rely solely on counters and time, in this embodiment of the invention, exceeding the secondary temperature warning threshold is itself the criterion for state machine transition. When the temperature monitoring thread parses the temperature data returned in frame 0x07 and finds that it exceeds the secondary temperature warning threshold, the system immediately forces the state machine to transition from its current state to the heat dissipation state, Safe_Mode. This achieves a leap from "open-loop preset" to "closed-loop adaptive"—the system no longer mechanically executes preset action sequences but actively adjusts its behavior based on its own temperature state.

[0252] The multimodal sensor-driven dynamic state machine of this invention is the first to directly use force and temperature sensor data as state machine transition conditions, achieving a leap from "open-loop preset" to "closed-loop adaptive," significantly improving the smoothness and safety of interaction. Regarding the "advantage of using force and temperature sensor data as transition conditions for the first time": compared to traditional technologies where state machine transition conditions only involve counters and time (e.g., automatic return to the starting position after 30 milliseconds of pressure, regardless of whether the finger touched the keyboard or how hard it was pressed), the method of this invention can use force and temperature sensor data as transition conditions, with advantages in four aspects: First, precise contact sensing: only when the detected pressure exceeds 0.5 Newtons is it considered that the finger has touched the keyboard, avoiding empty presses. Second, force adaptation: the state machine does not transition when the pressure is insufficient, continuing to increase the pressure until the target pressure is reached. Third, safety protection: when the pressure exceeds the maximum safety value, it immediately transitions back to the starting state to prevent damage to the device. Fourth, temperature self-adaptation: when the temperature exceeds 45 degrees Celsius, it switches to a deceleration state; when it exceeds 60 degrees Celsius, it switches to a heat dissipation state. These switching conditions do not exist in existing technologies.

[0253] Closed-loop adaptation refers to a system continuously collecting sensor feedback during execution and adjusting action parameters in real time based on the feedback. The force control adaptive logic in this embodiment of the invention: the system detects fingertip pressure in real time; when the pressure is insufficient, it automatically increases the downward pressure; when the pressure is excessive, it immediately returns to its original position. Temperature monitoring is also a form of closed-loop adaptation: the system detects joint temperature in real time; if the temperature is high, it reduces speed; if the temperature drops, it resumes speed. The essence of closed-loop adaptation is "doing and observing, adjusting as needed." Compared to the open-loop preset mechanism in traditional technologies, the adaptive closed-loop mechanism of this embodiment is more flexible, safer, and more human-like, making it easier to achieve refined, human-like interactions.

[0254] For example, in typing scenarios, the system monitors fingertip pressure in real time. When the pressure exceeds 0.5 Newtons, the state machine jumps to the force control state, demonstrating that force is the jump condition. In thermal protection, the system monitors joint temperature in real time. When the temperature exceeds the first-level temperature warning line (e.g., 45 degrees Celsius), the state machine jumps to the deceleration state, demonstrating that temperature is the jump condition. In TikTok scrolling scenarios, the system jumps to the disengagement state when it detects the disappearance of the screen's reaction force, again demonstrating that force is the jump condition. The force-position hybrid touchscreen interaction algorithm can propose a refined "contact-slide-disengagement" control strategy based on force feedback for scenarios like TikTok scrolling, solving the problem that traditional position control cannot simulate human touch.

[0255] Second, an active frequency reduction mechanism based on temperature trends.

[0256] Secondary protection is the final stage of the active frequency reduction mechanism. The temperature control strategy in this embodiment of the invention is divided into two levels: Primary warning (e.g., exceeding the primary temperature warning threshold of 45 degrees Celsius) slows down the speed, actively reducing speed to maintain operation; secondary protection (e.g., exceeding the secondary temperature warning threshold of 60 degrees Celsius) forces the system into a heat dissipation state and cuts off the enable function, an emergency shutdown to protect the hardware. These two mechanisms together constitute a complete active temperature protection system, significantly different from traditional over-temperature cutoff schemes (which often lack a primary warning and directly shut down the system), resulting in task interruption and inability to automatically recover. Furthermore, secondary protection performs a slow reset to a heat dissipation posture before cutting off the enable function, protecting the hardware and creating conditions for subsequent automatic recovery. Therefore, the active frequency reduction mechanism based on temperature trends, compared to traditional over-temperature cutoff methods, can achieve thermal balance while ensuring task continuity by dynamically adjusting the frequency and speed of the action.

[0257] Third, high-bandwidth multi-frame fusion communication protocol.

[0258] The implementation of secondary protection relies on the high real-time transmission of temperature data. This embodiment of the invention extends bidirectional communication rules, such as the CAN protocol, and designs a dedicated 0x07 frame header for transmitting high-precision temperature data, with each frame carrying temperature values ​​for five to ten joints. The temperature monitoring thread parses the 0x07 frame data in real time to promptly detect temperature exceedances and trigger secondary protection. With the help of this communication protocol conforming to preset bidirectional communication rules, temperature data can be uploaded to the host computer in a timely manner, thus ensuring the timely and effective triggering of secondary protection. Therefore, the improvement of the frame data in the communication protocol with preset bidirectional communication rules is the fundamental guarantee for the implementation of the secondary protection mechanism.

[0259] The high-bandwidth multi-frame fusion communication protocol involved, conforming to preset bidirectional communication rules, supports a CAN-like protocol stack capable of concurrent transmission of four-dimensional data (position, torque, force, and temperature), achieving full-state synchronization of 16 degrees of freedom within a 20ms cycle. Full-state synchronization can be understood as completing the instruction issuance and state acquisition of all 16 degrees of freedom within a 20ms control cycle, achieving real-time synchronization of uplink and downlink data. Specifically, this is manifested as follows: First, full synchronization of all 16 degrees of freedom. A dexterous hand has 16 degrees of freedom, including four joints (Pitch, Yaw, Roll, and Tip) of the thumb, and three joints (Pitch, Yaw, and Tip) of the other four fingers. The CAN protocol of this embodiment can use multiple frames such as 0x01, 0x04, and 0x05 to send all target angles, velocities, and torques of the 16 degrees of freedom within 20ms, while simultaneously receiving the actual angles of the 16 degrees of freedom and 5-channel pressure and 5-10-channel temperature feedback. Within each control cycle, the system can acquire the complete hand state, without any missing or delayed data for any joint. Second, concurrent synchronization of multi-dimensional data. The communication protocol of this embodiment supports concurrent transmission of four-dimensional data: position, torque, force, and temperature. Position and torque data are transmitted via frames 0x01, 0x04, 0x05, and 0x02 at a 20-millisecond cycle; force data is transmitted via frame 0x06 at a 10-millisecond cycle; and temperature data is transmitted via frame 0x07 at a 100-millisecond cycle. Although the transmission frequencies of different data types differ, they can share the bus bandwidth in an orderly manner within the time window without conflict. The state machine can obtain the latest force and temperature data in each decision cycle, achieving effective synchronization of multi-dimensional data. Third, bidirectional synchronization of uplink and downlink data. Within the same 20-millisecond cycle, the system both sends control commands downwards and receives sensor feedback upwards. The transmit and receive timers run independently, with transmission and reception proceeding in parallel without blocking. This means that within 20 milliseconds after issuing the "depress down by 5 degrees" command, the system can receive feedback that "the actual pressure was reduced by 4.8 degrees," forming a complete closed-loop control. This bidirectional synchronization is the foundation of real-time control.

[0260] Bandwidth verification: At a baud rate of 1 Mbps, the CAN bus 2.0 transmits each frame in approximately 100 to 130 microseconds. Approximately 150 to 200 frames can be transmitted within 20 milliseconds. In this embodiment of the invention, approximately 6 control commands are transmitted, approximately 2 force data frames are received, and 0.2 temperature data frames are received every 20 milliseconds, totaling approximately 8 frames, with a bandwidth utilization rate of less than 10%. Therefore, this not only effectively guarantees full-state synchronization but also maintains sufficient synchronization margin.

[0261] Based on the above-described screen touch optimization method for dexterous hands, this invention also provides a screen touch optimization device for dexterous hands. The device will be described in detail below with reference to Figure 4.

[0262] Figure 4A The diagram illustrates a structural block diagram of a screen touch optimization device 401 for a dexterous hand according to an embodiment of the present invention.

[0263] like Figure 4A As shown, the screen touch optimization device 401 for dexterous hands in this embodiment includes a touch operation module 411, a first data receiving module 412, a first state switching module 413, and a first state execution module 414.

[0264] The touch operation module 411 is used to respond to the received work operation command and perform a target touch operation on the target screen based on the starting position state of the dexterous hand and the current working state of the state machine. In one embodiment, it can be used to execute the operation S211 described above, which will not be repeated here.

[0265] The first data receiving module 412 is used to receive, based on a preset bidirectional communication rule, current joint temperature data transmitted in a first transmission cycle and current preliminary temperature data transmitted in a third transmission cycle during the target touch operation of the dexterous hand. In one embodiment, it can be used to perform the operation S212 described above, which will not be repeated here.

[0266] The first state switching module 413 is used to switch the current operating state of the state machine to a deceleration state or a heat dissipation state based on the current joint temperature data, the current preliminary temperature data, and the preset warning temperature range. In one embodiment, it can be used to execute the operation S213 described above, which will not be repeated here.

[0267] The first state execution module 414 is used to perform screen touch optimization applied to the dexterous hand based on at least one of the deceleration state and the heat dissipation state. In one embodiment, it can be used to perform the operation S214 described above, which will not be repeated here.

[0268] Figure 4B The diagram illustrates a structural block diagram of a screen touch optimization device 402 for a dexterous hand according to another embodiment of the present invention.

[0269] like Figure 4B As shown, the screen touch optimization device 402 for dexterous hands in this embodiment includes a second data receiving module 421, a second state switching module 422, and a second state execution module 423.

[0270] The second data receiving module 421 is used to receive current force sensing data transmitted in a second transmission cycle based on a preset bidirectional communication rule during the target touch operation of the dexterous hand. In one embodiment, it can be used to perform the operation S221 described above, which will not be repeated here.

[0271] The second state switching module 422 is used to adjust the downward pressure of the dexterous hand or switch the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range. In one embodiment, it can be used to perform the operation S222 described above, which will not be repeated here.

[0272] The second state execution module 423 is used to implement screen touch optimization for dexterous hands based on at least one of the pressure amount and the return state. In one embodiment, it can be used to perform the operation S223 described above, which will not be repeated here.

[0273] According to embodiments of the present invention, any multiple modules among the touch operation module 411, the first data receiving module 412, the first state switching module 413, and the first state execution module 414, or the second data receiving module 421, the second state switching module 422, and the second state execution module 423, can be combined into one module, or any one of these modules can be split into multiple modules. Alternatively, at least some of the functions of one or more of these modules can be combined with at least some of the functions of other modules and implemented in one module. According to embodiments of the present invention, at least one of the touch operation module 411, the first data receiving module 412, the first state switching module 413, and the first state execution module 414, or the second data receiving module 421, the second state switching module 422, and the second state execution module 423, can be at least partially implemented as hardware circuits, such as field-programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems-on-a-chip, systems-on-a-substrate, systems-on-package, application-specific integrated circuits (ASICs), or any other reasonable means of integrating or packaging circuits, or implemented in hardware or firmware, or in any one of software, hardware, and firmware implementations, or in a suitable combination of any of these. Alternatively, at least one of the touch operation module 411, the first data receiving module 412, the first state switching module 413, and the first state execution module 414, or the second data receiving module 421, the second state switching module 422, and the second state execution module 423, can be at least partially implemented as a computer program module, which can perform corresponding functions when the computer program module is run.

[0274] Figure 5 A block diagram of an electronic device adapted to implement a screen touch optimization method for dexterous hands, according to an embodiment of the present invention, is illustrated.

[0275] The electronic device provided in the embodiments of the present invention includes one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors execute the screen touch optimization method applied to dexterous hands described above.

[0276] like Figure 5 As shown, an electronic device 500 according to an embodiment of the present invention includes a processor 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage portion 508 into a random access memory (RAM) 503. The processor 501 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 501 may also include onboard memory for caching purposes. The processor 501 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0277] RAM 503 stores various programs and data required for the operation of electronic device 500. Processor 501, ROM 502, and RAM 503 are interconnected via bus 504. Processor 501 executes various operations of the method flow according to embodiments of the present invention by executing programs in ROM 502 and / or RAM 503. It should be noted that the programs may also be stored in one or more memories other than ROM 502 and RAM 503. Processor 501 may also execute various operations of the method flow according to embodiments of the present invention by executing programs stored in said one or more memories.

[0278] According to an embodiment of the present invention, the electronic device 500 may further include an input / output (I / O) interface 505, which is also connected to a bus 504. The electronic device 500 may also include one or more of the following components connected to the I / O interface 505: an input section 506 including a keyboard, mouse, etc.; an output section 507 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to the I / O interface 505 as needed. A removable medium 511, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 510 as needed so that computer programs read from it can be installed into the storage section 508 as needed.

[0279] The present invention also provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, cause the processor to perform the above-described screen touch optimization method for dexterous hands.

[0280] The computer-readable storage medium may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of the present invention.

[0281] According to embodiments of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium, such as including, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In the present invention, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of the present invention, the computer-readable storage medium may include ROM 502 and / or RAM 503 and / or one or more memories other than ROM 502 and RAM 503 described above.

[0282] Embodiments of the present invention also include a computer program product comprising a computer program that, when executed by a processor, implements the above-described screen touch optimization method for dexterous hands.

[0283] The computer program includes program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code is used to enable the computer system to implement the methods provided in the embodiments of the present invention.

[0284] When the computer program is executed by the processor 501, it performs the functions defined in the system / apparatus of this invention. According to embodiments of the invention, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0285] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and may be downloaded and installed via the communication section 509, and / or installed from a removable medium 511. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0286] In such an embodiment, the computer program can be downloaded and installed from a network via communication section 509, and / or installed from removable medium 511. When the computer program is executed by processor 501, it performs the functions defined in the system of this embodiment of the invention. According to embodiments of the invention, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0287] According to embodiments of the present invention, program code for executing the computer programs provided in the embodiments of the present invention can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0288] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0289] Furthermore, all actions involving the acquisition of information, signals, or data in this invention are carried out in compliance with the relevant data protection laws, regulations, and policies of the country where the invention is located, and with the authorization granted by the owner of the corresponding device.

[0290] Those skilled in the art will understand that the features described in the various embodiments and / or claims of the present invention can be combined or combined in various ways, even if such combinations or combinations are not explicitly described in the present invention. In particular, the features described in the various embodiments and / or claims of the present invention can be combined or combined in various ways without departing from the spirit and teachings of the present invention. All such combinations and / or combinations fall within the scope of the present invention.

[0291] The embodiments of the present invention have been described above. However, these embodiments are merely illustrative and not intended to limit the scope of the invention. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. The scope of the invention is defined by the appended claims and their equivalents. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of the invention, and all such substitutions and modifications should fall within the scope of the invention.

Claims

1. A screen touch optimization method for dexterous hands, characterized in that, include: During the target touch operation of the dexterous hand, based on the preset two-way communication rules, the current force sensing data transmitted in the second transmission cycle is received; Based on the current force sensing data and the preset force control threshold range, adjust the downward pressure of the dexterous hand or switch the current working state of the state machine to the return state; The screen touch optimization for dexterous hands is achieved based on at least one of the pressure amount and return state.

2. The method according to claim 1, characterized in that, During the target touch operation of the dexterous hand, the current force sensing data received in the second transmission cycle, based on a preset two-way communication rule, includes: Based on a preset force sensing frame format that conforms to the preset bidirectional communication rules, the current force sensing data transmitted in the second transmission cycle is received according to a preset dual timing mechanism.

3. The method according to claim 2, characterized in that, The preset force sensing frame format is used to define the data transmission format of the frame header, thumb pressure value, index finger pressure value, middle finger pressure value, ring finger pressure value, little finger pressure value, and check bit.

4. The method according to claim 3, characterized in that, The step of adjusting the downward pressure of the dexterous hand or switching the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range includes: Based on the force sensing deviation determined by the current force sensing data and the preset force control threshold range, and combined with the preset force sensing deviation coefficient, the downward pressure of the dexterous hand is determined. This is used to adjust the downward pressure of the dexterous hand fingers on the target screen based on the downward pressure and the first preset limit value while maintaining the current working state of the state machine.

5. The method according to claim 3, characterized in that, The step of adjusting the downward pressure of the dexterous hand or switching the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range includes: Based on the force sensing deviation determined by the current force sensing data and the preset force control threshold range, and combined with the contact stiffness coefficient, the downward pressure of the dexterous hand is determined. This is used to adjust the downward pressure of the dexterous hand fingers on the target screen based on the downward pressure and the second preset limit value while maintaining the current working state of the state machine.

6. The method according to claim 4 or 5, characterized in that, The step of adjusting the downward pressure of the dexterous hand or switching the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range further includes: When switching the current working state of the state machine to the return state, based on the current force sensing data and the preset force control threshold range, the preset return state execution sequence is invoked to perform return control of the dexterous hand to the starting position state.

7. The method according to claim 4 or 5, characterized in that, The step of adjusting the downward pressure of the dexterous hand or switching the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range further includes: In response to the received current joint temperature data, current preliminary temperature data, and preset warning temperature range, the dexterity hand adjusts its downward pressure or switches the current working state of the state machine to the return state based on the current force sensing data and preset force control threshold range. Then, based on the downward pressure or return state, the state machine switches its current working state to the deceleration state or the heat dissipation state.

8. A screen touch optimization device for dexterous hands, characterized in that, include: The second data receiving module is used to receive current force sensing data transmitted in the second transmission cycle based on a preset two-way communication rule during the target touch operation of the dexterous hand. The second state switching module is used to adjust the downward pressure of the dexterous hand or switch the current working state of the state machine to the return state based on the current force sensing data and the preset force control threshold range. The second state execution module is used to implement the screen touch optimization applied to the dexterous hand based on at least one of the pressure amount and the return state.

9. An electronic device, comprising: One or more processors; Memory, used to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors perform the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon executable instructions that, when executed by a processor, cause the processor to perform the method of any one of claims 1 to 7.

11. A computer program product comprising a computer program that, when executed by a processor, implements the method of any one of claims 1 to 7.

Images