Modular multi-mode tactile sensing method for robot and application of modular multi-mode tactile sensing method

Abstract

The invention discloses a modular multi-mode tactile sensing method for a robot and application thereof, and the method comprises the steps: detecting an external physical quantity through a plurality of distributed sensing units, each sensing unit comprising a tactile array sensor, a thermocouple and a flexible tension sensor; the electric signals output by the corresponding sensing units are conditioned through a plurality of multi-mode signal conditioning modules so as to be converted into analog voltage signals; the multi-mode signal conditioning module is provided with a superior input interface and a subordinate output interface; the high-speed data processing and communication module is used for providing a power supply and a line scanning control signal for all the multi-mode signal conditioning modules, synchronously acquiring an analog voltage signal output by the multi-mode signal conditioning modules, converting the analog voltage signal into a digital signal, packaging the digital signal into a data packet, and transmitting the data packet to an upper system; and an interrupt driving mechanism is adopted in the synchronous acquisition step. The performance of the robot touch sensing system is greatly improved, the expansibility of the robot touch sensing system is improved, and the cost is reduced.

Description

Technical Field

[0001] This invention relates to the field of robot sensing technology, and in particular to a modular multimodal tactile sensing method for robots and its applications. Background Technology

[0002] The successful execution of dexterous robotic operations relies heavily on the rich sensory information and real-time feedback that the device can acquire. In recent years, significant progress has been made in the mechanical structure design of robotic hands; however, the development of external sensing systems required to support fine manipulation still lags behind advancements in mechanical structure and actuation functions. Especially for multi-fingered hands, how to stably and reliably acquire high-precision, high-bandwidth force, position, or other sensory information remains a technical challenge that has not yet been fully resolved. Tactile sensing, as a core technology for realizing intelligent perception and human-robot interaction in robots, directly determines the accuracy, safety, and adaptability of robot operations.

[0003] The sensors configured in dexterous hands can be divided into two main categories: internal sensors and external sensors. Internal sensors include torque, position, velocity, and acceleration sensors, primarily used to sense the robot's own state and achieve motion control and dynamic balance. External sensors are used to sense the external environment, such as contact forces, object position, shape, texture, and temperature. However, there is a general problem of insufficient sensor configuration at the external sensing level, mainly reflected in the following aspects:

[0004] 1. Limitations of a single mode: Existing solutions are mostly limited to a single sensing mode, which can only provide tactile pressure perception and cannot meet the multi-dimensional task requirements such as contact detection, slip perception, and object attribute (hardness, texture, thermal conductivity) recognition.

[0005] 2. Poor scalability: Sensor arrays are mostly fixed configurations, usually designed for specific dexterous hands or their specific positions. They are difficult to adapt to dexterous hands and mechanical grippers with different configurations or degrees of freedom, resulting in high system reconfiguration costs and difficulty in meeting different needs.

[0006] 3. Data Transmission Bottleneck: As a dot matrix sensor, tactile sensor arrays typically use scanning for sampling, involving complex spatiotemporal protocol issues. The sampling rate is limited by various factors such as row and column scanning frequency, analog-to-digital conversion frequency, and communication protocol. Currently, the sampling rate of piezoresistive sensing systems on the market is generally below 50Hz, which is insufficient to support high-frequency controllers, becoming a key bottleneck restricting system performance.

[0007] 4. High integration complexity: Existing external sensor solutions are independent of the robot's dexterous hand, requiring external signal conditioning circuits, independent data acquisition equipment, and complex wiring schemes. Within the limited space of the dexterous hand, the number of wires is severely limited; in addition, the curved contact structure of the dexterous hand has irregular geometry, making it difficult to directly mount sensors on the surface of the robot's fingers.

[0008] 5. Insufficient cost-effectiveness: High-performance commercial tactile sensors are expensive, while low-cost commercial sensors often suffer from performance defects such as single-mode operation, poor scalability, and low frame rate, and do not have the potential to be applied to target scenarios through simple secondary development. Summary of the Invention

[0009] The purpose of this invention is to solve the technical problem of how to construct a robot tactile sensing system that can simultaneously achieve high-bandwidth data acquisition, multimodal information fusion, high system scalability, and easy integration, and proposes a modular multimodal tactile sensing method for robots and its application.

[0010] To achieve the above objectives, the present invention adopts the following technical solution: A modular multimodal tactile sensing method for robots includes the following steps: detecting external physical quantities through multiple distributed sensing units, each sensing unit including a tactile array sensor for detecting contact pressure, a thermocouple for detecting temperature, and a flexible tension sensor for detecting tensile tension; conditioning the electrical signals output by the corresponding sensing units through multiple multimodal signal conditioning modules to convert them into analog voltage signals; each multimodal signal conditioning module has an upper-level input interface and a lower-level output interface, connected via a bus-type cascaded configuration; and providing power and data to all cascaded multimodal signal conditioning modules through a high-speed data processing and communication module. The system scans control signals and simultaneously acquires the analog voltage signals output by all multi-modal signal conditioning modules, converting them into digital signals. The high-speed data processing and communication module encapsulates the digital signals into data packets according to a predefined communication protocol and transmits them to the host system. The synchronous acquisition step employs an interrupt-driven mechanism: a periodic conversion start signal simultaneously triggers multiple analog-to-digital converter chips to begin conversion; when conversion is complete, a conversion status signal triggers an interrupt service routine, in which each analog-to-digital converter chip is time-divisionally selected to read the converted data and store it in a buffer; in the main program loop, data is read from the buffer and encapsulated.

[0011] This invention also provides a modular multimodal tactile sensing system for robots, comprising: sensing units distributed on the surface of an object to be measured, the sensing units including a tactile array sensor for detecting pressure, a thermocouple for detecting temperature, and a flexible tension sensor for detecting tensile tension; at least one multimodal signal conditioning module electrically connected to the sensing units, the multimodal signal conditioning module including a signal conditioning unit for receiving and conditioning the electrical signals output by the tactile array sensor, thermocouple, and flexible tension sensor; the multimodal signal conditioning module further including a bus interface unit for receiving power lines, ground lines, and control signals from external sources, and for transmitting the power lines and ground lines... The system includes a control signal output module and a high-speed data processing and communication module electrically connected to the multimodal signal conditioning module. The high-speed data processing and communication module includes a main control unit and at least one analog-to-digital conversion unit. The main control unit generates the control signal and sends it to the multimodal signal conditioning module through the bus interface unit. The analog-to-digital conversion unit synchronously acquires multiple analog signals conditioned by the signal conditioning unit and converts them into digital signals, as well as reads and processes the digital signals. The main control unit coordinates the generation of the control signal, the start of the conversion of the analog-to-digital conversion unit, and the reading of the digital signals through interrupt-driven timing control logic to achieve synchronous high-speed acquisition of data from the entire system.

[0012] This invention also provides a multimodal signal conditioning module, comprising: a signal conditioning unit for conditioning and converting the electrical signal output from an access sensing unit into an analog voltage signal output, wherein the sensing unit includes a tactile array sensor, a thermocouple, and a flexible tension sensor; an upper-level input interface for connecting power lines and control signal lines from a data processing module or a previous-level multimodal signal conditioning module; a lower-level output interface for leading the power lines and control signal lines to a next-level multimodal signal conditioning module; and an analog voltage output interface for outputting the analog voltage signal to the data processing module; wherein the upper-level input interface and the lower-level output interface adopt the same bus protocol, enabling multiple multimodal signal conditioning modules to be cascaded via a bus through the upper-level input interface and the lower-level output interface.

[0013] This invention also provides a high-speed data processing and communication module, comprising: a microcontroller; multiple analog-to-digital converter (ADC) chips, wherein the analog input channels of each ADC chip are respectively connected to the analog voltage output interface of the multimodal signal conditioning module described above to receive analog voltage signals; all ADC chips share the data bus and control signal lines of the microcontroller except for the chip select signal; each ADC chip is respectively connected to an independent chip select pin of the microcontroller; wherein the microcontroller is configured to: generate periodic conversion start signals and simultaneously output them to all ADC chips to trigger synchronous conversion; in response to a conversion status signal issued by any ADC chip after conversion is completed, enter an interrupt service routine; in the interrupt service routine, each ADC chip is selected in a time-division multiplexing manner by controlling the chip select signal, and the conversion data of the selected chip is read through the data bus and stored in a buffer; in the main program loop, data is read from the buffer and uploaded through the communication interface.

[0014] The present invention also provides a robotic dexterous hand, including a palm and multiple fingers, and further including multiple multimodal signal conditioning modules as described above and a high-speed data processing and communication module as described above; the multiple multimodal signal conditioning modules are connected in a bus-cascade manner and distributed and integrated on the surface of the palm and fingers; the tactile array sensor and thermocouple connected to the multimodal signal conditioning module are disposed on the pressure surface of the palm and fingers, and the flexible tension sensor is connected across the joint of the finger.

[0015] The present invention also provides a robot, including the robotic dexterous hand described above.

[0016] The beneficial effects of this invention compared to the prior art include: By adopting a modular, bus-cascaded architecture and combining it with an interrupt-driven synchronous acquisition mechanism, this invention effectively integrates multimodal sensing units, distributed signal conditioning, and centralized high-speed data processing, thereby collaboratively achieving high-bandwidth data acquisition and transmission for the robot's tactile perception system under limited resources. Specifically, the multimodal signal conditioning module achieves plug-and-play and flexible expansion of sensing units through a unified interface, solving the problems of poor system scalability and complex integration; the interrupt-driven mechanism and multi-analog-to-digital converter chip multiplexing strategy efficiently utilize microcontroller resources, significantly improving the synchronous acquisition efficiency of multi-channel data and the system sampling rate, breaking through the data transmission bottleneck; and the integration of multimodal sensing units fundamentally expands the dimensions of the robot's perception of the physical environment. Through the synergistic effect of these technical features, this invention provides a high-performance, highly scalable, low-integration-complexity, and cost-effective modular multimodal tactile sensing overall solution.

[0017] Other beneficial effects of the embodiments of the present invention will be further described below. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the overall architecture of the modular multimodal tactile sensing system according to an embodiment of the present invention.

[0019] Figure 2 This is a schematic diagram of the arrangement of a distributed sensor array on a robot's dexterous hand according to an embodiment of the present invention.

[0020] Figure 3a This is a schematic diagram of the tactile array detection circuit according to an embodiment of the present invention.

[0021] Figure 3b This is a schematic diagram of the temperature sensor detection circuit according to an embodiment of the present invention.

[0022] Figure 3c This is a schematic diagram of the flexible tension sensor detection circuit according to an embodiment of the present invention.

[0023] Figure 4 This is a schematic diagram of the structure and interface of the multimodal signal conditioning module in an embodiment of the present invention.

[0024] Figure 5 This is a schematic diagram of a multi-AD7606 extended parallel port communication connection circuit according to an embodiment of the present invention.

[0025] Figure 6 This is a flowchart illustrating the collaborative operation of interrupt sampling and main function processing in an embodiment of the present invention.

[0026] Figure 7 This is a schematic diagram of the host computer visualization software interface according to an embodiment of the present invention. Detailed Implementation

[0027] The present invention will be further described below with reference to the accompanying drawings and preferred embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.

[0028] This invention proposes a modular multimodal tactile sensing method, system, and application for humanoid robot hands, aiming to address the shortage of cost-effective external sensors in current dexterous hand products and open-source projects. This method and system enable the robot hand to identify object attributes by grasping and manipulating objects, and to independently detect various physical quantities such as contact pressure, local ambient temperature, and thermal conductivity. This invention does not focus on a single sensor or circuit module; its core innovation lies in constructing a complete, scalable, and cost-effective system architecture. Based on this architecture, software and hardware components work collaboratively to achieve synchronous 200Hz multimodal information acquisition and to control the overall deployment cost of a single-handed robot to approximately 4000 yuan.

[0029] This innovation is reflected in a systematic and collaborative design across multiple levels: In the underlying sensing unit, this embodiment of the invention designs a multimodal sensor fusion and spatial stacking scheme, integrating tactile, temperature and tension sensors in the same physical area, realizing the synchronous acquisition of multiple physical signals in a limited space.

[0030] At the level of multimodal signal conditioning modules, this embodiment of the invention assigns each multimodal signal conditioning module to a knuckle space, continuing the spatial multiplexing concept of the sensing layer. Combined with modular structure and bus design, it completes the knuckle-level encapsulation and conditioning from physical signals to measurable electrical signals.

[0031] In the high-speed data processing and acquisition module, which is the core component determining system performance, this module adopts parallel interface multiplexing technology and a double-buffered transceiver mechanism, which significantly improves data throughput efficiency and channel expansion capability. This enables a single processing unit to uniformly schedule, process in real time, and reliably feed back to the back-end system from multiple front-end signal acquisition modules.

[0032] Finally, the accompanying host computer software is not only used to verify the overall system performance, but also has the function of real-time visualization monitoring of multi-channel data, forming a complete closed loop from signal perception to application verification.

[0033] In summary, this invention, starting from system architecture, constructs a high-performance, easily expandable, and low-cost multimodal tactile sensing system-level solution through the deep integration of "design concepts" and "engineering methods".

[0034] This invention provides a modular multimodal tactile sensing method for robots, comprising the following steps: detecting external physical quantities through multiple distributed sensing units, each sensing unit including a tactile array sensor for detecting contact pressure, a thermocouple for detecting temperature, and a flexible tension sensor for detecting tensile tension; conditioning the electrical signals output by the corresponding sensing units through multiple multimodal signal conditioning modules to convert them into analog voltage signals; each multimodal signal conditioning module has an upper-level input interface and a lower-level output interface, connected via a bus-type cascaded configuration; and transmitting data to all cascaded multimodal signal conditioning modules via a high-speed data processing and communication module. The block provides power and row scanning control signals, and synchronously acquires the analog voltage signals output by all multi-mode signal conditioning modules, converting them into digital signals. The high-speed data processing and communication module encapsulates the digital signals into data packets according to a predefined communication protocol and transmits them to the host system. The synchronous acquisition step employs an interrupt-driven mechanism: a periodic conversion start signal simultaneously triggers multiple analog-to-digital converter chips to begin conversion; when conversion is complete, a conversion status signal triggers an interrupt service routine, in which each analog-to-digital converter chip is time-divisionally selected to read the converted data and store it in a buffer; in the main program loop, data is read from the buffer and encapsulated.

[0035] The electrical signal output by the tactile array sensor is conditioned using an inverting amplifier circuit; the electrical signal output by the thermocouple is conditioned using a non-inverting amplifier circuit; and the electrical signal output by the flexible tension sensor is conditioned using a voltage divider circuit.

[0036] The interrupt service routine performs the following operations: selects an analog-to-digital converter chip based on the current index; reads the conversion data of multiple channels in the current row of the chip; stores the data in a first-in-first-out (FIFO) buffer; and updates the chip and row indexes.

[0037] In the main program loop, after reading data from the buffer, the data is reassembled according to the analog-to-digital converter chip and row address information of the data source. After each complete sensor unit's data is collected, it is encapsulated into a data packet. The data packet contains a frame header, module identifier, tactile data, temperature data, tension data, and checksum.

[0038] This invention also provides a modular multimodal tactile sensing system for robots, comprising: sensing units distributed on the surface of an object to be measured, the sensing units including a tactile array sensor for detecting pressure, a thermocouple for detecting temperature, and a flexible tension sensor for detecting tensile tension; at least one multimodal signal conditioning module electrically connected to the sensing units, the multimodal signal conditioning module including a signal conditioning unit for receiving and conditioning the electrical signals output by the tactile array sensor, thermocouple, and flexible tension sensor; the multimodal signal conditioning module further includes a bus interface unit for receiving power lines, ground lines, and control signals from external sources. It can output power lines, ground lines, and control signals; a high-speed data processing and communication module, which is electrically connected to a multi-mode signal conditioning module, includes a main control unit and at least one analog-to-digital converter (ADC). The main control unit generates control signals and sends them to the multi-mode signal conditioning module through a bus interface unit. The ADC is used to synchronously acquire multiple analog signals conditioned by the signal conditioning unit and convert them into digital signals, as well as read and process the digital signals. The main control unit coordinates the generation of control signals, the start of the ADC conversion, and the reading of digital signals through interrupt-driven timing control logic, thereby achieving synchronous high-speed acquisition of data from the entire system.

[0039] This invention also provides a multimodal signal conditioning module, comprising: a signal conditioning unit for conditioning and converting the electrical signal output by the connected sensing unit into an analog voltage signal output; the sensing unit including a tactile array sensor, a thermocouple, and a flexible tension sensor; an upper-level input interface for connecting power lines and control signal lines from a data processing module or a previous-level multimodal signal conditioning module; a lower-level output interface for leading the power lines and control signal lines to a next-level multimodal signal conditioning module; and an analog voltage output interface for outputting an analog voltage signal to the data processing module; wherein the upper-level input interface and the lower-level output interface adopt the same bus protocol, enabling multiple multimodal signal conditioning modules to be cascaded via a bus through the upper-level input interface and the lower-level output interface.

[0040] The signal conditioning unit includes: an analog switch for gating the signals of each row of the tactile array sensor; an inverting amplifier circuit for conditioning the signals of each column of the tactile array sensor; a two-stage non-inverting amplifier circuit for conditioning the output signal of the thermocouple; and a voltage divider circuit for conditioning the output signal of the flexible tension sensor.

[0041] This invention also provides a high-speed data processing and communication module, comprising: a microcontroller; multiple analog-to-digital converter (ADC) chips, wherein the analog input channels of each ADC chip are respectively connected to the analog voltage output interface of the multi-mode signal conditioning module described above to receive analog voltage signals; all ADC chips share the microcontroller's data bus and control signal lines except for the chip select signal; each ADC chip is connected to an independent chip select pin of the microcontroller; wherein the microcontroller is configured to: generate periodic conversion start signals and simultaneously output them to all ADC chips to trigger synchronous conversion; in response to a conversion status signal issued by any ADC chip after conversion is completed, enter an interrupt service routine; in the interrupt service routine, each ADC chip is selected in a time-division manner by controlling the chip select signal, and the conversion data of the selected chip is read through the data bus and stored in a buffer; in the main program loop, data is read from the buffer and uploaded through the communication interface.

[0042] This invention also provides a robotic dexterous hand, including a palm and multiple fingers, and further including multiple multimodal signal conditioning modules as described above and a high-speed data processing and communication module as described above; the multiple multimodal signal conditioning modules are connected in a bus-cascade manner and distributed and integrated on the surface of the palm and fingers; the tactile array sensor and thermocouple connected to the multimodal signal conditioning module are disposed on the pressure surface of the palm and fingers, and the flexible tension sensor is connected across the joint of the fingers.

[0043] This invention also provides a robot, including the robotic dexterous hand described above.

[0044] The following describes specific embodiments of the present invention.

[0045] This embodiment provides a modular multimodal tactile sensing system for robots, the overall architecture of which is as follows: Figure 1 As shown, it specifically includes the following four components: Sensing unit: The sensing unit is used to detect external multimodal physical quantities and convert them into electrical signals. This includes a tactile array sensor for pressure measurement via piezoresistive conversion, a thermocouple for temperature measurement using the Seebeck effect, and a flexible tension sensor where elongation is correlated with resistance. Their arrangement and integration are detailed below:

[0046] To achieve multimodal signal sensing of the hand, this embodiment proposes a sensing unit architecture consisting of a tactile array sensor, an adhesive thermocouple, and a flexible tension sensor.

[0047] The tactile array sensor uses high-performance PVDF (polyvinylidene fluoride) as the core material for tactile sensing. Based on this material, the piezoresistive sensor array, with its array characteristics, can measure the stress distribution at various points in a plane.

[0048] The adhesive-mount thermocouple uses a T-type thermocouple (copper-nickel) with a temperature range of -270℃ to 400℃, meeting the needs of everyday environments. Encapsulating the thermocouple and using double-sided adhesive for mounting facilitates spatial integration with piezoresistive sensors.

[0049] The flexible tension sensor is made of carbon fiber and has a stretching and measuring range of 5cm to 15cm, which matches the range of hand movements. Its flexible and stretchable packaging allows for quick integration by simply gluing both ends together.

[0050] All of the aforementioned sensing units employ flexible circuit design. Their flexibility allows them to adapt to the complex curved surface encapsulation requirements of dexterous hands and mechanical grippers, and they can be customized to fit mechanical grippers and dexterous hands with different geometric shapes. The integration of temperature sensors and flexible tension sensors provides richer multimodal tactile information on top of tactile feedback.

[0051] Based on the three types of flexible sensors described above that measure different physical quantities, this embodiment proposes a distributed sensor array structure to more comprehensively measure the sensing information of the entire hand, such as... Figure 2 As shown in the diagram. The specific distribution of the PVDF piezoresistive sensor, adhesive thermocouple, and flexible tension sensor in this structure is as follows:

[0052] Palm area: 36 piezoresistive sensor detection points are distributed in a 6×6 array within a 9cm×9cm space, covering the main contact area of ​​the palm, and are used to measure palm tactile information; Fingertip area: 36 piezoresistive sensor detection points are distributed in a 6×6 array within a 2cm×2cm space, covering the main area of ​​the foremost knuckle (fingertip); Knuckle area: 32 piezoresistive sensor detection points are distributed in an 8×4 array within a 2cm×4cm space, covering the fingertip area.

[0053] The area of ​​the adhesive thermocouple is the same as that of the piezoresistive sensor above it. A thermocouple is placed under each piezoresistive sensor to achieve coverage of both temperature and tactile modal information in the area.

[0054] The flexible tension sensor is attached to the joint and fixed at both ends of the joint, and it extends and retracts as the joint moves.

[0055] 2. Modular measurement circuit (multi-modal signal conditioning module): The multimodal signal conditioning module is used to condition the weak electrical signals output by the sensing unit and convert them into analog signals of an appropriate range. The resistance of each point of the tactile array sensor is processed by an inverting amplifier, the weak electrical signals output by the thermocouple are processed by a non-inverting amplifier, and the flexible tension sensor is directly connected in series with a resistor to output a voltage divider.

[0056] Specifically, the modular measurement circuit (multimodal signal conditioning module) is used to convert the changes in the various sensors in the sensing unit, namely the resistance changes of the tactile array, the weak electromotive force of the thermocouple, and the resistance changes of the flexible tension sensor, into electrical signals that can be detected by the analog-to-digital converter.

[0057] The multimodal signal conditioning module acts as a bridge connecting the multimodal sensors and the data processing system, undertaking the function of electrical signal conversion. A single multimodal signal conditioning module corresponds to a set of sensors (including a tactile array, a thermocouple, and a flexible tension sensor). Corresponding detection circuits need to be designed for each of these sensors, as shown in the diagram. Figures 3a to 3c As shown.

[0058] 2.1 Tactile Array Detection Circuit: The tactile array uses a row scanning method for sampling, which is controlled by an analog switch RS2251. RS2251 is a CMOS analog integrated circuit configured as an 8-channel multiplexer, which selects any one of the eight signals via row scanning control signals CH_A, CH_B, and CH_C. Its common terminal is connected to 3.3V. When CH_A, CH_B, or CH_C selects a row, the corresponding row receives a high level; unselected rows are grounded through pull-down resistors, maintaining 0V.

[0059] Based on this mechanism, the high-speed data processing and communication module selects rows 1 through 6 of the tactile array sensor sequentially by controlling the levels of the CH_A, CH_B, and CH_C signals. Rows 1 through 6 of the tactile array sensor are connected to channels 1 through 6 of the RS2251, respectively. Each dot of the tactile array acts as a resistor; when a row is selected, the corresponding channel of the analog switch outputs a high level, and all resistors in that row receive current.

[0060] like Figure 3a As shown, each column (i.e., the other end of the row and column resistors) is processed by an external inverting amplifier circuit. Columns 1 to 6 of the tactile array are connected to the inverting input of the operational amplifier, the non-inverting input is grounded, and a 5kΩ feedback resistor is connected between the inverting input and the output. The circuit will then output a voltage related to the resistance value of the array points. The output voltage formula is:

[0061]

[0062] Output voltage With the resistance value of the lattice point The resistance and pressure characteristics of the piezoresistive material are negatively correlated. After analysis using this curve, the contact force at that point can be obtained. 5V is the power supply voltage (input voltage) of the inverting amplifier.

[0063] The inverting amplifier circuit for the tactile sensor: The sensor's resistance is negatively correlated with pressure, and its no-load resistance is extremely high (megaohm level). To ensure that the output voltage is positively correlated with pressure and to suppress no-load output, it is placed in the feedback network of the inverting amplifier.

[0064] 2.2 Temperature sensor detection circuit: The temperature sensor uses a thermocouple. Due to the thermoelectric effect (Seebeck effect), an electromotive force is generated when there is a temperature difference between the hot and cold junctions. However, this electromotive force is extremely weak (on the order of 0.01mV), so it needs to be amplified by subsequent stages.

[0065] In circuit design (such as) Figure 3b (As shown) To avoid stability issues caused by excessively high single-stage amplification, a two-stage in-phase amplifier is used: the first stage amplifies by 11 times, the second stage amplifies by 25 times, and the total amplification of the two stages is 275 times. At the same time, to address the high-frequency noise present in the thermocouple, a low-pass filter is added before the op-amp stage, with a resistor of 100Ω and a capacitor of 100nF, and a cutoff frequency of approximately 15.9kHz, which can effectively filter out kHz-level noise.

[0066] Non-inverting operational amplifier circuit: Input signal U in It is directly connected to the non-inverting input of the operational amplifier. Output voltage. Through a resistor R f and R g The feedback network, composed of resistors R, is connected to the inverting input, thus forming negative feedback. f Connect the output terminal to the inverting input terminal; resistor R g Connect the inverting input terminal to ground; the amplifier's output voltage formula is:

[0067]

[0068] Output voltage proportional to the electromotive force of a thermocouple The real-time touch temperature can be obtained by consulting the T-type thermocouple calibration table.

[0069] Two-stage in-phase amplification of thermocouple signals: The thermocouple signal needs to be amplified by a factor of 100. If a single-stage operational amplifier is used, it will force the amplifier to operate in its high-gain range. This will not only significantly introduce nonlinear errors and offset voltages, but also severely restrict its gain-bandwidth product, leading to stability risks. Therefore, using two-stage amplification is a necessary compromise to achieve high gain, high stability, and excellent performance.

[0070] 2.3 Flexible tension sensor detection circuit: A flexible tension sensor is essentially a variable resistor; therefore, by connecting it in series with a resistor of a suitable value, the resistance value can be converted into a voltage value (e.g., ...). Figure 3c As shown), the output voltage formula is:

[0071] Output voltage With resistance of tension sensor The (fts (Flexible Tension Sensor)) has a mapping relationship; by substituting it into the tensile ratio-resistance relationship curve, the tensile length can be obtained.

[0072] Series voltage divider circuit for flexible tension sensor: The sensor's resistance remains in the kiloohm range throughout its entire range, and changes in resistance under no-load and load conditions do not pose a challenge to the circuit's input impedance. Therefore, a simple series voltage divider circuit can achieve reliable voltage conversion, and the design is more convenient.

[0073] 2.4 Multimodal signal conditioning module interface design: The multimodal signal conditioning module designed based on the above circuit has the following characteristics: Figure 4 The interface shown: Tactile sensor array interface: FPC (flexible printed circuit) 12Pin, connecting to 6 rows and 6 columns of tactile sensors; Temperature sensor input interface: SH (a connector standard manufactured by JST Corporation of Japan) 2Pin, connecting the two poles of the thermocouple cold junction (TC+, TC-). Flexible tension sensor input interface: SH 2Pin, connects to the two poles of the tension sensor (ZL+, ZL-); Upper-level input interface: SH 6Pin, connecting to the high-speed data processing and communication module or the previous level multimodal signal conditioning module, including three power lines (+5V, -5V, GND) and line scan control signals (CH_A, CH_B, CH_C). All multimodal signal conditioning modules share the same set of line scan control signals, which are uniformly driven by the microcontroller GPIO to realize synchronous line scanning of the tactile array of each module; the "-5V" power supply is to power the negative power rail of the operational amplifier to ensure that it can output a voltage close to 0V.

[0074] Lower-level output interface: SH 6Pin, which leads the three power lines and line scan control signal from the high-speed data processing and communication module or the previous stage to the next-level multi-mode signal conditioning module; Analog voltage output interface: SH 8Pin, connects to the high-speed data processing and communication module, outputs the converted electrical signal, including 6 signals (AD1 to AD6) of the tactile array sensor row, tension sensor signal (AD7), and temperature sensor signal (AD8).

[0075] The aforementioned upper-lower bus connection mechanism forms the basis of system scalability. On the same bus (i.e., three power lines and three line scan control signals), the number of multimodal signal conditioning modules can be flexibly expanded through upper-lower interfaces. For example, for a two-knuckle dexterous hand, only two multimodal signal conditioning modules are needed for a single finger; when expanding to three knuckles, an additional multimodal signal conditioning module can be directly connected to the lower-level output interface to achieve expansion, without the need to redesign the system architecture.

[0076] 2.5 Compact Design: To facilitate integration, in addition to the bus interface mechanism, this embodiment designs the modular measurement circuit to a knuckle-level size, using a four-layer PCB and a top-bottom double-sided layout to compress space. The final module dimensions are: length 25mm (adapting to the length of the proximal phalanx), width 15mm, and thickness 8mm. The size of a single measurement circuit is close to that of a human knuckle, and it can be integrated into a dexterous hand using 3D-printed shells or nuts and other structural components.

[0077] 3. High-speed data processing and communication module: The high-speed data processing and communication module is used to perform analog-to-digital conversion, encapsulation, and storage of analog signals, and upload them to the host computer through a high-speed communication interface.

[0078] Specifically, after processing by the detection circuit, the sensor converts changes in external physical quantities into analog electrical signals. These signals require analog-to-digital conversion to become meaningful digital quantities. The high-speed data processing and communication module, as the core processing unit of this system, is responsible for providing power and row scanning control signals to the sampling circuit, performing analog-to-digital conversion on the analog electrical signals output by the sampling circuit, buffering and encapsulating the data, and uploading it to the host computer system via a high-speed USB serial port.

[0079] 3.1 Hardware basics for multi-AD7606 extended parallel port communication: The high-speed data processing and communication module uses the high-performance STM32F407VET6 microcontroller based on the ARM Cortex-M4 architecture as its main control core, and connects to six 8-channel AD7606 analog-to-digital converter chips to meet the high parallel acquisition requirements of all sensors (channels 1 to 6 for the tactile array, channel 7 for the flexible tension sensor, and channel 8 for the thermocouple). The microcontroller and each AD7606 chip communicate at high speed via a parallel bus, following the FSMC (Flexible Static Memory Controller) parallel protocol (see [link to connection circuit diagram] for details). Figure 5The master controller sends bus control signals to the slave devices. After receiving these signals, the slave devices upload the converted digital voltage values ​​to the master controller. After completing a full read, the master controller uploads all data to the host computer system via USB in data packet format. Because the number of slave devices connected via the FSMC parallel port protocol is limited, this embodiment specifies 6 FSMC slave devices (the number can be 1-8). Each FSMC slave device, i.e., one ADC (analog-to-digital converter) of the data processing module (high-speed data processing and communication module), can process the signals of one multimodal signal conditioning module (including a tactile PVDF sensor array, a thermocouple, and one or more flexible tension sensors). A total of 15 multimodal signal conditioning modules are needed for the entire hand deployment, requiring 15 ADCs for processing, and at least 2 data processing modules are required. Since one data processing module corresponds to 8 ADCs, and the entire hand requires 15 ADCs, at least two data processing modules are needed.

[0080] To address this issue, multiple data processing modules can be used in parallel to expand the system.

[0081] Synchronous acquisition of multiple data processing modules is achieved through an external synchronization signal: an external synchronization clock circuit (such as the signal generation module of the peripheral MCU timer or the output of the host computer's GPIO pin) generates a 200Hz TTL level square wave signal, which is simultaneously input to each data processing module. Different data processing modules are configured to trigger sampling and data transmission at different edges or levels of this signal. For example, the first data processing module is set to be triggered by the rising edge, and the second data processing module is set to be triggered by the falling edge, thereby achieving synchronous sampling and staggered data transmission timing.

[0082] Each data processing module independently uploads data to the host computer through its own communication interface and configures its own AD ID independently, ensuring that they belong to different ranges and do not overlap. For example, slave 1 is 0-5 and slave 2 is 6-11; or slave 1 is 0-7 and slave 2 is 8-15. The host computer aggregates and integrates the data from different modules based on the AD ID contained in the data packet.

[0083] In terms of hardware connectivity, the 16-bit FSMC data bus DB0~DB15 is used for parallel reading of the AD7606's conversion results. Key control signals include:

[0084] Conversion Start Signal (CONVST): This signal is generated by the PWM output channel of the microcontroller's Timer TIM1 and is a periodic square wave signal. This signal is simultaneously connected to the CONVST pin of all six AD7606 chips. Each falling edge triggers all AD7606 chips to synchronously start an 8-channel analog-to-digital conversion, thereby ensuring that all chips sample in strict synchronization.

[0085] Conversion Status Signal (BUSY): This signal is output by the AD7606 chip and connected to the microcontroller's external interrupt pin PC10. When the AD7606 receives the conversion start signal, the BUSY pin is set to a high level, indicating that conversion is in progress; when all 8 channels of analog-to-digital conversion are completed, the BUSY pin generates a falling edge transition. The microcontroller triggers an external interrupt by detecting this falling edge, which serves as the timing reference for data read.

[0086] Chip select signals (CS1~CS6): Each AD7606 chip has an independent chip select pin, which is independently controlled by the microcontroller's six GPIO ports. The chip select signal is active low. When a chip select signal is pulled low, the corresponding AD7606 chip is selected, its data output buffer is enabled, and the data bus can access the chip's conversion result.

[0087] Read Enable Signal (RD): This signal is generated by the microcontroller and provides read timing during the chip select active period. The read enable signal is a low-level active pulse signal. During each low-level period, the selected AD7606 chip drives the conversion result of the current channel to the data bus DB0~DB15. Since the AD7606 uses an 8-channel sequential output mechanism internally, the main microcontroller needs to generate 8 consecutive read enable pulses during the chip select active period to sequentially read the conversion data of all 8 channels of the chip.

[0088] The aforementioned signals work in strict accordance with the timing sequence, ensuring high-speed and orderly data acquisition.

[0089] To enable the coordinated operation of six AD7606 chips, the system employs a chip select signal multiplexing mechanism. Specifically, all AD7606 chips share the data bus, address bus, and control signals (CONVST, RD, etc.) except for their respective chip select pins. During data acquisition, the microcontroller cyclically pulls down the corresponding chip select level for each chip, thereby selecting different AD7606 chips and enabling the reading of data from all chips using the same parallel bus. This hardware architecture lays the physical foundation for the system to achieve high-density, high-frequency data acquisition.

[0090] 3.2 Interrupt-driven sampling mechanism: To achieve an effective sampling rate of 200Hz for the entire system, it is necessary to understand the two-layer multiplexing structure of the system's data acquisition: First layer: Row scan multiplexing of the tactile array Each multimodal signal conditioning module's tactile array sensor employs a 6x6 matrix structure, with row scanning sampling achieved via analog switches. Row selection signals CH_A, CH_B, and CH_C are output from the microcontroller's GPIO pins, and all multimodal signal conditioning modules share the same set of row selection signals. The microcontroller selects rows 1 through 6 of the tactile array sequentially by controlling the level combinations of the row selection signals CH_A, CH_B, and CH_C (from 000 to 101, corresponding to decimal 0 to 5). Therefore, completing the acquisition of a single tactile array requires 6 row scans.

[0091] Second layer: Polling read multiplexing of multiple AD7606s For six AD7606 chips sharing the same 16-bit FSMC data bus, the system employs a polling read strategy: after each CONVST signal trigger, the interrupt service routine selects and reads only the 8 channels of data from one AD7606 chip; after the next CONVST trigger, it reads the data from the next chip, and so on. Therefore, completing one round of data acquisition for all six chips requires six consecutive CONVST triggers.

[0092] Combining the two-layer multiplexing structure described above, completing a full system data acquisition requires sequentially scanning six rows of six AD chips and the sampling circuits connected to each AD chip, which necessitates 36 CONVST triggers. Therefore, to achieve an effective system sampling rate of 200Hz, the actual trigger frequency of the CONVST signal is... 7200Hz is required:

[0093]

[0094] The system employs an interrupt-driven precision sampling control strategy. First, the microcontroller generates a 7.2kHz conversion start signal to trigger the AD7606 to perform continuous analog-to-digital conversion. After each conversion cycle, the AD7606's BUSY pin generates a falling edge. This falling edge signal is connected to the microcontroller's external interrupt pin, triggering an external interrupt service routine. The interrupt service routine further calls the AD7606_ISR() function, which performs the following operations:

[0095] Select the AD chip and row address to be read; Read the AD conversion results of the current row of 8 channels of the current AD chip; Store the data in a FIFO (First In First Out) buffer; Update the chip and row indexes to prepare for the next sampling.

[0096] This interrupt-driven approach ensures timely reading of conversion data from 6 AD chips (6 rows per chip, 8 channels per row) at a system effective sampling rate of 200Hz, totaling 288 channels. Continuous data acquisition and storage are achieved through a FIFO buffer mechanism.

[0097] 3.3 Main Function Data Filling and Sending Process The microprocessor repeatedly calls the `ProcessFifoMode()` function in its main function to read the raw data stored by the interrupt service routine from the FIFO circular buffer in real time, and then performs subsequent processing, encapsulation, and transmission operations. Within this function, the system executes the following key steps sequentially:

[0098] Data reading and reassembly: Circularly read all available data in the FIFO buffer, and based on the AD chip and row address information when the data was stored, organize the streaming data that was originally stored in time order into data blocks indexed by chip and row.

[0099] Data Packaging and Protocol Encapsulation: After processing all 6 lines of data (48 channel values ​​in total) from each AD chip, the system fills these data into the transmission buffer according to the dedicated communication protocol format defined in this embodiment. The data is stored in little-endian mode in the buffer. The specific format of this dedicated data packet is shown in Table 1.

[0100] Table 1 Data Packet Format

[0101] Data transmission: After the data packet is encapsulated, the checksum is calculated, and the complete data packet is sent to the host computer in batches through the USB CDC (USB communication device class) virtual serial port interface.

[0102] The coordination mechanism between interrupt sampling and main function processing is as follows: Figure 6 As shown, both share a circular buffer. The interrupt service routine acts as the producer, acquiring AD7606 data at high speed via the FSMC protocol and filling it into the circular buffer. The data processing flow in the main function acts as the consumer, polling the buffer and reading data to complete reassembly, packaging, and uploading. The two are decoupled through the shared circular buffer, collaboratively achieving high-throughput, low-latency continuous data acquisition and transmission.

[0103] 4. Host computer visualization and monitoring software: The host computer visualization and monitoring software is used to receive, parse, and display sensor data, as well as verify the system's acquisition effect. It includes serial port management functions and displays the data in real time in the grid control after parsing the data.

[0104] Based on the hardware architecture of the high-speed data processing and communication system, this embodiment includes a set of Windows Forms applications. As a crucial component of the system, this software not only undertakes the tasks of real-time reception, parsing, and display of sensor data, but more importantly, it provides an intuitive window for performance verification and implementation effect demonstration.

[0105] The host computer program establishes a connection with the high-speed data processing module of the lower-level machine through a virtual serial port USB CDC. It receives data packets from the lower-level machine in real time and performs frame header verification, data parsing, checksum verification, and unpacking operations according to the communication protocol defined by the high-speed data processing and communication module.

[0106] To intuitively demonstrate the system's high-density, high-bandwidth data acquisition capabilities, the core of the software interface adopts a matrix-based data visualization design (see the host computer software interface). Figure 7 The system uses six independent controls to correspond to and display the outputs of six data acquisition modules in real time. Each control maps the channel structure of a single module in a 6-row × 8-channel matrix layout. In the matrix, the first six columns dynamically display the real-time voltage of each dot matrix of the tactile array sensor, reflecting the contact pressure distribution; the seventh column displays the voltage signal of the flexible tension sensor, characterizing the joint stretching state; and the eighth column displays the voltage signal of the temperature sensor, reflecting the temperature information of the contact area.

[0107] To further enhance data recognition, the software dynamically renders the color of each data unit based on the real-time voltage value, using a hierarchical color coding scheme to map different voltage ranges to blue, light green, yellow, orange, and red, thereby enabling users to quickly perceive signal strength and changing trends.

[0108] This host computer software provides real-time, raw, and high-bandwidth visualization of the overall data acquisition performance of the hardware system. Users can directly verify the system's sampling rate, channel stability, and data throughput through this interface, and it offers intuitive support for system debugging and performance verification.

[0109] Through the coordinated work of its various modules, from the sensing unit to the measurement circuit and then to the data processing module, the system achieves high-bandwidth acquisition and processing of multimodal tactile information from the bottom up.

[0110] The innovative aspects of this embodiment are as follows: 1. Overall Architecture of Multimodal Tactile Sensing System The architecture, from bottom to top, consists of: a distributed multimodal sensor array, modular sampling circuitry, multiple AD7606 analog signal sampling chips, the FSMC multiplexing protocol and interrupt sampling real-time upload mechanism in the main control microprocessor, and upper-level visualization and monitoring software. This architecture is the core of this embodiment and is fundamental to achieving high-speed multimodal sampling based on a dexterous hand. The rest of this embodiment revolves around this system.

[0111] 2. Multimodal sensor array combination scheme To achieve multimodal signal sensing of the hand, this embodiment proposes a sensing unit architecture composed of a tactile array sensor, an adhesive thermocouple, and a flexible tension sensor. This embodiment proposes combining these three heterogeneous sensing devices to obtain richer modal sensing data. Corresponding sampling mechanisms and circuits are designed for each of these three heterogeneous sensing devices: the tactile array uses a row scanning mechanism combined with an inverting amplification circuit, the thermocouple uses a two-stage inverting amplification circuit, and the flexible tension sensor uses a voltage divider circuit. The spatial combination structure of these sensors on a time-dependent dexterous hand is also described in detail.

[0112] 3. Modular measurement circuit The protection of this measurement circuit involves the circuit's composition, mechanism, and structure: Circuit composition: It consists of a multi-mode signal conditioning module and an analog switch (RS2251). The front stage is connected to various sensor inputs, and the back stage outputs 8 analog signals to the AD converter. Multiple multi-mode signal conditioning modules can be connected in series on the same bus. Row scanning mechanism: A CMOS analog integrated circuit configured as an 8-channel multiplexer using RS2251 is used as an analog switch. The row scanning is controlled by the CH_A, CH_B, and CH_C chip select signals to sample the matrix points of each row of the tactile array sensor in sequence. Easy expansion mechanism: New multimodal signal conditioning modules can be directly connected to the lower-level output interface on the same bus to achieve expansion without redesigning the system architecture; Module structure: A four-layer PCB and a top-bottom double-sided layout are used to compress space and realize the measurement circuit design at the knuckle level.

[0113] 4. Multiple AD chip reuse mode This model includes a reuse mechanism and a reuse protocol: Hardware multiplexing mechanism: Six AD7606 chips share a 16-bit FSMC data bus and control signals (CONVST, RD, etc.), and are selected in a time-division manner only through their respective independent chip select pins (CS1~CS6), so as to achieve large-scale parallel data acquisition with limited pin resources; FSMC multiplexing protocol: The microcontroller sequentially pulls down the chip select level of each chip by polling, and uses the same set of parallel port buses to complete the reading of data from all chips.

[0114] 5. MCU sampling and reading mechanism This mechanism comprises two parts: interrupt sampling and main function processing, and their collaborative working mode: The interrupt service routine acts as a producer, triggered by the falling edge of the BUSY signal, and acquires AD7606 data at high speed via the FSMC protocol and fills it into the circular buffer. The main function's data processing flow acts as a consumer, polling the buffer and reading data to complete data reassembly, protocol encapsulation, and uploading. Collaboration mechanism: The two are decoupled through a shared FIFO ring buffer, and work together to achieve high throughput and low latency continuous data acquisition and transmission.

[0115] The modular multimodal tactile sensing system for robots proposed in this embodiment has the following beneficial effects: 1. High-bandwidth real-time data acquisition capability This system achieves an effective sampling rate of 200Hz, meeting the real-time requirements of high-frequency controllers for tactile feedback. This achievement is attributed to the AD7606 high-speed analog-to-digital converter chip used in the data processing module. The AD7606 boasts extremely high conversion speed, completing a single 8-channel synchronous conversion in just 60μs, far superior to the sampling rate of traditional piezoresistive sensing systems, which is generally below 50Hz. Combined with an interrupt-driven precise sampling control strategy and a FIFO ring buffer mechanism, the system can stably complete continuous data acquisition and transmission of 288 channels at a conversion frequency of 7.2kHz, fundamentally overcoming the bottleneck of insufficient data throughput in existing technologies.

[0116] 2. Multimodal information fusion perception Compared to the limitations of existing technologies that rely solely on tactile pressure sensing, this embodiment integrates a PVDF piezoresistive sensor array, a bonded T-type thermocouple, and a carbon fiber flexible tension sensor to achieve coordinated sensing of multiple physical quantities. The system has the capability to simultaneously acquire tactile, temperature, and tension sensory modal information.

[0117] Meanwhile, this embodiment proposes a reasonable sensor unit arrangement and spatial stacking integration scheme: thermocouples and piezoresistive sensors are spatially reused in the same area, and flexible tension sensors are attached to the joints. The three work together to enable the robotic hand to identify various attributes of objects, such as contact pressure distribution, surface temperature, thermal conductivity, and joint motion state, through grasping and manipulation. Based on this scheme, external multimodal sensors can be integrated into the robotic hand in the future, and the system has the potential to be embedded in the robotic hand or other irregularly shaped objects.

[0118] 3. High-efficiency multi-chip reuse mechanism This system achieves efficient multiplexing of multiple AD7606 chips by extending the FSMC protocol, completing large-scale parallel data acquisition under the constraint of limited microcontroller pin resources. This achievement relies on a chip select signal multiplexing mechanism: six AD7606 chips share the same 16-bit FSMC data bus and control signals, and are time-divisionally selected only through their independent chip select pins (CS1~CS6), thus utilizing the same parallel bus to read data from all chips. This design increases the system's basic channel expansion capability to 48 (6 chips × 8 channels) without increasing hardware complexity. Furthermore, for the tactile sensor array, the system adopts an analog switch row scanning scheme, further expanding the acquisition channels from the basic 48 channels to 288 (6 chips × 8 channels × 6 rows) based on refined timing control, effectively resolving the contradiction between the number of pins and the number of acquisition channels in traditional solutions.

[0119] 4. Modular architecture and high scalability This system adopts a modular architecture design with bus input and output, possessing excellent scalability and ease of secondary development. Sampling circuit modules are cascaded via upper and lower level bus interfaces, allowing users to flexibly add or remove multimodal signal conditioning modules according to the needs of different dexterous hand configurations or mechanical grippers without redesigning the system architecture. For example, when expanding from two-knuckle to three-knuckle, only one multimodal signal conditioning module needs to be connected to the lower-level output interface to complete the expansion. This technical advantage stems from the unified 6-pin bus interface design of the multimodal signal conditioning modules, ensuring consistency between electrical connections and control logic.

[0120] 5. Compact integrated design The multimodal signal conditioning module of this system adopts a four-layer PCB and a top-bottom double-sided layout, achieving a compact size at the knuckle level (25mm×15mm×8mm). It can be directly integrated into the surface of the dexterous hand through 3D printing of shells or nuts and other structural components. Compared with the space-consuming external sensor solutions and bulky visual-tactile solutions in the existing technology, this system significantly reduces the integration difficulty, enabling external sensors to be truly embedded in the limited space of the dexterous hand.

[0121] 6. High cost-performance ratio This system achieves significant cost advantages while ensuring high performance. The cost of a complete single-hand configuration (including all sensors, multimodal signal conditioning modules, and data processing modules) is approximately RMB 4,000, as detailed in Table 2. Compared to existing high-performance commercial tactile sensors (such as the Pacini PX-6AX series, which cost over RMB 10,000 for a full-hand deployment), this system significantly reduces costs. Furthermore, compared to low-cost commercial sensors (such as the Ligan Technology RPPS series), this system has advantages in key indicators such as sampling rate (200Hz vs 20Hz), number of modalities (trimodal vs single-modal), and scalability.

[0122] Table 2 Price Details

[0123] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, several equivalent substitutions or obvious modifications can be made without departing from the concept of the present invention, and all such modifications, achieving the same performance or purpose, should be considered within the scope of protection of the present invention.

Claims

1. A modular multimodal tactile sensing method for robots, characterized in that, Includes the following steps: External physical quantities are detected by multiple distributed sensing units, including a tactile array sensor for detecting contact pressure, a thermocouple for detecting temperature, and a flexible tension sensor for detecting tensile tension. Multiple multimodal signal conditioning modules condition the electrical signals output by the corresponding sensing units to convert them into analog voltage signals. Each multimodal signal conditioning module has an upper-level input interface and a lower-level output interface, connected via a bus-style cascaded configuration. A high-speed data processing and communication module provides power and line scanning control signals to all cascaded multimodal signal conditioning modules and synchronously acquires data. The analog voltage signal output by the multi-modal signal conditioning module is converted into a digital signal; the digital signal is encapsulated into data packets according to a predefined communication protocol by the high-speed data processing and communication module and transmitted to the host system; wherein, the synchronous acquisition step adopts an interrupt-driven mechanism: a periodic conversion start signal simultaneously triggers multiple analog-to-digital converter chips to start conversion; when the conversion is completed, an interrupt service routine is triggered by a conversion status signal, in which each analog-to-digital converter chip is selected in a time-division multiplexing manner to read the converted data and store it in a buffer; in the main program loop, data is read from the buffer and encapsulated.

2. The method according to claim 1, characterized in that, The electrical signal output by the tactile array sensor is conditioned using an inverting amplifier circuit; the electrical signal output by the thermocouple is conditioned using a non-inverting amplifier circuit. The electrical signal output by the flexible tension sensor is conditioned using a voltage divider circuit.

3. The method according to claim 1 or 2, characterized in that, The interrupt service routine performs the following operations: selects an analog-to-digital converter chip based on the current index; reads the conversion data of multiple channels in the current row of the chip; stores the data in a first-in-first-out (FIFO) buffer; and updates the chip and row indexes.

4. The method according to claim 1, characterized in that, In the main program loop, after reading data from the buffer, the data is reassembled according to the analog-to-digital converter chip and row address information of the data source. After each complete sensor unit's data is collected, it is encapsulated into a data packet. The data packet contains a frame header, module identifier, tactile data, temperature data, tension data, and checksum.

5. A modular multimodal tactile sensing system for robots, characterized in that, include: A sensing unit, distributed on the surface of the object to be measured, includes a tactile array sensor for detecting pressure, a thermocouple for detecting temperature, and a flexible tension sensor for detecting tensile tension. At least one multimodal signal conditioning module is electrically connected to the sensing unit. This module includes a signal conditioning unit for receiving and conditioning the electrical signals output by the tactile array sensor, thermocouple, and flexible tension sensor. The module also includes a bus interface unit for receiving and outputting power, ground, and control signals from external sources. High-speed data processing and... A communication module is electrically connected to the multimodal signal conditioning module. The high-speed data processing and communication module includes a main control unit and at least one analog-to-digital conversion unit. The main control unit generates the control signal and sends it to the multimodal signal conditioning module through the bus interface unit. The analog-to-digital conversion unit synchronously acquires multiple analog signals conditioned by the signal conditioning unit and converts them into digital signals, as well as reads and processes the digital signals. The main control unit coordinates the generation of the control signal, the start of the conversion of the analog-to-digital conversion unit, and the reading of the digital signals through interrupt-driven timing control logic to achieve synchronous high-speed acquisition of data from the entire system.

6. A multimodal signal conditioning module, characterized in that, include: The signal conditioning unit is used to condition the electrical signal output by the connected sensing unit and convert it into an analog voltage signal output. The sensing unit includes a tactile array sensor, a thermocouple, and a flexible tension sensor. The upper-level input interface is used to connect power lines and control signal lines from the data processing module or the previous-level multimodal signal conditioning module; The lower-level output interface is used to lead the power line and control signal line to the next-level multi-mode signal conditioning module; An analog voltage output interface is used to output the analog voltage signal to the data processing module; wherein, the upper-level input interface and the lower-level output interface adopt the same bus protocol, enabling multiple multi-mode signal conditioning modules to be cascaded in a bus manner through the upper-level input interface and the lower-level output interface.

7. The multimodal signal conditioning module according to claim 6, characterized in that, The signal conditioning unit includes: an analog switch for selecting the signals of each row of the tactile array sensor; an inverting amplifier circuit for conditioning the signals of each column of the tactile array sensor; a two-stage non-inverting amplifier circuit for conditioning the output signal of the thermocouple; and a voltage divider circuit for conditioning the output signal of the flexible tension sensor.

8. A high-speed data processing and communication module, characterized in that, include: microcontroller; Multiple analog-to-digital converter (ADC) chips are provided, with each ADC's analog input channel connected to the analog voltage output interface of the multimodal signal conditioning module as described in claim 6 or 7 to receive analog voltage signals. All ADC chips share the microcontroller's data bus and control signal lines (excluding the chip select signal). Each ADC chip is connected to an independent chip select pin of the microcontroller. The microcontroller is configured to: generate periodic conversion start signals and simultaneously output them to all ADC chips to trigger synchronous conversion; enter an interrupt service routine in response to a conversion status signal issued by any ADC chip after conversion completion; in the interrupt service routine, select each ADC chip in a time-division multiplexing manner by controlling the chip select signal, and read the conversion data of the selected chip through the data bus and store it in a buffer; in the main program loop, read data from the buffer and upload it via the communication interface.

9. A robotic dexterous hand, comprising a palm and multiple fingers, characterized in that, It also includes a plurality of multimodal signal conditioning modules as described in claim 6 or 7 and a high-speed data processing and communication module as described in claim 8; the plurality of multimodal signal conditioning modules are connected in a bus-cascade manner and distributed and integrated on the surface of the palm and fingers; the tactile array sensor and thermocouple connected to the multimodal signal conditioning module are disposed on the tactile surface of the palm and fingers, and the flexible tension sensor is connected across the joint of the finger.

10. A robot, characterized in that, Including the robotic dexterous hand as described in claim 9.

Images