A humanoid robot fall control method, device, equipment and storage medium

Abstract

The application provides a humanoid robot fall control method, device, equipment and storage medium, relating to the technical field of humanoid robots. The control method comprises: acquiring state data of the humanoid robot in real time; predicting the current state of the humanoid robot according to the state data; and controlling the humanoid robot to perform a fall protection action if the current state is a high-risk fall state. The humanoid robot fall control method of the application improves the accuracy of fall prediction by monitoring the center of gravity position, movement speed and acceleration in real time. The method can quickly start the protection action in a high-risk situation, reduce the risk of injury, and protect the safety of surrounding personnel. It enhances the adaptability and autonomy of the robot, improves the stability and safety, reduces the maintenance cost, and prolongs the service life, thereby enhancing the practicality and market competitiveness of the robot and widening the application range.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of humanoid robots, and in particular to a humanoid robot fall control method, device, equipment and storage medium. BACKGROUND

[0002] As an important branch of robot technology, humanoid robots have shown great potential for application in various fields such as service, rescue, and entertainment due to their ability to mimic human behavior. The design of such robots aims to simulate the bipedal walking mechanism of humans, enabling them to perform tasks in complex and changing environments. However, due to the high dynamic nature and environmental uncertainty, humanoid robots face the risk of falling when performing tasks, which not only relates to the safety of the robot itself but also involves the safety of surrounding personnel.

[0003] In the prior art, the fall prediction of humanoid robots mainly relies on the data provided by the inertial measurement unit (IMU). The IMU can measure the linear acceleration and angular velocity of the robot, providing basic data for fall detection. However, this single sensor method has limitations in capturing subtle changes in the robot's posture, especially in complex environments, where the robot may lose balance due to various factors such as uneven ground, external force interference, etc. This limits the accuracy and real-time performance of fall prediction, causing the robot to fail to respond in time when facing the risk of falling.

[0004] In addition, existing fall protection schemes mostly rely on mechanical structures or simple emergency braking mechanisms. These schemes often lack flexibility in design and cannot adapt to changing fall scenarios. For example, some protection mechanisms may be effective when the robot falls backward, but may not provide sufficient protection when falling sideways. This limitation makes existing protection measures ineffective in practical applications, failing to effectively reduce the damage caused by falls to the robot.

[0005] In summary, the existing technology has significant shortcomings in the fall prediction and protection of humanoid robots. The limitations of IMU data affect the accuracy of fall prediction, while existing protection schemes lack flexibility and adaptability, making it difficult to cope with complex fall scenarios. These problems limit the stability and safety of humanoid robots in complex environments, increasing the risk of robot operation. Therefore, developing more accurate and flexible fall prediction and protection technology is of great significance for improving the practicality and reliability of humanoid robots. SUMMARY

[0006] In a first aspect, the present application provides a humanoid robot fall control method, comprising:

[0007] real-time acquisition of state data of the humanoid robot; the state data includes center of gravity position, motion speed and acceleration;

[0008] predicting a current state of the humanoid robot according to the state data;

[0009] if the current state is a high-risk falling state, controlling the humanoid robot to perform a falling protection action.

[0010] In an optional implementation, the predicting a current state of the humanoid robot according to the state data comprises:

[0011] adopting a model predictive control algorithm, taking the center of gravity position, the motion speed and the acceleration as inputs, and judging whether the humanoid robot reaches a preset trigger condition by using a robot whole-body dynamics model;

[0012] if yes, determining that the current state of the humanoid robot is a high-risk falling state;

[0013] if no, determining that the current state of the humanoid robot is not a high-risk falling state.

[0014] In an optional implementation, the preset trigger condition comprises at least one of the following conditions:

[0015] A. the center of gravity position exceeds a preset safety range;

[0016] B. a change of the acceleration exceeds a preset acceleration change threshold;

[0017] C. a descending speed of the motion speed exceeds a preset descending threshold.

[0018] In an optional implementation, the falling protection action comprises:

[0019] confirming a falling direction and a current robot key protection area corresponding to the falling direction; the current robot key protection area comprises a leg, a fragile joint and a collision joint; the fragile joint comprises a hand joint and a head joint; the collision joint comprises at least one of a knee joint, a hip joint and an elbow joint;

[0020] controlling the leg to form a bent leg state to lower the center of gravity of the humanoid robot;

[0021] controlling the fragile joint to adjust to a safe posture;

[0022] controlling the collision joint to open a compliant collision function;

[0023] wherein, the safe posture comprises at least one of the following methods:

[0024] A. adjusting the hand joint to form a fist state;

[0025] B, adjust the elbow joint of the hand to the chest area;

[0026] C, adjust the head to tilt towards the chest direction to form a bowing state.

[0027] In an optional embodiment, the confirming the falling direction and the current robot focus protection area corresponding to the falling direction comprises:

[0028] Real-time monitoring of the attitude angle data of the humanoid robot using an inertial measurement unit; the attitude angle data includes the absolute value of the pitch angle and / or the roll angle;

[0029] If the attitude angle reaches 0.5 radian, it is determined that the humanoid robot is currently in an unstable state, and the leg and / or the fragile joint are taken as the current robot focus protection area;

[0030] If the attitude angle reaches 1.05 radian, it is determined that the humanoid robot is currently in a tipping state, and the collision joint is taken as the current robot focus protection area.

[0031] In an optional embodiment, the compliant collision function comprises:

[0032] Real-time monitoring of the external force and actual speed acting on the collision joint; when the external force acting on the collision joint is perceived, the joint acceleration generated by the joint is calculated according to a dynamics model; the expression of the dynamics model is:

[0033]

[0034] Where F represents the falling collision force; M represents the mass parameter of the collision joint; D represents the damping parameter of the collision joint; v real represents the actual speed; represents the rate of change of the joint acceleration of the collision joint;

[0035] Controlling the collision joint to conform to the external force movement using the joint acceleration; and dynamically adjusting the damping parameter and the mass parameter according to the actual speed of the collision joint.

[0036] In an optional embodiment, after the humanoid robot performs the falling protection action, the method further comprises:

[0037] Performing fault self-checking, attitude position detection and environment perception on the humanoid robot, and obtaining self-checking results, attitude position information and environment perception results, respectively;

[0038] Establishing motion planning data according to the self-checking results, the attitude position information and the environment perception results;

[0039] determining whether the humanoid robot can simulate recovery to a stable posture according to the motion planning data;

[0040] If yes, adjusting the current posture of the humanoid robot according to the motion planning data; using a segmented motion strategy to adjust the position and angle of the upper body and / or lower body until the stable posture is recovered;

[0041] If no, determining that the motion is blocked, stopping posture adjustment, and generating a warning.

[0042] In a second aspect, the present application provides a humanoid robot fall control device, comprising:

[0043] a monitoring module for acquiring state data of the humanoid robot in real time; the state data includes center of gravity position, motion speed and acceleration;

[0044] a prediction module for predicting the current state of the humanoid robot according to the state data;

[0045] a protection module for controlling the humanoid robot to perform a fall protection action when the current state is a high-risk fall state.

[0046] In a third aspect, the present application provides a computer device, comprising a processor and a memory, the memory storing a computer program, and the processor is configured to execute the computer program to implement the humanoid robot fall control method according to any one of the preceding embodiments.

[0047] In a fourth aspect, the present application provides a computer storage medium storing a computer program, which, when executed on a processor, implements the humanoid robot fall control method according to any one of the preceding embodiments.

[0048] The humanoid robot fall control method provided in the present application effectively overcomes the limitations of traditional methods that rely on single IMU data by acquiring state data including center of gravity position, motion speed and acceleration in real time, improving the accuracy of fall prediction. It can immediately control the robot to perform a protection action when a high-risk fall state is predicted, providing a more flexible and timely response mechanism, reducing the risk of structural damage and the safety threat to surrounding personnel. In addition, this method enhances the environmental adaptability and autonomy of the robot, improves its stability and safety in complex environments, reduces maintenance and operating costs, and prolongs the service life of the robot. In summary, this method significantly improves the practicality and market competitiveness of humanoid robots, making them have a wider application prospect in various scenarios such as home, industry and rescue. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the present application, the drawings required to be used in the embodiments will be briefly introduced as follows. It should be understood that the following drawings only show some of the embodiments of the present application, and therefore should not be regarded as limiting the scope of protection of the present application. For those of ordinary skill in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0050] Figure 1 The structural schematic diagram of the hardware running environment related to the humanoid robot fall control method embodiment of the present application;

[0051] Figure 2 The flowchart of the first embodiment of the humanoid robot fall control method of the present application;

[0052] Figure 3 The flowchart of the step S200 in the second embodiment of the humanoid robot fall control method of the present application;

[0053] Figure 4 The overall flowchart of the third embodiment of the humanoid robot fall control method of the present application, including the step S300;

[0054] Figure 5 The flowchart of the third embodiment of the humanoid robot fall control method of the present application, including the step S310;

[0055] Figure 6 The supplementary flowchart of the steps S400-S800 in the fourth embodiment of the humanoid robot fall control method of the present application;

[0056] Figure 7 The module connection schematic diagram of the humanoid robot fall control device of the present application;

[0057] Figure 8 The overall structure and working flowchart of the humanoid robot fall control device of the present application;

[0058] Figure 9 The flowchart of the humanoid robot fall control device execution method of the present application, for three different stages. DETAILED DESCRIPTION

[0059] The technical solutions in the embodiments of the present application will be described clearly and completely in the embodiments of the present application in combination with the drawings. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments.

[0060] The components of the application embodiments described and illustrated herein can be arranged and designed in a wide variety of different configurations. Therefore, the following detailed description of the application, read with reference to the accompanying drawings, is not intended to limit the scope of the application as claimed, but is merely representative of selected embodiments of applications. The description of the embodiments of applications based herein will allow those skilled in the art to

[0061] Hereinafter, the terms "include", "have", and their conjugates herein can only mean to indicate a certain characteristic, number, step, operation, element, component or a combination thereof, and should not be construed as excluding the existence or possibility of adding one or more other characteristics, numbers, steps, operations, elements, components or combinations thereof.

[0062] In addition, the terms "first", "second", "third", and the like are used only to distinguish descriptions, and should not be understood as indicating or implying relative importance.

[0063] Unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which various embodiments of the present application belong. The terms (such as those defined in commonly used dictionaries) will be interpreted as having a meaning that is the same as the contextual meaning in the relevant technical field and will not be interpreted in an idealized or overly formal sense unless clearly defined in various embodiments of the present application.

[0064] Some embodiments of the present application will be described below in detail with reference to the accompanying drawings. The following embodiments and features in the embodiments can be combined with each other without conflict.

[0065] As Figure 1 shown is a structural schematic diagram of a hardware operating environment of a terminal related to an embodiment of the present application.

[0066] The humanoid robot fall control system according to the embodiments of the present application can be a PC, a smart phone, a tablet computer, a portable computer or a mobile terminal device, etc. The humanoid robot fall control system can include a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005 and a communication bus 1002. The communication bus 1002 is used to realize the connection and communication between the components. The user interface 1003 can include a display screen, an input unit such as a keyboard, a remote controller, and the optional user interface 1003 can further include a standard wired interface, a wireless interface. The network interface 1004 can optionally include a standard wired interface, a wireless interface (such as a WI-FI interface). The memory 1005 can be a high-speed RAM memory or a stable memory such as a disk memory. The memory 1005 can optionally be a storage device independent of the aforementioned processor 1001. Optionally, the humanoid robot fall control system can further include an RF (Radio Frequency, RF) circuit, an audio circuit, a WiFi module, etc. In addition, the humanoid robot fall control system can also be configured with a gyroscope, a barometer, a hygrometer, a thermometer, an infrared sensor and other sensors, which will not be described here.

[0067] Those skilled in the art can understand that Figure 1 The humanoid robot fall control system shown in the above embodiments does not constitute a limitation thereon, and can include more or fewer components than shown, or combine certain components, or different component arrangements. For example, Figure 1 As shown, the memory 1005 as a computer readable storage medium can include an operating system, a data interface control program, a network connection program and a humanoid robot fall control program.

[0068] In summary, the method provided by the present application improves the accuracy of fall prediction by monitoring key state data in real time, and can quickly respond to high-risk fall states and perform protection actions. This method reduces the damage to the robot structure, reduces the safety risk to the surrounding personnel, and improves the environmental adaptability and economy of the robot, and enhances its stability and reliability in a variable environment.

[0069] Embodiment 1:

[0070] With reference to Figure 2 The present embodiment provides a humanoid robot fall control method, comprising:

[0071] Step S100, real-time acquisition of state data of the humanoid robot; the state data includes the position of the center of gravity, the movement speed and the acceleration.

[0072] The humanoid robot mentioned in this embodiment, i.e. the humanoid robot, is a high-tech product designed to imitate the appearance and behavior of humans. They have structures similar to human heads, torsos, arms, and legs, can move flexibly and perform various actions. These robots usually have multiple degrees of freedom, can walk on two feet, adapt to uneven ground and stairs, making them seamlessly integrated into the environment of human life and work. Humanoid robots are also equipped with advanced sensors and artificial intelligence technologies, including machine learning, natural language processing, and computer vision, to make autonomous decisions and perform complex tasks. They have wide applications in service industries, education, entertainment, industrial automation, disaster rescue, and other fields, and as research platforms, they have driven the development of related technologies such as motion control, human-computer interaction, and cognitive science. With the continuous advancement of technology, the autonomy and intelligence level of humanoid robots are constantly improving, and their role and influence in human society are also expanding.

[0073] Humanoid robots require specific fall control methods, mainly because they imitate the bipedal walking mechanism of humans, which makes them face more complex challenges in dynamic balance control, especially when adapting to complex environments such as uneven ground, stairs, and narrow spaces. In addition, to ensure safety when working in an environment interacting with humans and to avoid injuries to personnel and equipment caused by falls, effective fall control methods are crucial. Finally, fall control methods help reduce maintenance costs, extend service life, and improve task execution efficiency, thereby enhancing the economic and practicality of humanoid robots.

[0074] In the above steps, real-time monitoring of key state data of the humanoid robot using sensors is involved. State data includes center of gravity position, motion speed, and acceleration, which are crucial for assessing the stability of the robot and predicting the risk of falling. By obtaining these data in real time, the motion state and possible imbalance trend of the robot can be accurately captured, providing necessary information for subsequent fall prediction.

[0075] Specifically, it can be achieved through sensors integrated inside the robot, such as inertial measurement units (IMU), force sensors, etc. IMU can provide data on acceleration and angular velocity, while force sensors can provide information on the force of the robot in contact with the ground, which together help determine the center of gravity position.

[0076] Step S200, predicting the current state of the humanoid robot according to the state data.

[0077] As mentioned above, the state data obtained in the first step is analyzed, and an algorithm is used to predict the current stability state of the robot to determine whether there is a risk of falling. The prediction result can help identify whether the robot is in a high-risk falling state, so that measures can be taken in advance.

[0078] Specifically, machine learning algorithms such as random forest, support vector machine (SVM), or neural network can be used to train the model based on historical data to predict the risk of falling.

[0079] Step S300, if the current state is a high-risk falling state, the humanoid robot is controlled to perform a fall protection action.

[0080] When the prediction result shows that the robot is in a high-risk falling state, the system will automatically trigger the preset fall protection action to reduce the damage caused by falling. Performing a fall protection action can significantly reduce the damage to the robot when it falls, protect critical components, and ensure that the robot can quickly recover to normal state.

[0081] The fall protection action can be achieved through a pre-set control strategy, such as adjusting the joint angle to absorb the impact, starting the compliant control to reduce joint damage, etc. The specific control algorithm can use PID control or more advanced adaptive control algorithm.

[0082] The humanoid robot fall control method provided in this embodiment effectively overcomes the limitations of traditional methods that rely on single IMU data by acquiring state data including center of gravity position, motion speed and acceleration in real time, improving the accuracy of fall prediction. It can immediately control the robot to perform a protection action when a high-risk falling state is predicted, providing a more flexible and timely response mechanism, reducing the risk of structural damage, and reducing the safety threat to surrounding personnel. In addition, this method enhances the environmental adaptability and autonomy of the robot, improves its stability and safety in complex environments, reduces maintenance and operating costs, and prolongs the service life of the robot. In summary, this method significantly improves the practicality and market competitiveness of humanoid robots, making them have a wider application prospect in various scenarios such as home, industry, rescue, etc.

[0083] Embodiment 2:

[0084] Referring to Figure 3 , this embodiment provides a humanoid robot fall control method, based on the above embodiment 1, the step S200, predicting the current state of the humanoid robot according to the state data, including:

[0085] Step S210, using a model predictive control algorithm, taking the center of gravity position, the motion speed and the acceleration as input, using the robot whole-body dynamics model to determine whether the humanoid robot reaches a pre-set trigger condition.

[0086] The above step involves using a Model Predictive Control (MPC) algorithm to process the real-time acquired state data. MPC is an advanced control strategy that uses the robot's dynamics model to predict future behavior and optimize control inputs. By using MPC, the robot's future state can be predicted based on its dynamics model, allowing for more accurate fall risk assessment.

[0087] Implementing MPC requires establishing a mathematical model of the robot that describes its dynamic behavior, including the relationship between center of mass position, motion velocity, and acceleration.

[0088] The above, the real-time acquired center of mass position, motion velocity, and acceleration data are used as input parameters for the MPC algorithm. These data provide the current motion state of the robot, which the MPC algorithm can use to predict future states and make more accurate control decisions.

[0089] These data can be obtained directly by sensors, such as IMU sensors providing acceleration and angular velocity data, and motion velocity can be obtained by integrating acceleration data.

[0090] The above, the full-body dynamics model of the robot is used to evaluate the current state data and determine whether the pre-set fall risk trigger condition is met. Through the full-body dynamics model, the motion state of the robot can be more comprehensively understood, allowing for more accurate identification of fall risk. Specifically, a complex mathematical model may be required that can describe the dynamic behavior of all joints and body parts of the robot. This model can be a physics-based model or a data-driven model.

[0091] Step S220, if yes, then determine that the current state of the humanoid robot is a high-risk fall state.

[0092] Step S230, if no, then determine that the current state of the humanoid robot is not a high-risk fall state.

[0093] The above, according to the prediction result of the MPC algorithm to finally determine whether the robot is in a high-risk fall state. This judgment provides a clear basis for decision-making, allowing the robot to quickly take protective measures when a high-risk fall is predicted.

[0094] A series of thresholds or rules can be set, and when the prediction result of the MPC algorithm exceeds these thresholds, it is determined to be a high-risk fall state.

[0095] Further, the pre-set trigger condition includes at least one of the following conditions:

[0096] A, the center of mass position exceeds the pre-set safety range.

[0097] The above, the center of gravity position of the humanoid robot is monitored, and compared with the preset safety range. If the center of gravity position of the robot exceeds this range, it may mean that the robot is about to lose balance. By monitoring the center of gravity position, risks can be identified in time before the robot falls, so that preventive measures can be taken. The center of gravity position can be monitored in real time by sensors integrated inside the robot, such as IMU. Then, the monitored center of gravity position is compared with the preset safety range.

[0098] For example, assuming C is the current center of gravity position, C safe is the preset safety range, then: if ∣C-C safe ∣>∈, trigger the fall protection action. Where ∈ is a set threshold value, representing the boundary of the safety range.

[0099] B, the change of acceleration exceeds the preset acceleration change threshold.

[0100] The above condition involves monitoring the acceleration change of the humanoid robot, and comparing it with the preset acceleration change threshold. If the acceleration change exceeds this threshold, it may mean that the robot encounters sudden external force or violent motion. By monitoring the acceleration change, sudden movements or impacts that may cause falls can be detected.

[0101] The change of acceleration can be monitored by IMU sensors. Then, the rate of change of acceleration is calculated and compared with the preset threshold.

[0102] For example, assuming a is the current acceleration, a th is the preset acceleration change threshold, then: if ∣a-a prev ∣>a th , trigger the fall protection action, where a prev is the acceleration at the previous moment.

[0103] C, the speed of motion decreases at a rate exceeding the preset decrease threshold.

[0104] The above condition involves monitoring the speed of motion of the humanoid robot, and comparing it with the preset decrease threshold. If the speed decreases too quickly, it may mean that the robot is losing power or control. By monitoring the decrease of motion speed, the possible out-of-control state of the robot can be identified, and measures can be taken in time to prevent falls. The change of motion speed can be monitored by speed sensors. Then, the rate of change of speed is calculated and compared with the preset threshold.

[0105] For example, assuming v is the current speed, v th is the preset speed decrease threshold, then: if d v / dt <-v th , then a fall protection action is triggered; wherein d v / d t represents the rate of change of velocity.

[0106] Embodiment 3:

[0107] With reference to Figure 4 , this embodiment provides a humanoid robot fall control method, based on the above embodiment 1, further, in step S300, the fall protection action includes:

[0108] Step S310, confirming the fall direction and the current robot key protection area corresponding to the fall direction. The current robot key protection area includes the leg, fragile joint and collision joint; the fragile joint includes the hand joint and head joint; the collision joint includes at least one of the knee joint, hip joint and elbow joint.

[0109] It should be noted that confirming the fall direction and determining the corresponding key protection area is crucial for the fall control of humanoid robots, as it allows the robot to make targeted protection, optimize the allocation of limited control resources, reduce the risk of secondary injury, improve the efficiency of recovery from the fall state, plan adaptive actions to mitigate impact, enhance safety in human environments, improve the reliability and robustness of the robot, and improve the interactive experience with humans. Through such prediction and protection measures, the humanoid robot can respond more quickly and effectively when facing a fall situation, thereby minimizing damage and ensuring the safety of the robot and surrounding personnel.

[0110] The above steps involve quickly identifying the direction of the fall after predicting the fall risk, and determining the areas that need to be protected by the robot, such as the leg, fragile joint (hand joint and head joint) and joint that may collide (knee joint, hip joint and elbow joint). By quickly identifying the fall direction and key protection area, targeted protection actions can be performed to reduce damage.

[0111] The fall direction can be determined by analyzing sensor data, such as accelerometer and gyroscope data. Then, according to the fall direction, the area that needs to be protected is determined.

[0112] Step S320, controlling the leg to form a bent leg state, lowering the center of gravity of the humanoid robot.

[0113] This step involves, when a fall is predicted, bending the robot's legs by controlling the leg joints to lower the center of gravity and reduce the impact force during the fall. Lowering the center of gravity can increase stability and reduce injury during the fall. Specifically, the leg joints can be controlled by electric servo motors to achieve the leg bending action.

[0114] Step S330, control the fragile joint to adjust to a safe posture.

[0115] As described above, before falling or touching the ground, adjust the hand joint and head joint to a safe posture, such as a fist, elbow flexion to the chest area, and head tilt to a low head state. Adjusting the fragile joint to a safe posture can reduce the damage to these parts when falling. A series of safe postures can be preset, and when a falling risk is detected, the joints are adjusted to these postures through a control algorithm.

[0116] The safe posture includes at least one of the following methods:

[0117] A, adjust the hand joint to a fist state;

[0118] B, adjust the hand joint to flex the elbow to the chest area;

[0119] C, adjust the head to tilt towards the chest direction to form a low head state.

[0120] As described above, at the initial stage of falling, the robot uses motion planning to clench the hand joint and bend the elbow to protect the hand in front of the chest. When falling backward, in order to avoid head impact, at the initial stage of falling, the robot uses motion planning to lower the head downward.

[0121] Step S340, control the collision joint to open the soft collision function.

[0122] As described above, the soft collision function refers to the joint following the direction and size of the external force when subjected to external force, and through this elastic collision method, the impact on the joint can be reduced.

[0123] As described above, when falling, the soft function of joints such as knee joint, hip joint and elbow joint that are prone to collision is opened to reduce the impact. The soft function can provide a certain deformation space for the joint when subjected to impact, reducing damage to the joint. The stiffness parameter of the joint can be changed, or a special soft control algorithm can be used to achieve it.

[0124] Specifically, the specific method of implementing the fall protection action can include, first, using sensor data (such as IMU) to predict the falling direction and calculate the falling time; then, using PID control or other robot control algorithms to accurately adjust the joint angle; and using a soft control algorithm to dynamically adjust the stiffness of the joint to adapt to different impact situations.

[0125] It should be noted that steps S320 to S340 can be performed in any order, or simultaneously.

[0126] Further, with reference to Figure 5 , the step S310 of confirming the falling direction and the current robot key protection area corresponding to the falling direction comprises:

[0127] Step S311, real-time monitoring of the attitude angle data of the humanoid robot using an inertial measurement unit; the attitude angle data includes the absolute value of the pitch angle and / or the roll angle.

[0128] It should be noted that the pitch angle, also known as the pitch angle, is used to describe the rotation angle of the robot around its horizontal axis (left-right direction). A positive value indicates that the robot's head is rotated upward and the tail is rotated downward, and a negative value indicates rotation in the opposite direction. The roll angle, also known as the roll angle, is used to describe the rotation angle of the robot around its vertical axis (front-back direction). A positive value indicates that the right side of the robot is rotated upward and the left side is rotated downward, and a negative value indicates rotation in the opposite direction.

[0129] The pitch angle and roll angle are key parameters for evaluating the stability of the humanoid robot's posture. They not only help predict the falling direction, such as by analyzing their absolute values to determine whether the robot is leaning forward or backward, but also trigger protection actions when they exceed a pre-set threshold to stabilize the robot or initiate emergency protection measures. In addition, controlling these attitude angles is crucial for maintaining dynamic balance, allowing the robot to adjust gait and joint angles in real time to adapt to different ground conditions and movement requirements. Accurate monitoring of the pitch angle and roll angle is also very important for improving the safety of humanoid robots, helping to avoid falls and reduce the risk of injury, especially in environments interacting with humans. Therefore, the absolute values of pitch and roll are crucial for evaluating and controlling the stability of the humanoid robot's posture, predicting the risk of falling, triggering protection actions, and maintaining dynamic balance.

[0130] In the above steps, an inertial measurement unit (IMU) is used to monitor the pitch angle and / or roll angle of the humanoid robot in real time. By monitoring these attitude angles, the dynamic stability of the robot can be understood in real time, providing key data for subsequent fall risk assessment.

[0131] IMU sensors can provide continuous attitude angle data that can be directly used for subsequent analysis and processing.

[0132] Step S312, if the attitude angle reaches 0.5 radians, it is determined that the humanoid robot is currently in an unstable state, and the legs and / or vulnerable joints are considered as the current robot key protection area.

[0133] The monitored attitude angle data is compared with a preset threshold value (0.5 radian) to determine whether the robot is in an unstable state. Setting the threshold value can help quickly identify the unstable state of the robot and take timely measures, such as focusing on protecting the leg and fragile joints as the key protection area.

[0134] A simple comparison algorithm can be used, and when the absolute value of the attitude angle exceeds 0.5, the corresponding protection measures are triggered. If | attitude angle | > 0.5 radian, it is determined to be in an unstable state.

[0135] In step S313, if the attitude angle reaches 1.05 radians, it is determined that the humanoid robot is currently in a tipping state, and the collision joint is set as the current robot key protection area.

[0136] The attitude angle data is compared with another preset threshold value (1.05 radian) to determine whether the robot is in a tipping state. A higher threshold value indicates a more serious unstable state, and the joint that may collide needs to be protected.

[0137] Similar to the aforementioned step S312, a comparison algorithm is used, but a higher threshold value is used to trigger more urgent protection measures. If | attitude angle | > 1.05 radian, it is determined to be in a tipping state.

[0138] Further, in step S340, the compliant collision function includes:

[0139] In step S341, the external force and actual speed acting on the collision joint are monitored in real time; when the external force acting on the collision joint is sensed, the joint acceleration generated by the joint is calculated according to the dynamics model.

[0140] The external force and actual speed acting on the collision joint (such as the knee joint, hip joint, and elbow joint) of the humanoid robot are monitored in real time using sensors. By monitoring the external force and speed, the external impact can be responded to in time, and joint damage can be reduced.

[0141] These parameters can be monitored by force sensors and speed sensors, or IMU sensors can be used to estimate speed and acceleration.

[0142] When the external force acts on the joint, the joint acceleration is calculated using the dynamics model.

[0143] For example, taking the x direction of the hip joint as an example, the robot and the external environment interaction model can be simplified as a mass-spring-damper model, and its dynamics equation can be expressed as:

[0144]

[0145] wherein F represents the falling impact force; M represents the mass parameter of the impact joint; D represents the damping parameter of the impact joint; v real represents the actual velocity; x real represents the actual spatial position; represents the rate of change of the joint acceleration of the impact joint; v des represents the desired velocity; x des represents the desired spatial position; K represents the stiffness parameter of the impact joint;

[0146] Since the humanoid robot falls and follows the impact external force movement, there is:

[0147] v des = 0; the stiffness parameter K = 0, so the above formula 1 can be further simplified, and the following expression of the kinetic model is obtained:

[0148]

[0149] wherein F represents the falling impact force; M represents the mass parameter of the impact joint; D represents the damping parameter of the impact joint; v real represents the actual velocity; represents the rate of change of the joint acceleration of the impact joint.

[0150] The above model helps to calculate the movement of the joint under the action of the external force, so that the joint can conform to the external force and reduce the impact damage.

[0151] Step S342, controlling the impact joint to conform to the external force movement according to the joint acceleration; and dynamically adjusting the damping parameter and the mass parameter according to the actual velocity of the impact joint.

[0152] According to the calculated joint acceleration, the movement of the joint is controlled to conform to the external force, and the compliant impact is realized. By conforming to the external force, the joint can reduce the impact and damage, and improve the safety and reliability of the robot. The compliant control can be realized by adjusting the control algorithm of the joint, such as the PID controller.

[0153] Then, the damping parameter D and the mass parameter M are dynamically adjusted according to the actual velocity of the joint to optimize the effect of compliant impact. Dynamic adjustment of parameters can make the response of the joint more adaptive to the current motion state, and improve the protection effect. Specifically, the feedback control algorithm can be used to adjust D and M in real time according to the change of the actual velocity.

[0154] Embodiment 4:

[0155] Referring to Figure 6The embodiment provides a humanoid robot fall control method, based on the above embodiment 1, after the step S300 of controlling the humanoid robot to perform the fall protection action, further comprising:

[0156] Step S400, performing fault self-checking, posture position detection and environment perception on the humanoid robot, and obtaining self-checking results, posture position information and environment perception results respectively.

[0157] It should be noted that the fault self-checking, posture position detection and environment perception after performing the fall protection action are to comprehensively evaluate the state and safety of the robot. The fault self-checking ensures that the robot has no structural damage or functional failure after falling, and the posture position detection provides accurate position and posture information of the robot, which is crucial for planning the recovery action. The environment perception ensures that the surrounding environment is safe during the robot recovery process, and there are no obstacles to hinder the recovery action or cause additional risks to the robot and surrounding personnel. These steps help to avoid secondary injury, improve recovery success rate, and enable the robot to adapt to changing environmental conditions. At the same time, the collected data can be used for subsequent analysis to improve the design and fall recovery algorithm of the robot, reduce the risk of future falls, enhance the autonomy and intelligence level of the robot, and ensure compliance with safety standards and regulatory requirements.

[0158] The above steps involve a series of checks and detections on the robot after performing the fall protection action, including fault self-checking, posture position detection and environment perception. Through these detections, the health status, current posture and surrounding environment of the robot can be evaluated to provide necessary information for subsequent recovery actions. Fault self-checking can be achieved through built-in diagnostic systems, posture position detection can be achieved through IMU and other sensors, and environment perception can be achieved through vision systems and sensors.

[0159] For example, first, the humanoid robot performs fault self-checking on each part to detect whether it is structurally damaged, such as joint damage, motor failure, etc. Then, use built-in sensors (such as gyroscopes, accelerometers) to detect the posture position of the robot, i.e. to detect the current posture and position. And, use the vision system of the robot to perform environment perception to confirm that there are no obstacles in the surrounding environment that will hinder the standing process. Use a physics engine or dynamic model to simulate the process of the robot standing up from the current state to evaluate whether it is possible to succeed. Determine that the robot has normal movement conditions, and then start the recovery program.

[0160] Step S500, according to the self-checking results, the posture position information and the environment perception results, establishing motion planning data.

[0161] It is necessary to point out that before adjusting the current posture of the humanoid robot, the motion planning data is first established and simulated, in order to ensure that the planned recovery action is both safe and feasible, while optimizing the recovery path to improve efficiency. This step allows the robot to predict and avoid potential risks before actual action execution, verify the adaptability and flexibility of the recovery action, and reduce damage and maintenance costs caused by incorrect operation. In addition, it enhances the robot's autonomous decision-making ability, improves the reliability of executing complex tasks, and helps to establish user trust in the performance of the robot. Through data-driven decision-making, the robot can more scientifically and accurately handle abnormal situations, ensuring safe and effective recovery to a stable state in uncertain or dynamic environments.

[0162] According to the detection results, the motion planning data is established to guide the recovery action of the robot. The motion planning data can help the robot determine how to safely and effectively recover from the current posture to a stable posture. Path planning algorithms such as A* algorithm or RRT algorithm can be used to generate motion planning data in combination with the robot's dynamics model and environmental information.

[0163] The A* algorithm (A Star Algorithm) is a heuristic search algorithm used to find the shortest path from the starting point to the target point in a graph. It combines the characteristics of Dijkstra's algorithm (guaranteed to find the shortest path) and greedy best-first search (selects the most promising node for expansion at each step).

[0164] The RRT algorithm (Rapidly-exploring Random Tree) is a sampling-based tree search algorithm used to solve path planning problems in unstructured environments. It is particularly suitable for high-dimensional spaces and complex obstacle environments.

[0165] Both algorithms are suitable for robot path planning, and the choice of algorithm depends on the specific application scenario and requirements. A* algorithm is suitable for grid maps and known environment path planning, while RRT algorithm is suitable for continuous space and complex environment path planning.

[0166] Step S600, determine whether the humanoid robot can simulate recovery to a stable posture according to the motion planning data.

[0167] This step involves simulating the possibility of the robot recovering to a stable posture according to the motion planning data. Through simulation, the feasibility of the recovery action can be evaluated without actually executing the action, avoiding possible dangers. A physics engine or dynamic model can be used to simulate the robot's action and predict potential problems during the recovery process.

[0168] Step S700, if yes, adjusting the current posture of the humanoid robot according to the motion planning data; using a segmented motion strategy to adjust the position and angle of the upper body and / or lower body until the stable posture is restored.

[0169] If the simulation result shows that the stable posture can be restored, the robot's posture is adjusted according to the motion planning data. By adjusting the posture, the robot can be restored to a stable state to avoid injury caused by falling. The joint angle and force can be adjusted by a control algorithm, such as a PID controller, to achieve posture adjustment.

[0170] In addition, a segmented motion strategy is used to adjust the robot's upper body and / or lower body in stages until the stable posture is restored. The segmented motion strategy can more finely control the robot's movements, improving the success rate and safety of recovery. A segmented motion control algorithm can be designed to first adjust the upper body and then the lower body to gradually restore the stable posture.

[0171] For example, if the stable posture can be reached through motion planning data simulation, posture adjustment is first performed, i.e., according to the current posture, the robot needs to first adjust the position of each part of the body to prepare for standing. Then the robot will use a segmented motion strategy, first lifting the upper body to a certain angle, and then the lower body. For example, using the weight of the arms and upper body as a lever, the body is pushed upward by the legs. During the standing process, the robot needs to constantly adjust the center of gravity to maintain dynamic balance.

[0172] A PID controller or other control algorithm is used to fine-tune joint movement to maintain balance.

[0173] Once the robot approaches the standing state, it needs to fine-tune the joint angle and force to finally complete the standing. If the standing attempt fails, the robot needs to be able to re-evaluate the state and decide whether to try again or seek other recovery strategies.

[0174] Step S800, if no, determine that the motion is blocked, stop posture adjustment, and generate a warning.

[0175] If it is determined that the motion is blocked and cannot be restored to the stable posture according to the motion planning data, the posture adjustment is stopped and a warning is generated. The invalid or dangerous movement is stopped in time to avoid further injury and to remind the operator or system to take other measures. The motion can be determined to be blocked by monitoring the robot's movements and sensor feedback, and the movement can be stopped and a warning can be issued if necessary.

[0176] In addition, with reference to Figure 7 The embodiment of the present application also provides a humanoid robot fall control device, comprising:

[0177] A monitoring module 10 is configured to acquire state data of the humanoid robot in real time, wherein the state data comprises a center of gravity position, a motion speed and an acceleration.

[0178] A prediction module 20 is configured to predict a current state of the humanoid robot according to the state data.

[0179] A protection module 30 is configured to control the humanoid robot to perform a fall protection action when the current state is a high-risk fall state.

[0180] In addition, in some embodiments, with reference to Figure 8 The application also provides a humanoid robot fall control device, comprising:

[0181] A sensor module 40 is integrated with various sensors, including but not limited to a gyroscope, an accelerometer, a plantar six-axis force sensor, etc., and is configured to acquire posture, speed, ground contact force and other information of the humanoid robot in real time.

[0182] A data processing module 50 is configured to process the data collected by the sensor module in real time, and provide accurate data basis for the subsequent prediction module through signal filtering and feature value extraction.

[0183] A prediction module 60 is configured to use machine learning algorithms and models to evaluate the fall possibility of the robot in combination with the feature values extracted by the data processing module, and determine the fall direction and the joint that is likely to touch the ground when the fall risk is predicted.

[0184] A control module 70 is configured to adjust the motion state of the robot in real time according to the evaluation result of the prediction module, so as to avoid falling or reduce the damage caused by falling.

[0185] With reference to Figure 9 Based on the humanoid robot fall control device, the fall control of the humanoid robot can be divided into three stages: body stabilization stage, fall protection stage and fall recovery stage.

[0186] (1) Body stabilization stage: Before the fall occurs, the posture detection system and the balance stabilization controller work together to obtain the body posture angle through the IMU, and measure the contact force between the robot and the ground through the plantar six-axis force sensor. Combined with these data, the system calculates the zero moment point (ZMP), and adjusts the waist and hip joint positions of the robot in real time to ensure that the center of gravity of the body remains within the support polygon. If the ZMP exceeds the support polygon due to external interference, the robot will adjust the gait according to the IMU and ZMP data to try to restore the balance. When these adjustment strategies cannot restore the balance, the system will determine the fall direction and start the fall protection function.

[0187] (2) Fall protection phase: Once a fall occurs, the system will perform the following protection actions:

[0188] According to the predicted fall direction, determine which joints will first contact the ground and their force direction.

[0189] The leg knee and hip joints quickly perform a protection action to lower the center of gravity to reduce the impact force when falling.

[0190] Joints with soft dragging function open the soft state before touching the ground to contact the ground in an elastic way, reducing damage to the joints.

[0191] Vulnerable joints such as fingers, wrists, and heads are self-adjusted to a safe posture to avoid direct impact with the ground.

[0192] For joints and parts prone to collision, increase the soft control elasticity factor and add protective materials externally to further reduce the impact force.

[0193] (3) Fall recovery phase: After a fall occurs, the robot determines the body state through IMU data, joint motor position data, and force sensor data, including checking for motor abnormalities and determining the current posture. For different fall postures, such as lying flat, lying down, lying on one side, or other situations, the robot will be processed in a classified manner. Through motion planning, the robot adjusts to a stable posture or recovers to the normal working state to achieve autonomous recovery.

[0194] The application also provides a computer device, which exemplarily comprises a processor and a memory, wherein the memory stores a computer program, and the processor runs the computer program, so that the computer device executes the functions of each module in the above-mentioned humanoid robot fall control method or the above-mentioned humanoid robot fall control device.

[0195] The processor can be an integrated circuit chip with a signal processing capability. The processor can be a general-purpose processor, including a central processing unit (CPU), a graphics processing unit (GPU), and a network processor (NP), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or at least one of the above. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc., which can implement or execute the disclosed methods, steps, and logic block diagrams in the embodiments of the application.

[0196] The memory can be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), and the like. Among them, the memory is used to store a computer program, and the processor can execute the computer program correspondingly after receiving an execution instruction.

[0197] The application further provides a computer storage medium for storing the computer program used in the computer device. The computer storage medium can be a readable storage medium, a non-volatile storage medium or a volatile storage medium. For example, the computer storage medium can include, but is not limited to, a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various program code storage media.

[0198] In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can also be implemented by other manners. The apparatus embodiments described above are only schematic, for example, the flow charts and structural diagrams in the drawings show the possible implementation architectures, functions and operations of the apparatus, method and computer program product according to the embodiments of the present application. In this regard, each block in the flow chart or block diagram can represent a module, a program segment or a part of code, which contains one or more executable instructions for implementing the specified logical function. It should also be noted that, in alternative implementation manners, the functions noted in the blocks can also occur in different orders from that noted in the drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and sometimes they can be executed in reverse order, depending on the functions involved. It should also be noted that each block in the structural diagram and / or flow chart, and the combination of blocks in the structural diagram and / or flow chart, can be implemented by a special hardware-based system for executing the specified functions or actions, or can be implemented by a combination of special hardware and computer instructions.

[0199] In addition, each functional module or unit in each embodiment of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0200] The functions, if implemented in the form of software functional modules and sold or used as independent products, can be stored in a computer readable storage medium. Based on such an understanding, the technical solutions of the present application, in essence or the parts that contribute to the prior art, or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a smart phone, a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present application.

[0201] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application.

Claims

1. A method for controlling the fall of a humanoid robot, characterized in that, include: The state data of the humanoid robot is acquired in real time; the state data includes the center of gravity position, motion speed, and acceleration. Predict the current state of the humanoid robot based on the state data; If the current state is a high-risk fall state, then control the humanoid robot to perform fall protection actions; The fall protection actions include: The process of identifying the direction of the fall and the corresponding key protection area of ​​the robot includes: real-time monitoring of the humanoid robot's attitude angle data using an inertial measurement unit; the attitude angle data includes the absolute values ​​of the elevation angle and / or roll angle; if the attitude angle reaches 0.5 radians, the humanoid robot is determined to be in an unstable state, and the legs and / or vulnerable joints are designated as the key protection area of ​​the robot; if the attitude angle reaches 1.05 radians, the humanoid robot is determined to be in a tilted state, and the impact joints are designated as the key protection area of ​​the robot.

2. The humanoid robot fall control method as described in claim 1, characterized in that, The step of predicting the current state of the humanoid robot based on the state data includes: A model predictive control algorithm is used, with the center of gravity position, the motion speed and the acceleration as inputs, and the robot's whole-body dynamics model is used to determine whether the humanoid robot has reached the preset triggering conditions. If so, the current state of the humanoid robot is determined to be a high-risk fall state; If not, then the current state of the humanoid robot is determined to be not a high-risk fall state.

3. The humanoid robot fall control method as described in claim 2, characterized in that, The preset triggering condition includes at least one of the following conditions: A. The center of gravity position exceeds the preset safety range; B. The change in acceleration exceeds a preset acceleration change threshold; C. The rate of decrease of the motion speed exceeds the preset decrease threshold.

4. The humanoid robot fall control method as described in claim 1, characterized in that, The current key protection areas of the robot include the legs, vulnerable joints, and impact joints; the vulnerable joints include the hand joints and head joints; the impact joints include at least one of the knee joint, hip joint, and elbow joint. The fall protection measures also include: Controlling the legs to be in a bent-leg state lowers the center of gravity of the humanoid robot; Control the fragile joint to adjust to a safe posture; Control the collision joint to activate the compliant collision function; The safe posture includes at least one of the following methods: A. Adjust the hand joints to form a fist; B. Adjust the hand joint, bending the elbow towards the chest area; C. Adjust your head to tilt towards your chest to form a bowed head position.

5. The humanoid robot fall control method as described in claim 4, characterized in that, The compliant collision function includes: The system monitors the external force and actual velocity acting on the collision joint in real time; when the external force is sensed acting on the collision joint, the joint acceleration generated is calculated based on the dynamic model; the expression of the dynamic model is: ； Where F represents the fall impact force; M represents the mass parameter of the impact joint; D represents the damping parameter of the impact joint; v real This represents the actual speed; The rate of change of the joint acceleration of the collision joint; The collision joint is controlled to move in accordance with the external force using the joint acceleration; and the damping parameter and the mass parameter are dynamically adjusted according to the actual velocity of the collision joint.

6. The humanoid robot fall control method as described in claim 1, characterized in that, After controlling the humanoid robot to perform fall protection actions, the method further includes: The humanoid robot is subjected to fault self-check, posture and position detection, and environmental perception, and the self-check results, posture and position information, and environmental perception results are obtained respectively. Based on the self-test results, the posture and position information, and the environmental perception results, motion planning data is established; Determine whether the humanoid robot can simulate and recover to a stable posture based on the motion planning data; If so, the current posture of the humanoid robot is adjusted according to the motion planning data; a segmented motion strategy is used to adjust the position and angle of the upper body and / or lower body until the stable posture is restored. If not, the movement is deemed to be obstructed, posture adjustment is stopped, and a warning is generated.

7. A fall control device for a humanoid robot, characterized in that, include: The monitoring module is used to acquire the status data of the humanoid robot in real time; the status data includes the center of gravity position, movement speed, and acceleration. A prediction module is used to predict the current state of the humanoid robot based on the state data; The protection module is used to control the humanoid robot to perform fall protection actions when the current state is a high-risk fall state; The fall protection actions include: The process of identifying the direction of the fall and the corresponding key protection area of ​​the robot includes: real-time monitoring of the humanoid robot's attitude angle data using an inertial measurement unit; the attitude angle data includes the absolute values ​​of the elevation angle and / or roll angle; if the attitude angle reaches 0.5 radians, the humanoid robot is determined to be in an unstable state, and the legs and / or vulnerable joints are designated as the key protection area of ​​the robot; if the attitude angle reaches 1.05 radians, the humanoid robot is determined to be in a tilted state, and the impact joints are designated as the key protection area of ​​the robot.

8. A computer device, characterized in that, The computer device includes a processor and a memory, the memory storing a computer program, and the processor executing the computer program to implement the humanoid robot fall control method according to any one of claims 1-6.

9. A computer storage medium, characterized in that, It stores a computer program, which, when executed on a processor, implements the humanoid robot fall control method according to any one of claims 1-6.

Images