Behavior safety control method for humanoid robot and intelligent robot with humanoid robot body

Abstract

The invention provides a behavior safety control method for a humanoid robot and an intelligent robot with a body, and relates to the technical field of robots, and the method comprises the steps: through a mode of combining global obstacle avoidance and local obstacle avoidance, the path planning efficiency, flexibility and safety of the robot and the cooperative capability of a multi-robot system are remarkably improved. According to the method, the behavior safety control level of the robot can be improved, the performance and response speed of the whole system can be optimized, and it is ensured that the robot can operate efficiently and safely when facing static and dynamic obstacles. Meanwhile, the behavior mode of the robot is judged, if abnormal behaviors including hardware and software faults, running path abnormity, special environment exceeding a set range and the like are found, emergency braking is needed, and all power supplies of the robot are cut off through a built-in safety chip of the robot or a special instruction.

Description

Technical Field

[0001] The present invention relates to the field of robotics technology, and in particular to a behavior safety control method for a humanoid robot and an embodied intelligent robot. Background Art

[0002] In robotic handling tasks in warehouses, despite continuous technological advances, there are still some safety control issues. Warehouse environments are usually dynamic, and workers, other robots, or transportation equipment may appear at any time, causing robots to constantly handle and respond to these changes. Especially in highly complex environments, robots need to instantly identify and predict potential obstacles or people, and perform path planning and obstacle avoidance operations. In addition, current robot obstacle avoidance algorithms and systems react slowly or make inaccurate judgments when dealing with complex situations. For example, when multiple obstacles or people appear in front of the robot at the same time, it may lead to improper obstacle avoidance path selection, or the robot fails to respond in time, resulting in collisions or safety accidents.

[0003] The above problems not only affect the robot's task execution efficiency, but also directly affect the safety of workers, especially when interacting with the robot at close range. Therefore, the low safety of robot behavior control has become an urgent problem to be solved. Summary of the invention

[0004] The present invention provides a behavior safety control method for a humanoid robot and an embodied intelligent robot, so as to solve the defect of low behavior control safety of robots in the prior art and improve the behavior control safety of robots.

[0005] The present invention provides a behavior safety control method for a humanoid robot and an embodied intelligent robot, comprising the following steps:

[0006] Receive shared information sent by multiple robots in the surrounding environment, and collect environmental information corresponding to the target robot;

[0007] Establishing an environmental model based on the shared information and the environmental information to generate an environmental map;

[0008] Determining a global path according to the environment map and the starting point information of the target robot;

[0009] During the movement based on the global path, if a dynamic obstacle is detected, local obstacle avoidance is performed to avoid the dynamic obstacle;

[0010] If there is abnormal behavior during driving, all power supplies to the target robot are disconnected based on the built-in safety chip of the target robot or special instructions.

[0011] According to a behavior safety control method for a humanoid robot and an embodied intelligent robot provided by the present invention, the environmental model is established according to the shared information and the environmental information to generate an environmental map, including:

[0012] Determining the state information of the target robot at the last moment according to the shared information;

[0013] According to the state transfer matrix, control input and control matrix of the target robot, and the state information of the target robot at the previous moment, the state information of the target robot at the current moment is predicted; wherein the state transfer matrix is ​​used to describe the evolution of the robot state from one moment to the next moment when there is no external control and noise influence; the control input includes the control instruction of the robot; the control matrix is ​​used to determine the state change of the robot caused by the control input;

[0014] Performing data fusion on the current state information and the environmental information to obtain fused data;

[0015] An environmental model is established according to the fused data to generate the environmental map.

[0016] According to a behavior safety control method for a humanoid robot and an embodied intelligent robot provided by the present invention, determining a global path according to the environment map and the starting point information of the target robot comprises:

[0017] Constructing a map data structure according to the environment map;

[0018] Initializing an open list for storing nodes to be explored and a closed list for storing explored nodes, and setting an initial actual cost, an estimated cost, and a comprehensive evaluation cost for each node in the map data structure;

[0019] Determine the starting point and the target point of the target robot according to the starting point information, and add the estimated cost and the comprehensive evaluation cost of the starting point to the open list;

[0020] Entering a path search loop, in each loop, selecting a node with the smallest comprehensive evaluation cost from the open list for processing, exploring neighbor nodes through multiple loops, updating the actual cost, estimated cost and comprehensive evaluation cost of the node, and adding the qualified node to the open list or the closed list;

[0021] After the target point is found, a path list is constructed by tracing back from the target point to the predecessor node, and the path list is reversed to obtain a global path from the starting point to the target point.

[0022] According to a behavior safety control method for a humanoid robot and an embodied intelligent robot provided by the present invention, in the process of moving based on the global path, if a dynamic obstacle is detected, local obstacle avoidance is performed to avoid the dynamic obstacle, comprising:

[0023] During the movement based on the global path, real-time acquisition of perception data of the surrounding environment;

[0024] If the dynamic obstacle is identified based on the perception data, obtaining a motion feature of the dynamic obstacle;

[0025] Determining an obstacle avoidance strategy for the local obstacle avoidance according to the motion characteristics;

[0026] The dynamic obstacle is avoided according to the obstacle avoidance strategy.

[0027] According to a behavior safety control method for a humanoid robot and an embodied intelligent robot provided by the present invention, the method further comprises:

[0028] Receive obstacle avoidance control instructions;

[0029] The private key is used to decrypt the session key encrypted by the public key to obtain the session key;

[0030] decrypting the obstacle avoidance control instruction according to the session key;

[0031] According to the decrypted obstacle avoidance control instructions, the obstacle avoidance operation is performed.

[0032] According to a behavior safety control method for a humanoid robot and an embodied intelligent robot provided by the present invention, the method further includes:

[0033] In case of a fault being detected, a task transfer request signal is broadcast to surrounding robots;

[0034] If a response signal to the task transfer request signal is received, the task to be executed is transferred to the robot that sent the response signal.

[0035] The present invention also provides a behavior safety control device for a humanoid robot and an embodied intelligent robot, comprising the following modules:

[0036] A receiving module is used to receive shared information sent by multiple robots in the surrounding environment and collect environmental information corresponding to the target robot;

[0037] An environment map generation module, used to establish an environment model according to the shared information and the environment information, and generate an environment map;

[0038] A global path determination module, used to determine a global path according to the environment map and the starting point information of the target robot;

[0039] An obstacle avoidance module, configured to perform local obstacle avoidance if a dynamic obstacle is detected during movement based on the global path, so as to avoid the dynamic obstacle;

[0040] The abnormal behavior processing module is used to disconnect all power supplies of the target robot based on the built-in safety chip of the target robot or special instructions if abnormal behavior occurs during driving.

[0041] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, a behavioral safety control method for a humanoid robot and an embodied intelligent robot as described above is implemented.

[0042] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a behavioral safety control method for a humanoid robot and an embodied intelligent robot as described in any one of the above.

[0043] The present invention also provides a computer program product, comprising a computer program, which, when executed by a processor, implements the behavior safety control method of any of the humanoid robots and embodied intelligent robots described above.

[0044] The behavioral safety control method of the humanoid robot and the embodied intelligent robot provided by the present invention receives shared information sent by multiple robots in the surrounding environment and collects the environmental information corresponding to the target robot; establishes an environmental model based on the shared information and environmental information to generate an environmental map; determines the global path based on the environmental map and the starting point information of the target robot; in the process of moving based on the global path, if a dynamic obstacle is detected, local obstacle avoidance is performed to avoid the dynamic obstacle. The present invention significantly improves the path planning efficiency, flexibility, safety and coordination ability of the robot through the combination of global obstacle avoidance and local obstacle avoidance. This method can not only improve the behavioral safety control level of the robot, but also optimize the performance and response speed of the overall system, ensuring that the robot can operate efficiently and safely when facing static and dynamic obstacles. At the same time, by judging the behavior of the robot, if abnormal behavior is found, including hardware and software failures, abnormal operation path, special environment exceeding the set range, etc., emergency braking is required, and all power supplies of the robot are disconnected through the robot's built-in safety chip or special instructions. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 It is a flow chart of the behavior safety control method of the humanoid robot and the embodied intelligent robot provided by the present invention.

[0047] Figure 2 It is a schematic diagram of the framework of the behavior safety control of the humanoid robot and the embodied intelligent robot provided by the present invention.

[0048] Figure 3 It is a structural schematic diagram of the behavior safety control device of the humanoid robot and the embodied intelligent robot provided by the present invention.

[0049] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] Combine the following Figure 1-Figure 4 The behavioral safety control method of the humanoid robot and the embodied intelligent robot of the present invention is described.

[0052] Figure 1 : is a flow chart of the behavior safety control method of the humanoid robot and the embodied intelligent robot provided by the present invention, such as Figure 1 As shown, the method includes the following:

[0053] Step 101, receiving shared information sent by multiple robots in the surrounding environment, and collecting environmental information corresponding to the target robot.

[0054] It should be noted that the robot of the present invention can be a humanoid robot and an embodied intelligent robot.

[0055] In a multi-robot collaboration scenario, robots can share information, such as their relative position with other robots, perceived environmental data (such as obstacles, dynamic objects, weather information, etc.), mission information, target location, collected sensor data, constructed environmental maps, etc. By sharing information, robots can work together to improve mission efficiency, avoid conflicts, optimize paths and emergency responses, etc.

[0056] The target robot can be any robot in the multi-robot collaborative scene. The target robot receives shared information sent by multiple robots in the surrounding environment, and collects environmental information in its own area, for example, through sensors such as visual sensors, lidar, ultrasonic sensors, infrared sensors, tactile sensors, temperature sensors, and environmental sensors, to collect multi-dimensional perception data such as vision, sound, gas, temperature, humidity, position, and motion.

[0057] For example, when the laser radar is working, it emits a laser beam into the surrounding space. The laser is reflected back after encountering an object and is received by the laser radar. By measuring the time difference between laser emission and reception, the distance r between the laser radar and the object is calculated based on the principle of constant light speed. i At the same time, the laser radar will record the emission angle θ of the laser beam during the process of emitting and receiving the laser beam. i The data obtained by LiDAR measurement are in polar coordinates (r i ,θ i ), i = 1, 2, ..., n means there are n measurement points. The distance r of each measurement point i and angle θ i Combined, they accurately describe the position of the point in the lidar coordinate system.

[0058] Step 102: Establish an environment model based on the shared information and the environment information to generate an environment map.

[0059] Specifically, the shared information and environmental information are fused, and then an environmental model is established based on the fused data to generate an environmental map. The environmental map is a multi-dimensional, dynamic information collection that shows the structure, characteristics, and properties of the environment in which the robot or device is located.

[0060] In one embodiment, the state information of the target robot at the previous moment is determined based on the shared information; the state information of the target robot at the current moment is predicted based on the state transfer matrix, control input and control matrix of the target robot, as well as the state information of the target robot at the previous moment; wherein the state transfer matrix is ​​used to describe the evolution of the robot state from one moment to the next moment when there is no external control and noise influence; the control input includes the control instructions of the robot; the control matrix is ​​used to determine the state change of the robot caused by the control input; the state information and environmental information at the current moment are fused to obtain fused data; an environmental model is established based on the fused data to generate an environmental map.

[0061] Specifically, the timestamp in the shared information is extracted, and the state information of the target robot at the last moment is obtained from the shared information based on the timestamp. It is understandable that the state information of the target robot at the last moment may include the position coordinates (such as x, y, z axis coordinates) of the target robot in space, speed (linear velocity and angular velocity), posture (such as heading angle, pitch angle, roll angle) and other parameters, which are used to describe the motion state of the robot at the last moment and the position relationship in the environment.

[0062] Furthermore, the state information of the target robot at the current moment is predicted, wherein the state estimation prediction formula is:

[0063]

[0064] in, Represents the state prediction value, that is, the state information of the target robot at the current moment, represents the optimal state estimate of the target robot at the previous moment, that is, the state information of the target robot at the previous moment; A represents the state transfer matrix, which is used to describe the evolution of the robot state from one moment to the next when there is no external control and noise influence; u k represents the control input, including the control instructions of the robot, such as the speed and acceleration instructions of the robot; B represents the control matrix, which is used to determine the state change of the robot caused by the control input.

[0065] Assume the robot's state vector X = [x, y, θ] T , respectively represent the abscissa, ordinate and orientation angle of the robot in the plane coordinate system; the robot's control input vector u=[v,ω] T , where v represents the linear velocity and ω represents the angular velocity. At the discrete time step Δt, the state transfer matrix A and the control matrix B can be derived according to the kinematic principle.

[0066] The kinematic equation is:

[0067]

[0068] Writing this in matrix form:

[0069]

[0070] Among them, the state transfer matrix is:

[0071]

[0072] The control matrix is:

[0073]

[0074] It can be seen that the elements in the matrix are related to the robot's motion parameters and posture.

[0075] For example, if the position and speed of the target robot at the previous moment are known, the position and speed of the target robot at the current moment can be predicted by combining the current speed command and its own motion law (state transfer matrix).

[0076] Furthermore, the predicted state information at the current moment is updated to achieve the fusion between the state information at the current moment and the environmental information. The state update formula is:

[0077]

[0078] Among them, z k represents new measurement data, i.e., collected environmental information; H represents the observation matrix; K k represents the gain. Specifically, The state prediction value Transform to the same dimensional space as the measured data so that it can be combined with the new measured data z k Compare the difference between the two Reflects the deviation between the state prediction value and the actual measurement. Gain K k It is determined by the covariance of system noise and measurement noise, which determines how to balance the state prediction value and measurement deviation If the measurement noise is small, K k will be larger, which means that the new measurement data is more trusted, thus the state prediction value On the contrary, if the measurement noise is large, K k Will be smaller and more likely to rely on state predictions In this way, the new measurement data z k and state prediction value Fusion to obtain more accurate state estimation

[0079] Furthermore, based on the fused data, the robot can construct different types of environment maps, where the map types may include grid maps and feature maps.

[0080] For example, taking a grid map as an example, the environment is divided into a two-dimensional grid of M×N, and each grid cell (i, j) has a probability value p ij , indicating the possibility that the location is an obstacle. The larger the probability value, the greater the possibility of an obstacle. The map is represented by P = {p ij |i∈[1,M],j∈[1,M]}. The construction process is as follows: Based on the fused data, the target robot infers the state of each grid (obstacle or blank area) and updates the corresponding probability value. For example, when the sensor detects an obstacle in a certain area, the obstacle probability value p of the grid corresponding to the area is ij will increase.

[0081] Taking the feature map as an example, the map is constructed by extracting significant features in the environment (such as walls, door frames, trees, etc.). The feature points are represented by F = {f 1 ,f 2 ,...,f k}, where f k Including location, type, size and other information. The construction process is: identify the features in the environment based on the fused data, and determine the relationship between feature points to generate a map. For example, determine the location and length of the wall based on sensor data and add it as a feature point to the feature map.

[0082] Optionally, as the robot explores the environment, newly acquired measurements z k The map needs to be updated in real time. For example, using the Bayesian update formula:

[0083]

[0084] Among them, P(M k |z k ) is in the measurement data z k Given the environment map M k The posterior probability, P(z k |M k ) is a given environment map M k The measurement likelihood function under k ) is the prior probability of the environmental map (i.e. the previous map information). Through the above formula, new data is integrated into the map to continuously improve and update the environmental model.

[0085] The embodiment of the present invention can fully utilize the advantages of different data sources by fusing the shared information of other robots with the environmental information collected by itself, thereby generating a more accurate environmental map and improving the reliability of the map. On the other hand, the data fusion method can reduce the dependence on sensors and avoid the failure of the entire system due to the failure or performance degradation of a sensor.

[0086] Step 103, determining a global path according to the environment map and the starting point information of the target robot.

[0087] Global path planning can be understood as planning a path for the robot from the starting point to the target point based on the entire environment map. In this case, the goal is to determine the best path that takes into account obstacles, terrain, map restrictions, etc., and tries to make the robot travel the shortest or most efficient path.

[0088] In one embodiment, a map data structure is constructed based on an environment map; an open list for storing nodes to be explored and a closed list for storing explored nodes are initialized, and an initial actual cost, estimated cost, and comprehensive evaluation cost are set for each node in the map data structure; the starting point and target point of the target robot are determined based on the starting point information, and the estimated cost and comprehensive evaluation cost of the starting point are added to the open list; a path search loop is entered, and in each loop, a node with the smallest comprehensive evaluation cost is selected from the open list for processing, and after multiple cycles of exploring neighboring nodes, the actual cost, estimated cost, and comprehensive evaluation cost of the node are updated, and eligible nodes are added to the open list or closed list; after finding the target point, a path list is constructed by backtracking from the target point to the predecessor node, and the path list is reversed to obtain a global path from the starting point to the target point.

[0089] Specifically, assume that the environment map is a two-dimensional grid map, and use M to represent the map. ij represents the grid at row i and column j in the map. ij =1, indicating that the grid is a passable area; if M ij = 0, indicating that the grid is an obstacle. At the same time, let the starting point coordinates of the robot be (x s ,y s ), the target point coordinates are (x g ,y g ).

[0090] Furthermore, two lists are created: the open list OL, which is used to store the nodes to be explored; and the closed list CL, which is used to store the nodes that have been explored. For each node (grid), an initial cost is set. The cost refers to the cumulative cost or path length from the starting point to a certain node, which can include:

[0091] g(n): the actual cost from the starting point to node n;

[0092] h(n): estimated cost from node n to the target point (heuristic estimate), Manhattan distance or Euclidean distance can be used as the heuristic function;

[0093] f(n) is the comprehensive evaluation cost of node n, f(n) = g(n) + h(n).

[0094] Further, calculate the starting point (x s ,y s ) of h((x s ,y s )) and f((x s ,y s )), add the starting point and its related information (g, h, f values) to the open list.

[0095] Further, entering the main loop (path search), when the open list OL is not empty, perform the following steps:

[0096] 1) Select a node: Select the node n with the smallest f(n) value from the open list OL, remove it from the open list OL and add it to the closed list CL.

[0097] 2) Target arrival judgment: If node n is the target point (x g ,y g ), it means that a path has been found. By backtracking the predecessor node of node n (the predecessor node of each node can be recorded additionally when recording node information), starting from the target point, the predecessor nodes are added to the path list one by one until returning to the starting point, and the global path is obtained.

[0098] 3) Explore neighbor nodes: If node n is not the target point, explore the neighbor nodes of node n. Assume that the robot can move in four directions: up, down, left, and right. Node n = (x n ,y n )’s neighbor nodes are (x n +1,y n )、(x n -1,y n )、(x n ,y n +1)、(x n ,y n For each neighbor node m=(x m ,y m ):

[0099] 3.1) Legality check: Check whether the neighbor node m is within the map range, and (i.e., not an obstacle). If the condition is not met, skip this neighbor node.

[0100] 3.2) Update cost: Calculate the new g value from the starting point through node n to neighbor node m, denoted as:

[0101] g'(m) = g(n) + cost(n, m);

[0102] where cost(n, m) is the cost of moving from node n to node m. In the grid map, if the movement cost between adjacent nodes is the same, it is set to 1. If g'(m) < g(m), then update g'(m) = g(m) for neighbor node m, recalculate h(m) according to the heuristic function, and f(m) = g(m) + h(m). At the same time, set node n as the predecessor node of node m.

[0103] 3.4) Add to the open list: If neighbor node m is not in the open list, add it and its related information (updated g, h, f values) to the open list.

[0104] Furthermore, perform path backtracking and construction: Assume the target point is found. When recording the information of each node, in addition to the g, h, f values, the predecessor node of each node is also recorded. Starting from the target point, continuously search for the predecessor node to construct a path list. Let the current node be cur = (x cur , y cur ), its predecessor node be pre = (x pre , y pre ), and the path list be Path. Initially, cur = (x g , y g ), add cur to Path. Then, cur = pre, repeat this process until cur = (x s , y s ). Finally, reverse the path list Path to obtain the global path from the starting point to the target point.

[0105] Based on the known environmental map, the embodiment of the present invention determines a reasonable path through the starting position and target position of the target robot using a path planning algorithm, which can ensure that the robot avoids obstacles and prevent collisions with static or dynamic obstacles in the environment, thereby ensuring the safe driving of the robot.

[0106] Step 104, during the movement based on the global path, if a dynamic obstacle is detected, perform local obstacle avoidance to avoid the dynamic obstacle.

[0107] Specifically, the target robot first determines a route based on global path planning, and during the process of moving, it continuously monitors the surrounding environment through real-time perception data. When a dynamic obstacle is detected, the target robot predicts its future position based on the movement characteristics of the obstacle, and selects an appropriate local obstacle avoidance strategy (such as avoidance, deceleration, re-planning the path, etc.) based on the prediction results. Finally, the target robot safely avoids the dynamic obstacle according to the obstacle avoidance strategy and continues to move along the path.

[0108] In one embodiment, during the movement based on the global path, perception data of the surrounding environment is acquired in real time; if a dynamic obstacle is identified based on the perception data, the motion characteristics of the dynamic obstacle are acquired; based on the motion characteristics, an obstacle avoidance strategy for local obstacle avoidance is determined; and the dynamic obstacle is avoided according to the obstacle avoidance strategy.

[0109] When the target robot moves based on the global path, it obtains perception data in real time through the sensors it is equipped with, which may include lidar data, visual data, ultrasonic sensor data, sound sensor data, etc.

[0110] Dynamic obstacle detection mainly relies on the position change of objects in a continuous time series. Dynamic obstacles can be identified by using motion models, image analysis, obstacle separation based on motion estimation, and other methods. For example, taking motion models as an example, the movement of objects can be detected by comparing consecutive frames. For example, in the point cloud data scanned by the lidar, some point clouds will move over time, and the changing point clouds belong to dynamic obstacles. By calculating the changes in the position of objects at different time points, the movement trajectory and speed of the object can be estimated, thereby accurately identifying dynamic obstacles. Taking image analysis as an example, by analyzing the movement of pixels between consecutive image frames, the movement direction and speed of the object can be estimated, and dynamic objects and backgrounds can be further distinguished. Deep learning is used to detect objects in the image, identify dynamic objects, and combine tracking algorithms to identify the movement trajectory of objects.

[0111] Furthermore, based on the perception data, the motion characteristics of the dynamic obstacles are obtained, which may include the obstacle position, moving speed, moving trajectory, moving direction, acceleration, and movement mode. Then, according to the motion characteristics, the obstacle avoidance strategy for local obstacle avoidance is determined, and the obstacle avoidance strategy may include an obstacle avoidance strategy based on trajectory planning, an obstacle avoidance strategy based on potential field method, an obstacle avoidance strategy based on behavior, an obstacle avoidance strategy based on artificial potential field, an obstacle avoidance strategy based on artificial intelligence and machine learning, and the like.

[0112] Taking the obstacle avoidance strategy of artificial potential field as an example, the embodiment of the present invention adopts multiple potential fields of different levels to handle different types of targets and obstacles respectively. The basic idea of ​​multi-layer potential field is that multiple different potential fields can handle targets and obstacles in a hierarchical manner, thereby avoiding the local minimum problem and taking into account the avoidance of global targets and local obstacles at the same time.

[0113] Assume there are N targets, M obstacles, and the robot is located at position p. and obstacles Each layer will generate its own attractive potential field and repulsive potential field. Multi-layer potential fields can combine the influence of each layer of potential fields through weighted sum.

[0114] Multi-target attraction potential field:

[0115]

[0116] Among them, i is an index variable used to traverse each target from 1 to N, that is, to represent different target individuals. It represents the attraction coefficient of the i-th target, which is used to measure the attraction strength of the i-th target to the robot. Different targets can have different attraction coefficients. Represents the position vector of the i-th target, that is, the specific position of the i-th target in space.

[0117] Multiple obstacle repulsive potential field:

[0118]

[0119] Among them, j is an index variable used to traverse each obstacle from 1 to M, that is, to represent different obstacle individuals. Represents the repulsion coefficient of the j-th obstacle, which is used to measure the strength of the repulsion force of the j-th obstacle on the robot. Different obstacles can have different repulsion coefficients. Represents the position vector of the jth obstacle, that is, the specific position of the jth obstacle in space. threshold It is a distance threshold, which is used to define that when the distance between the robot and the obstacle is less than the threshold, the corresponding repulsive potential energy will be generated. When the distance is greater than this threshold, the repulsive potential energy is 0 or can be ignored.

[0120] Composite potential field: The final potential field is a weighted sum of the target attraction and obstacle repulsion:

[0121] U(p)=α.U att (p)+β.U rep (p);

[0122] Among them, α and β represent weight factors, which are used to adjust the balance between attraction and repulsion.

[0123] For example, suppose there are multiple target points (such as target 1 and target 2) and some obstacles (such as obstacle 1 and obstacle 2) in the environment. Each target and each obstacle will generate their own attraction and repulsion. Assume that the robot starts from the starting point (0, 0), calculates the attraction potential field of each target and the repulsion potential field of each obstacle, synthesizes these potential fields to obtain the total potential field; calculates the gradient of the robot's current position, and moves along the gradient direction. For example, the robot will first avoid obstacle 1 and obstacle 2. At the same time, the robot will be attracted by target 1 and target 2, and finally find a suitable balance path between the two.

[0124] Through the multi-layer potential field model, the impact of the target and obstacles can be considered and synthesized separately, thus achieving more flexible and accurate path planning. In practical applications, adjusting the weights and the form of the potential field can cope with complex environments and multiple targets, avoiding obstacles while optimizing the path to reach the target.

[0125] Furthermore, the dynamic potential field of machine learning can also be combined. For example, through reinforcement learning algorithms, the robot can gradually learn how to adjust the attraction and repulsion forces in the interaction with the environment to avoid local minima and reach the goal efficiently. A reward function R(p) is defined to guide the robot's learning.

[0126] Reward function R(p):

[0127] R(p)=-U att (p)-U rep (p);

[0128] Through the reinforcement learning algorithm, the parameters of the potential field are updated according to the robot's experience in the environment to adapt to different obstacle configurations and target changes. The multi-layer potential field can handle multiple targets and obstacles and avoid local minimum problems; combined with machine learning methods, the potential field can be optimized according to dynamic changes in the environment.

[0129] Furthermore, in order to prevent the robot from being subjected to excessive repulsive force when approaching an obstacle, a smooth repulsive potential field function can be used to avoid overreaction. A Gaussian function can be used to smooth the repulsive force to avoid the sharp changes in the traditional potential field method.

[0130] Smooth repulsive potential field:

[0131]

[0132] Among them, σ represents the smoothing factor of the repulsive force, which controls the expansion range of the potential field.

[0133] Step 105: If there is any abnormal behavior during the driving process, all power supplies of the target robot are disconnected based on the built-in safety chip of the target robot or special instructions.

[0134] Specifically, a variety of methods are used to determine whether the robot's behavior is normal. At the hardware and software levels, the performance parameters of the hardware (such as CPU usage, sensor signal strength, etc.) and the running status of the software (whether there is error information, whether the program is executed according to the predetermined process, etc.) are continuously monitored. For the running path, the robot will compare the actual trajectory with the preset path. Once the deviation exceeds the threshold, it will be identified as an abnormality. In special environments, with the help of various environmental sensors (temperature sensors, humidity sensors, pressure sensors, etc.), when the detected environmental data exceeds the set range, an abnormal judgment is triggered.

[0135] Once abnormal behavior is detected, emergency braking is required to prevent the robot from causing more serious consequences. There are two ways to achieve this. One is through the built-in safety chip of the machine. As the core component to ensure the safe operation of the robot, the safety chip can quickly execute the pre-set program when receiving an abnormal signal, disconnect all power supplies to the robot, and stop the robot immediately. The second is through special instructions, which can be input by the operator at the external control terminal, or automatically triggered by the robot's own control system in certain specific circumstances. After the special instruction is issued, all power supplies to the robot will also be cut off to achieve the purpose of emergency braking.

[0136] In one embodiment, it is assumed that there is a humanoid robot responsible for cargo handling in a logistics warehouse. The robot needs to shuttle between preset shelf aisles and carry cargo from designated shelves to the shipping area. During operation, the robot's visual sensor suddenly fails, causing it to be unable to accurately identify the shelf position. This causes the robot's running path to deviate from the preset route and gradually approach the wall of the warehouse. At the same time, due to the failure of the visual system, the robot's software system constantly attempts to recalibrate the sensor, resulting in a sharp increase in CPU usage and software jamming. At this time, the robot's built-in monitoring program detects hardware failure (abnormal visual sensor), software failure (CPU usage is too high and software jams), and abnormal running path (deviation from the preset path). After the monitoring program determines that the robot has abnormal behavior, it immediately sends a signal to the built-in security chip. After receiving the signal, the security chip quickly cuts off all power supplies to the robot. Alternatively, the warehouse manager finds that the robot is abnormal through the monitoring system and enters a special command in the control terminal, which can also achieve the same effect of cutting off the power supply to the robot. Finally, the robot stops running before approaching the wall, avoiding serious consequences such as collision with the wall and damage to itself and falling of goods.

[0137] The behavior safety control method of the humanoid robot and the embodied intelligent robot provided by the embodiment of the present invention receives the shared information sent by multiple robots in the surrounding environment, and collects the environmental information corresponding to the target robot; establishes an environmental model according to the shared information and the environmental information, and generates an environmental map; determines the global path according to the environmental map and the starting point information of the target robot; in the process of moving based on the global path, if a dynamic obstacle is detected, local obstacle avoidance is performed to avoid the dynamic obstacle. The present invention can significantly improve the path planning efficiency, flexibility, safety and coordination ability of the robot, especially in a dynamic and complex environment, by combining global obstacle avoidance with local obstacle avoidance. This method can not only improve the behavior safety control level of the robot, but also optimize the performance and response speed of the overall system, ensuring that the robot can operate efficiently and safely when facing static and dynamic obstacles. At the same time, by judging the behavior of the robot, if abnormal behavior is found, including hardware and software failures, abnormal operation path, special environment exceeding the set range, etc., emergency braking is required, and all power supplies of the robot are disconnected through the robot's built-in safety chip or special instructions.

[0138] Based on the above embodiment, the method further includes:

[0139] Step 110, receiving an obstacle avoidance control instruction;

[0140] Step 111, using the private key to decrypt the session key encrypted by the public key to obtain the session key;

[0141] Step 112, decrypting the obstacle avoidance control instruction according to the session key;

[0142] Step 113: Execute an obstacle avoidance operation according to the decrypted obstacle avoidance control instruction.

[0143] Specifically, the robot receives obstacle avoidance control instructions from an external control system (such as a remote control terminal, the robot's own environmental perception and decision-making system, etc.). The instructions are encrypted before transmission to ensure the security of the transmission process and prevent the instructions from being stolen or tampered with, which would affect the normal operation of the robot.

[0144] During the communication process, the sender (controller) generates a temporary session key, which is used to encrypt the obstacle avoidance control instructions in the future. It is a symmetric encryption key. To ensure the secure transmission of the session key, the sender encrypts the session key with the public key of the receiver (robot) and sends it. After receiving the encrypted session key, the robot decrypts it with its own private key. The public key and private key are generated based on an asymmetric encryption algorithm. The public key can be made public, and the private key is kept securely by the robot. Only the robot's private key can decrypt the content encrypted with its public key, thereby securely obtaining the session key.

[0145] After obtaining the session key, the robot uses the symmetric key to decrypt the received obstacle avoidance control command. The symmetric encryption algorithm uses the same key for encryption and decryption, which has high computational efficiency. The sender encrypts the obstacle avoidance control command with the session key, and the robot can restore the original command content with the same session key.

[0146] After the robot completes command decryption and extracts specific operation information such as obstacle avoidance actions and path planning, it converts this information into motor drive signals, servo control signals, etc., controls the movement of the robot's mechanical structure, avoids obstacles, and ensures the robot's safe operation in complex environments.

[0147] The embodiment of the present invention adopts hybrid encryption, combining the advantages of symmetric encryption and asymmetric encryption, to ensure the security, efficiency and flexibility of the transmission and processing of robot obstacle avoidance control instructions. On the other hand, for resource-constrained robot devices, the high efficiency of symmetric encryption can reduce the computing burden; at the same time, asymmetric encryption ensures the security of session keys, allowing robots to communicate reliably in unsafe network environments and adapt to complex and changing application scenarios.

[0148] In one embodiment, the method further comprises:

[0149] Step 120, in the event of a fault being detected, broadcasting a task transfer request signal to surrounding robots;

[0150] Step 121: if a response signal to the task transfer request signal is received, the task to be executed is transferred to the robot that sent the response signal.

[0151] Specifically, each robot has a self-monitoring system that monitors the robot's various states (such as battery power, sensor status, motion module health, etc.) in real time. When the robot detects a failure in a key system (such as a power system, sensor, control system, etc.), the robot will determine whether it has entered a fault state. Among them, the fault confirmation process may include: the robot first executes a self-diagnosis program to confirm whether it is a fault. For example, if the battery power is too low, the robot will first check whether it can be charged or complete the task. If the self-diagnosis confirms that there is a fault, the robot will evaluate the fault level. For example, a minor fault may be resolved by restarting, and a major fault requires other robots to take over.

[0152] Once the fault is confirmed, the robot broadcasts a task transfer request signal to surrounding robots through wireless communication protocols (such as Wi-Fi, Bluetooth, ZigBee, 5G, etc.), which may include: robot identifier (ID), current task information (task type, location, progress, etc.), fault type and urgency.

[0153] Broadcast signals can be limited to a certain range (e.g., within 50 meters) to ensure that requests are only sent to other robots that are close to the current robot. Communication between robots can use a dedicated collaborative communication protocol, such as broadcast communication based on UDP (User Datagram Protocol), to ensure that signals are transmitted quickly and efficiently.

[0154] After receiving the task transfer request signal, other robots can decide whether to take over the task based on the following factors:

[0155] 1) Task compatibility: Check whether the current task is suitable for its own functions (for example, whether it has the same load capacity, the same path planning requirements, etc.).

[0156] 2) Distance and location: Select the robot closest to the faulty robot to reduce the delay in task takeover.

[0157] 3) Task priority and idle status: According to the priority of the task, a robot that is currently idle and has sufficient resources is selected to take over the task. If there are not enough idle robots, the task request can be forwarded to a robot farther away.

[0158] 4) Load and power: The robot should also consider its remaining power, load capacity, etc. to ensure that it can successfully complete the task after taking over.

[0159] After selecting a suitable robot, it will reply to the faulty robot, expressing its willingness to take over the task.

[0160] The robot that takes over the task needs to obtain the latest status of the current task (such as progress, path, target position, etc.). To this end, a task data synchronization protocol can be designed to transmit the task status data to the control system of the taking over robot through wireless communication.

[0161] After taking over the task, the robot will perform path planning and optimize the path from the current point to the task target point to ensure the successful completion of the task. During the execution process, the taking-over robot will continue to monitor the surrounding environment in real time, avoid collisions, and respond to possible new faults in a timely manner.

[0162] When the faulty robot completes self-diagnosis and recovers, it can resume work through self-repair or restart process. The robot will send a recovery signal to other robots to notify them that they can return to the task.

[0163] The embodiment of the present invention ensures that the robot can effectively request other robots to take over the task when a failure occurs and ensures the smooth completion of the task through the steps of fault detection, task request broadcasting, task transfer and takeover, task synchronization, recovery and regression. The system should have real-time, robust and adaptive capabilities, be able to handle various emergencies, and ensure efficient collaboration of the robot team.

[0164] In order to further analyze and explain the behavior safety control methods of humanoid robots and embodied intelligent robots, refer to Figure 2 And the following examples.

[0165] The framework of behavioral safety control of humanoid robots and embodied intelligent robots mainly includes the following contents:

[0166] External network interface: connected to the external network, used for the robot to receive external instructions, data and other information. It is the entrance for the robot to communicate with the outside world.

[0167] Security chip: It is at the core and plays a key role in security protection. It receives information from the external network interface and performs secure encryption processing on it. It is also responsible for key management and special command processing. The security chip is also connected to the power supply and the core control circuit switch, and can control the power supply when necessary. For example, when the robot has a safety problem, the security chip controls the core control circuit switch to cut off the power supply.

[0168] Behavior control module: connected to the security chip, receives information processed by the security chip, and is used to control the robot's specific behavior actions, such as movement, grasping, etc.

[0169] AI chip: connected to the security chip and CPU, responsible for processing artificial intelligence-related tasks, such as image recognition, speech recognition, etc., to provide support for the robot's intelligent decision-making.

[0170] CPU: As the central processing unit, it is connected to the security chip and AI chip, processes various data and instructions, and coordinates the operation of various parts of the robot.

[0171] GPU: Responsible for graphics processing and accelerated computing tasks in the robot system, and can assist AI chips and CPUs in performing complex calculations.

[0172] Power supply: Provides power support for the entire system, and is connected to the security chip through the core control circuit switch and is controlled by the security chip.

[0173] In the above framework provided by the embodiment of the present invention, the security chip is responsible for encryption, key management and special commands, and centrally controls the information received by the robot from the external network to ensure data security; the security chip can control the core control circuit switch, and can cut off the power supply in time when there is a problem to avoid abnormal behavior of the robot; through encrypted communication with the security chip, it is ensured that the instructions received by the behavior control module are true and correct, and the instructions are prevented from being tampered with; each module is connected to the security chip, the data transmission is secure, and the safety of the behavior control module is guaranteed when cooperating with other components.

[0174] The behavior safety control device of the humanoid robot and the embodied intelligent robot provided by the present invention is described below. The behavior safety control device of the humanoid robot and the embodied intelligent robot described below and the behavior safety control method of the humanoid robot and the embodied intelligent robot described above can be referenced to each other.

[0175] refer to Figure 3 The behavior safety control device for a humanoid robot and an embodied intelligent robot provided by the present invention includes a receiving module 301, an environment map generating module 302, a global path determining module 303 and an obstacle avoiding module 304.

[0176] The receiving module 301 is used to receive shared information sent by multiple robots in the surrounding environment and collect environmental information corresponding to the target robot;

[0177] An environment map generation module 302, used to establish an environment model according to the shared information and the environment information, and generate an environment map;

[0178] A global path determination module 303, used to determine a global path according to the environment map and the starting point information of the target robot;

[0179] The obstacle avoidance module 304 is used to perform local obstacle avoidance if a dynamic obstacle is detected during the movement based on the global path to avoid the dynamic obstacle;

[0180] The abnormal behavior processing module 305 is used to disconnect all power supplies of the target robot based on the built-in safety chip of the target robot or special instructions if there is abnormal behavior during driving.

[0181] The behavior safety control device of the humanoid robot and the embodied intelligent robot provided by the embodiment of the present invention receives the shared information sent by multiple robots in the surrounding environment, and collects the environmental information corresponding to the target robot; establishes an environmental model according to the shared information and the environmental information, and generates an environmental map; determines the global path according to the environmental map and the starting point information of the target robot; in the process of moving based on the global path, if a dynamic obstacle is detected, local obstacle avoidance is performed to avoid the dynamic obstacle. The present invention can significantly improve the path planning efficiency, flexibility, safety and coordination ability of the robot, especially in a dynamic and complex environment, by combining global obstacle avoidance with local obstacle avoidance. This method can not only improve the behavior safety control level of the robot, but also optimize the performance and response speed of the overall system, ensuring that the robot can operate efficiently and safely when facing static and dynamic obstacles. At the same time, by judging the behavior of the robot, if abnormal behavior is found, including hardware and software failures, abnormal operation path, special environment exceeding the set range, etc., emergency braking is required, and all power supplies of the robot are disconnected through the robot's built-in safety chip or special instructions.

[0182] In one embodiment, the environment map generation module 302 is specifically configured to:

[0183] Determining the state information of the target robot at the last moment according to the shared information;

[0184] According to the state transfer matrix, control input and control matrix of the target robot, and the state information of the target robot at the previous moment, the state information of the target robot at the current moment is predicted; wherein the state transfer matrix is ​​used to describe the evolution of the robot state from one moment to the next moment when there is no external control and noise influence; the control input includes the control instruction of the robot; the control matrix is ​​used to determine the state change of the robot caused by the control input;

[0185] Performing data fusion on the current state information and the environmental information to obtain fused data;

[0186] An environmental model is established according to the fused data to generate the environmental map.

[0187] In one embodiment, the global path determination module 303 is specifically configured to:

[0188] Constructing a map data structure according to the environment map;

[0189] Initializing an open list for storing nodes to be explored and a closed list for storing explored nodes, and setting an initial actual cost, an estimated cost, and a comprehensive evaluation cost for each node in the map data structure;

[0190] Determine the starting point and the target point of the target robot according to the starting point information, and add the estimated cost and the comprehensive evaluation cost of the starting point to the open list;

[0191] Entering a path search loop, in each loop, selecting a node with the smallest comprehensive evaluation cost from the open list for processing, exploring neighbor nodes through multiple loops, updating the actual cost, estimated cost and comprehensive evaluation cost of the node, and adding the qualified node to the open list or the closed list;

[0192] After the target point is found, a path list is constructed by tracing back from the target point to the predecessor node, and the path list is reversed to obtain a global path from the starting point to the target point.

[0193] In one embodiment, the obstacle avoidance module 304 is specifically configured to:

[0194] During the movement based on the global path, real-time acquisition of perception data of the surrounding environment;

[0195] If the dynamic obstacle is identified based on the perception data, obtaining a motion feature of the dynamic obstacle;

[0196] Determining an obstacle avoidance strategy for the local obstacle avoidance according to the motion characteristics;

[0197] The dynamic obstacle is avoided according to the obstacle avoidance strategy.

[0198] In one embodiment, the obstacle avoidance module 304 is further configured to:

[0199] Receive obstacle avoidance control instructions;

[0200] The private key is used to decrypt the session key encrypted by the public key to obtain the session key;

[0201] decrypting the obstacle avoidance control instruction according to the session key;

[0202] According to the decrypted obstacle avoidance control instructions, the obstacle avoidance operation is performed.

[0203] In one embodiment, the behavior safety control device for a humanoid robot and an embodied intelligent robot further includes a task transfer module for:

[0204] In case of a fault being detected, a task transfer request signal is broadcast to surrounding robots;

[0205] If a response signal to the task transfer request signal is received, the task to be executed is transferred to the robot that sent the response signal.

[0206] Figure 4An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430 and a communication bus 440, wherein the processor 410, the communication interface 420 and the memory 430 communicate with each other through the communication bus 440. The processor 410 may call the logic instructions in the memory 430 to execute the behavior safety control method of the humanoid robot and the embodied intelligent robot, the method comprising: receiving shared information sent by multiple robots in the surrounding environment, and collecting the corresponding environmental information of the target robot; establishing an environmental model according to the shared information and the environmental information, and generating an environmental map; determining a global path according to the environmental map and the starting point information of the target robot; in the process of moving based on the global path, if a dynamic obstacle is detected, local obstacle avoidance is performed to avoid the dynamic obstacle; if there is abnormal behavior during driving, all power supplies of the target robot are disconnected based on the built-in safety chip of the target robot or special instructions.

[0207] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0208] On the other hand, the present invention also provides a computer program product, which includes a computer program, and the computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the behavioral safety control method of a humanoid robot and an embodied intelligent robot provided by the above methods, and the method includes: receiving shared information sent by multiple robots in the surrounding environment, and collecting environmental information corresponding to the target robot; establishing an environmental model based on the shared information and the environmental information, and generating an environmental map; determining a global path based on the environmental map and the starting point information of the target robot; in the process of moving based on the global path, if a dynamic obstacle is detected, local obstacle avoidance is performed to avoid the dynamic obstacle; if there is abnormal behavior during driving, all power supplies to the target robot are disconnected based on the built-in safety chip of the target robot or special instructions.

[0209] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the behavioral safety control method for a humanoid robot and an embodied intelligent robot provided by the above-mentioned methods, the method comprising: receiving shared information sent by multiple robots in the surrounding environment, and collecting environmental information corresponding to a target robot; establishing an environmental model based on the shared information and the environmental information, and generating an environmental map; determining a global path based on the environmental map and the starting point information of the target robot; during movement based on the global path, if a dynamic obstacle is detected, performing local obstacle avoidance to avoid the dynamic obstacle; if there is abnormal behavior during driving, disconnecting all power supplies to the target robot based on the built-in safety chip of the target robot or special instructions.

[0210] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0211] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0212] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A behavior safety control method for a humanoid robot and an embodied intelligent robot, characterized in that: include: Receive shared information sent by multiple robots in the surrounding environment, and collect environmental information corresponding to the target robot; Establishing an environmental model based on the shared information and the environmental information to generate an environmental map; Determining a global path according to the environment map and the starting point information of the target robot; During the movement based on the global path, if a dynamic obstacle is detected, local obstacle avoidance is performed to avoid the dynamic obstacle; If there is abnormal behavior during driving, all power supplies to the target robot are disconnected based on the built-in safety chip of the target robot or special instructions.

2. The behavior safety control method of a humanoid robot and an embodied intelligent robot according to claim 1, characterized in that: The step of establishing an environment model according to the shared information and the environment information to generate an environment map includes: Determining the state information of the target robot at the last moment according to the shared information; According to the state transfer matrix, control input and control matrix of the target robot, and the state information of the target robot at the previous moment, the state information of the target robot at the current moment is predicted; wherein the state transfer matrix is ​​used to describe the evolution of the robot state from one moment to the next moment when there is no external control and noise influence; the control input includes the control instruction of the robot; the control matrix is ​​used to determine the state change of the robot caused by the control input; Performing data fusion on the current state information and the environmental information to obtain fused data; An environmental model is established according to the fused data to generate the environmental map.

3. The behavior safety control method of a humanoid robot and an embodied intelligent robot according to claim 1, characterized in that: Determining a global path according to the environment map and the starting point information of the target robot includes: Constructing a map data structure according to the environment map; Initializing an open list for storing nodes to be explored and a closed list for storing explored nodes, and setting an initial actual cost, an estimated cost, and a comprehensive evaluation cost for each node in the map data structure; Determine the starting point and the target point of the target robot according to the starting point information, and add the estimated cost and the comprehensive evaluation cost of the starting point to the open list; Entering a path search loop, in each loop, selecting a node with the smallest comprehensive evaluation cost from the open list for processing, exploring neighbor nodes through multiple loops, updating the actual cost, estimated cost and comprehensive evaluation cost of the node, and adding the qualified node to the open list or the closed list; After the target point is found, a path list is constructed by tracing back from the target point to the predecessor node, and the path list is reversed to obtain a global path from the starting point to the target point.

4. The behavior safety control method of a humanoid robot and an embodied intelligent robot according to claim 1, characterized in that: In the process of moving based on the global path, if a dynamic obstacle is detected, local obstacle avoidance is performed to avoid the dynamic obstacle, including: During the movement based on the global path, real-time acquisition of perception data of the surrounding environment; If the dynamic obstacle is identified based on the perception data, obtaining a motion feature of the dynamic obstacle; Determining an obstacle avoidance strategy for the local obstacle avoidance according to the motion characteristics; The dynamic obstacle is avoided according to the obstacle avoidance strategy.

5. The behavior safety control method of a humanoid robot and an embodied intelligent robot according to claim 4, characterized in that: The method further comprises: Receive obstacle avoidance control instructions; The private key is used to decrypt the session key encrypted by the public key to obtain the session key; decrypting the obstacle avoidance control instruction according to the session key; According to the decrypted obstacle avoidance control instructions, the obstacle avoidance operation is performed.

6. The behavior safety control method of a humanoid robot and an embodied intelligent robot according to claim 1, characterized in that: The method further comprises: In case of a fault being detected, a task transfer request signal is broadcast to surrounding robots; If a response signal to the task transfer request signal is received, the task to be executed is transferred to the robot that sent the response signal.

7. A behavior safety control device for a humanoid robot and an embodied intelligent robot, characterized in that: include: A receiving module is used to receive shared information sent by multiple robots in the surrounding environment and collect environmental information corresponding to the target robot; An environment map generation module, used to establish an environment model according to the shared information and the environment information, and generate an environment map; A global path determination module, used to determine a global path according to the environment map and the starting point information of the target robot; An obstacle avoidance module, configured to perform local obstacle avoidance if a dynamic obstacle is detected during movement based on the global path, so as to avoid the dynamic obstacle; The abnormal behavior processing module is used to disconnect all power supplies of the target robot based on the built-in safety chip of the target robot or special instructions if abnormal behavior occurs during driving.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, it implements the behavior safety control method of the humanoid robot and the embodied intelligent robot as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the behavior safety control method of the humanoid robot and the embodied intelligent robot as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the behavior safety control method of the humanoid robot and the embodied intelligent robot as described in any one of claims 1 to 6 is implemented.

Images