Robot safety interaction path planning method based on dynamic danger criterion and bidirectional search

Abstract

The invention provides a robot safety interaction path planning method based on a dynamic danger criterion and bidirectional search, and the method comprises the steps: introducing the dynamic danger criterion in a planning stage, carrying out the quantification of a potential collision risk between a robot and a person, and comprehensively considering the inertia and rigidity of the robot, and the influence of the relative distance between the robot and the person; in combination with forward global search and reverse local search, a safe path is quickly generated in a complex environment. The forward search preferentially reduces danger indexes, the reverse search reversely avoids obstacles from a target position, the path continuity is ensured through an attitude potential energy function, and the local convergence problem of a traditional planning algorithm in an obstacle dense area is solved; the system also combines a real-time monitoring and safety control module, fuses a danger criterion, an obstacle potential field and a target potential field, and supports online path optimization in a dynamic environment so as to ensure that the robot can be always in a safe state in a final interaction process.

Description

Technical Field

[0001] The present invention relates to the technical field of human - robot collaborative safety, and particularly to a robot safe interaction path planning method based on dynamic danger criteria and bidirectional search. Background Art

[0002] Currently, in industrial environments, to ensure safety, human - robot interaction is usually achieved by isolating robots from humans. However, as robots gradually enter non - industrial scenarios such as homes, offices, and medical settings, direct human - robot interaction is becoming increasingly common, and simple isolation can no longer meet safety requirements. Traditional path planning methods mostly adopt reactive control strategies based on potential fields, generating motion paths through attractive target and obstacle - avoidance potential fields. However, in complex and crowded environments, these methods are prone to falling into local minima, resulting in extended planning times and discontinuous paths, making it difficult to ensure the safety of robots during interaction. In the prior art, some studies have proposed danger indices or danger criteria to quantify the impact of factors such as the relative distance between the robot and the human, robot inertia, and stiffness on the collision risk; but these methods usually only target the end - effector or single factors and fail to comprehensively consider the interaction risk between the overall robot configuration and the human. In addition, traditional planning is mostly forward search, making it difficult to quickly find a safe path when the target is in a densely - populated obstacle area. Summary of the Invention

[0003] To solve the existing technical problems, the main object of the present invention is to provide a robot safe interaction path planning method based on dynamic danger criteria and bidirectional search. This method comprehensively evaluates the robot inertia and the relative distance between the robot and the human during the planning stage to generate an index reflecting the actual danger; at the same time, when the target is in a crowded environment, inverse kinematics is used to generate the target configuration, and reverse search is implemented to reduce the planning time and risk, and the forward and reverse paths are seamlessly connected through an attitude potential energy function to ensure the continuity of the overall path in the robot joint space. Through this improved method, potential collision risks can be effectively reduced, and a safer and more reliable motion path can be provided for the robot in a dynamic human - robot interaction environment.

[0004] To achieve the above - mentioned technical features, the object of the present invention is realized as follows: A robot safe interaction path planning method based on dynamic danger criteria and bidirectional search uses dynamic danger criteria to quantify the interaction risk between the robot and the human, comprehensively considering robot inertia, stiffness, and the relative distance between the robot and the human to evaluate the collision impact force and danger; when the target is in a densely - populated obstacle area, inverse kinematics is used to generate the target configuration, and a low - potential energy path is searched backward from the target, and then seamlessly connected with the path obtained by forward search from the starting point through an additional attitude potential energy function to ensure the continuity and reasonableness of the path in the robot joint space.

[0005] Preferably, the method specifically includes the following steps: Step 1, task initialization: Receive the user's natural language command and convert it into the target position, task requirements, and constraint conditions, and obtain the current initial configuration of the robot; Step 2, quantification of dynamic danger criteria: Calculate the dynamic danger index according to the inertia, stiffness of the robot in each configuration, and the relative distance between the robot and the human, and form a danger criterion, and the danger criterion adopts an expression based on summation or multiplication; Step 3, reverse search stage: When the target position is in a densely populated obstacle area, calculate the target configuration through inverse kinematics according to the target workspace position and the desired end effector direction; Step 4, global forward search stage: Starting from the initial configuration, use the forward search method to generate a preliminary path. During this process, evaluate each configuration based on the dynamic danger criterion, and preferentially select a low-danger state; if the danger criterion reaches the preset low safety threshold during the current search process, mark this path segment as an acceptable state; Step 5, path merging and attitude correction: Use an additional attitude potential function to perform seamless and smooth connection between the path obtained by forward search and the path obtained by reverse search, ensuring that the two parts of the path are continuous and have the same attitude in the robot joint space; Step 6, real-time monitoring and safety control: Integrate a real-time monitoring module to continuously monitor the user feedback and environmental changes; Step 7, output the final safe path comparison and analysis: Output the final safe path after forward and reverse search fusion, attitude correction, and real-time monitoring adjustment to the robot control system for the robot to perform specific tasks.

[0006] Preferably, in Step 1, receive and parse the command input by the user in natural language form, convert the command into information including the target position, action requirements, and related task constraints; at the same time, obtain the current initial configuration of the robot to provide basic data for subsequent path planning, safety assessment, and real-time control.

[0007] Preferably, in Step 2, calculate the inertia of the robot in each configuration, evaluate the relative distance between the robot and the human, and quantify the distance factor through a repulsive potential function. Combine the above factors to generate a danger index to evaluate the potential collision impact force and risk level in the current configuration, and use this danger index as the key input in the subsequent path planning stage to facilitate the selection of a low-danger configuration and provide a basis for the bidirectional search strategy.

[0008] Preferably, the types of danger criteria include: 1) Summation type criterion: ; Among them, is an independent factor, representing the risk index of the independent contribution of inertia and distance terms in the additive criterion through linear superposition; is the current effective inertia of the robot; is the total mass of the robot; is the distance between the centroid of the robot and the human; is the inertia factor, is the centroid distance factor, with weight ; is the centroid distance repulsive force function; 2) Product type criterion: ; Among them, is the dependence factor, representing the risk index contributed by the product coupling relationship between inertia and the human-robot distance; is the scaling factor to ensure that the risk index range is within [0, 1]; is the preset maximum safe inertia threshold; is the maximum distance when the risk criterion takes effect, and there is no risk contribution beyond this distance, is the minimum distance when the risk criterion takes effect.

[0009] Preferably, in step 3, reverse search for the low-potential energy path from the target configuration to determine the intermediate position for safe transition, providing a basis for the seamless connection of the forward search.

[0010] Preferably, in step 4, generate a preliminary path from the initial configuration of the robot using forward search, with key points that meet the task requirements as the segmentation boundary points; in each search step, calculate the dynamic risk criterion based on the inertia of the robot and the relative centroid distance between the robot and the human, and preferentially select low-risk configurations; divide the search process into multiple segments to quickly generate a continuous and safety-constrained global path, providing a basis for subsequent local trajectory planning.

[0011] Preferably, in step 5, use the attitude potential energy function to seamlessly merge the forward and reverse paths and perform attitude correction to ensure continuous matching of the robot path in the joint space. The potential energy of the attitude function: ; Among them, is the attitude potential energy function, is the joint angle in the search configuration, is the joint angle obtained from the end configuration of the reverse path to ensure that the attitude of the end point of the forward path is consistent with the starting point of the reverse path.

[0012] Preferably, in step 6, the environment and user feedback are monitored in real time, the security is dynamically evaluated, and the planned path is automatically adjusted to ensure that the robot is in a low-risk state.

[0013] Preferably, in step 7, the final path that has been optimized for security and corrected for continuity is output to the robot control system for task execution.

[0014] The present invention has the following beneficial effects: Aiming at the security problems of local minima, path discontinuity and low computational efficiency existing in traditional path planning methods in a complex and crowded human-robot interaction environment, the present invention proposes a robot safety interaction path planning method based on dynamic danger criteria and bidirectional search, which specifically includes the following effects: 1. Improve the safety of robot interaction: Quantify the interaction risk between the robot and humans based on dynamic danger criteria, optimize path planning, make the robot stay away from high-risk areas as much as possible, and reduce the collision risk.

[0015] 2. Efficient bidirectional search strategy: Combine forward search and backward search methods, and achieve seamless connection of the path through the attitude potential function, effectively reduce the local minimum problem, and improve the path planning efficiency.

[0016] 3. Real-time monitoring and dynamic adjustment: Integrate a real-time monitoring module, which can adjust the path according to user feedback and environmental changes to ensure the adaptability and safety of the robot in complex scenarios.

[0017] 4. Optimize the motion posture of the robot: Consider factors such as the inertia and stiffness of the robot during path planning, ensure that the robot always maintains a low-risk configuration during movement, reduce the need for emergency obstacle avoidance, and improve the stability of task execution.

[0018] 5. Improve the task execution efficiency: Adopt bidirectional search combined with dynamic danger criteria, make the path search more accurate, reduce unnecessary detours, and improve the movement efficiency and task completion rate of the robot in complex environments.

[0019] 6. The present invention provides a comprehensive path planning solution for complex and crowded human-robot interaction environments, and has broad application prospects. Especially in scenarios that require high safety and efficiency, such as the manufacturing industry, medical robots and other fields, it may be widely used. Description of the Drawings

[0020] The present invention will be further described below with reference to the drawings and embodiments.

[0021] Figure 1 Flowchart of the combined backward-forward search algorithm of the present invention. Detailed Embodiments

[0022] To make the objectives, innovative aspects, application fields, and operation methods of the present invention clearer and more transparent, the present invention will be further described in detail below in conjunction with embodiments and additional flowcharts.

[0023] Aiming at the problems of local minima, discontinuous paths, and low computational efficiency existing in the traditional path planning method of robots in the existing human-robot coexistence scenario, which result in the robot being in a high-risk state during the interaction with the collaborator. The present invention proposes a robot safe interaction path planning method based on dynamic danger criteria and bidirectional search, and the specific implementation method is as follows: Embodiment 1: A robot safe interaction path planning method based on dynamic danger criteria and bidirectional search quantifies the interaction risk between the robot and the human using dynamic danger criteria, comprehensively considering the inertia and stiffness of the robot and the relative distance between the robot and the human (especially between the centers of mass) to evaluate the collision impact force and danger; when the target is located in a dense obstacle area, the inverse kinematics is used to generate the target configuration, and the low-potential energy path is searched backward from the target, and then seamlessly connected with the path obtained by the forward search from the starting point through an additional attitude potential energy function to ensure that the path is continuous and reasonable in the robot joint space. This bidirectional search method can more effectively find the low-risk path to the target and significantly reduce the search time; the system integrates a real-time monitoring module, which can evaluate the user's feedback on the robot's actions and automatically adjust the planned path when detecting environmental changes or potential dangers to ensure that the robot is always in a safe state.

[0024] Embodiment 2: A robot safe interaction path planning method based on dynamic danger criteria and bidirectional search specifically includes the following steps: Step 1: Task initialization, receive the user's natural language command, convert it into the target position, task requirements, and constraint conditions, and obtain the current initial configuration of the robot.

[0025] Step 2: Quantification of dynamic danger criteria. According to the inertia and stiffness of the robot in each configuration and the relative distance between the robot and the human (especially the center of mass of the human), calculate the dynamic danger index to form the danger criteria (which can adopt the expression based on summation or multiplication).

[0026] Step 3: Reverse search stage. When the target position is in a dense obstacle area, calculate the target configuration through inverse kinematics according to the target workspace position and the desired end effector direction.

[0027] Step 4: Global forward search phase. Starting from the initial configuration, a preliminary path is generated using the forward search method. During this process, each configuration is evaluated based on the dynamic danger criterion, and the low-danger states are preferentially selected. If, during the current search process, the danger criterion reaches the preset low safety threshold, then this path segment is marked as an acceptable state.

[0028] Step 5: Path merging and attitude correction. Using an additional attitude potential energy function, seamless and smooth connection is carried out between the path obtained from the forward search and the path obtained from the backward search to ensure that the two parts of the path are continuous and have the same attitude in the robot joint space.

[0029] Step 6: Real-time monitoring and safety control. Integrate a real-time monitoring module to continuously monitor user feedback and environmental changes.

[0030] Step 7: Output and comparative analysis of the final safe path. Output the final safe path that has been fused by forward and backward search, attitude corrected, and adjusted by real-time monitoring to the robot control system for the robot to execute specific tasks.

[0031] Furthermore, in Step 1, receive and parse the command input by the user in natural language form, convert the command into information including the target position, action requirements, and related task constraints, etc. At the same time, obtain the current initial configuration of the robot to provide basic data for subsequent path planning, safety assessment, and real-time control. This step ensures that the system starts from the user's needs and the actual environmental state, laying the initial conditions for subsequent dynamic danger assessment and bidirectional search.

[0032] Furthermore, in Step 2, calculate the inertia of the robot in each configuration (considering the mass distribution and total mass of each link of the robot), evaluate the relative distance between the robot and the human, and quantify the distance factor through a repulsive potential function. Combine the above factors to generate a danger index for evaluating the potential collision impact force and risk level in the current configuration. This danger index serves as a key input in the subsequent path planning stage, helping to select low-danger configurations and providing a basis for the bidirectional search strategy. The danger criterion DC (Danger Criterion), whose purpose is to quantify the potential risk of the robot's motion trajectory to human-robot interaction safety through a mathematical model, thereby dynamically optimizing the safety of the path in path planning. Select the type of danger criterion: 1) Additive criterion (independent factor): ; Among them, is the independent factor, representing the danger index independently contributed by the linear superposition of the inertia and distance terms in the additive criterion; is the centroid distance repulsive force function; is the current effective inertia of the robot; is the total mass of the robot; is the centroid distance between the robot and the human; is the inertia factor, is the centroid distance factor and weight ; 2) Product - type criterion (dependency factor): ; Among them, is the dependency factor, representing the risk index contributed by the product coupling relationship between inertia and the human - robot distance; is the scaling factor to ensure that the risk index ranges from [0, 1]; is the preset maximum safe inertia threshold; is the centroid distance between the robot and the human; is the maximum distance when the risk criterion takes effect, and there is no risk contribution beyond this distance, is the minimum distance when the risk criterion takes effect.

[0033] In step 3, it is characterized in that a low - potential - energy path is searched reversely from the target configuration to determine the intermediate position for safe transition ( IL ), providing a basis for the seamless connection of the forward search.

[0034] Furthermore, in step 4, a preliminary path is generated by forward search starting from the initial configuration of the robot. This path uses key points that meet the task requirements (such as grasping and releasing points) as segment boundary points; in each search step, a dynamic risk criterion is calculated based on the robot's inertia and the relative centroid distance between the robot and the human, and a low - risk configuration is preferentially selected; the search process is divided into multiple segments to quickly generate a continuous and safety - constraint - compliant global path, providing a basis for subsequent local trajectory planning.

[0035] Furthermore, in step 5, the forward and reverse paths are seamlessly merged using the attitude potential energy function, and attitude correction is performed to ensure continuous matching of the robot path in the joint space. Attitude function potential energy: ; Among them, is the attitude potential energy function, is the joint angle in the search configuration, is the joint angle obtained from the end - configuration of the reverse path to ensure that the attitude of the end - point of the forward path is the same as that of the start - point of the reverse path.

[0036] Furthermore, in step 6, the environment and user feedback are monitored in real - time, the safety is dynamically evaluated, and the planned path is automatically adjusted to ensure that the robot is in a low - risk state.

[0037] Further, in step 7, the final path after safety optimization and continuity correction is output to the robot control system for task execution.

[0038] Embodiment 2: Refer to Figure 1 , the present invention proposes a robot safety interaction path planning method and system based on dynamic danger criteria and bidirectional search. The specific implementation method is as follows: 1. Set the danger area for robot-human interaction based on information factors such as the relative position, speed, and posture between the robot and the human, and input the environment and robot model. Input the robot dynamic parameters: mass m , inertia tensor I ; human-robot interaction parameters: human centroid position , safety distance threshold ; obstacle information: position, geometric shape, and potential field influence radius .

[0039] 2. By real-time monitoring of information such as the distance, relative speed, and posture angle between the robot and the human, use a mathematical model (such as a potential energy function) to quantify the interaction risk and make dynamic adjustments according to the risk level. Danger criterion selection and formula definition: Additive criterion (applicable to robots with low degrees of freedom): ; Wherein, is the inertia factor, is the centroid distance factor, is the inertia scalar, is the centroid distance repulsive force function.

[0040] Product criterion (applicable to redundant robots): ; Wherein, is the scaling factor to ensure that the danger index range is in [0,1].

[0041] 3. Weight and threshold setting, is the inertia factor, is the centroid distance factor, is the weight of the target attraction criterion, is the weight of the danger criterion. Initial weight values in stage 1: ; Initial weight values in stage 2: . Danger threshold: , and when it exceeds the limit, it triggers path adjustment.

[0042] 4. Perform the reverse generation of the target configuration in the bidirectional search path planning algorithm. Through inverse kinematics solution, when the target is in the dense obstacle area, use inverse kinematics to calculate the joint configuration set of the end effector target pose , and screen the danger criterion DC The configuration with the lowest danger .

[0043] 5. Reverse path search, starting from Reverse search, avoiding the obstacle potential field , generate a path segment to the intermediate safety point IL (the obstacle potential field is zero). The termination condition is that the path segment extends to the free space or reaches the maximum search steps (such as 1000 steps).

[0044] 6. Next, perform forward path search and phased optimization, which are divided into two stages: Stage 1 (mainly minimizing danger), search forward from the starting configuration to IL , and the cost function is: ; Among them, is the target attraction potential field function, is the obstacle repulsion potential field function. Continuously optimize until the danger criterion , and the attitude potential energy .

[0045] Stage 2 (mainly target-oriented), switch the cost function weights: ; Among them, is the target attraction potential field function, is the obstacle repulsion potential field function. Combine the local trajectory planner to generate a smooth trajectory and adjust the speed in real time (such as reducing the speed by 30% according to the user intention feedback).

[0046] 7. Connect the reverse path and the forward path. Perform path connection and attitude correction, and the attitude potential energy function is: ; Force the joint angles of the forward path end point and the reverse path start point to be the same, avoiding configuration jumps. Path merging, generate a continuous trajectory through cubic spline interpolation at IL .

[0047] 8. Real-time monitoring and dynamic adjustment, first estimate the user intention. Through RGB-DThe camera is integrated with a force sensor to detect the user's gestures (such as waving to pause) and contact force in real time, and generate an intention confidence level. If , trigger speed decay: .

[0048] 9. Dynamically perceive the environment, use SLAM technology to update the position of obstacles. If a dynamic obstacle is detected to enter the range, recalculate the potential field .

[0049] 10. When a change in the environment (such as the appearance of a new obstacle) or user feedback indicates a potential risk, control the safety during the human-robot collaboration process and re-plan the path. Deviation trigger: When the danger criterion or obstacle mutation occurs, initiate an emergency stop and generate a temporary safe path. Recovery mechanism: Call the local planner to generate an obstacle avoidance trajectory within 100 ms to ensure a response delay of 200 ms.

[0050] 11. Final path and feedback output. After completing the forward and backward searches, attitude correction, and real-time adjustment, the system will generate an optimized safe path and output it to the robot control system for task execution. During the task execution process, the system will continuously monitor the planning effect and safety assessment results, and transmit this feedback information back to the planning module. According to the actual situation, the system can update the planning parameters and strategies in a timely manner to form a closed-loop feedback, further optimizing the subsequent path planning and robot motion control to ensure that the robot always maintains the best safety state in a dynamic environment.

[0051] So far, the present invention has described the technical method of the present invention in detail through the drawings and specific flowcharts. However, the protection scope of the present invention is not limited to the specific embodiments described. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and these modified or substituted technical solutions still fall within the protection scope of the present invention.

Claims

1. A robot safety interaction path planning method based on dynamic danger criteria and bidirectional search, characterized in that: Quantify the interaction risk between the robot and humans using dynamic danger criteria, comprehensively considering the robot's inertia, stiffness, and the relative distance between the robot and humans to evaluate the collision impact force and danger level; when the target is in a dense obstacle area, use inverse kinematics to generate the target configuration, search backward from the target for a low-potential energy path, and then seamlessly connect it with the path obtained by forward search from the starting point through an additional attitude potential energy function to ensure that the path is continuous and reasonable in the robot joint space.

2. The robot safety interaction path planning method based on the dynamic danger criterion and bidirectional search according to claim 1, wherein: The method specifically includes the following steps: Step 1, task initialization: Receive the user's natural language command and convert it into the target position, task requirements, and constraint conditions to obtain the current initial configuration of the robot. Step 2, quantification of dynamic danger criteria: Calculate the dynamic danger index according to the inertia, stiffness of the robot in each configuration, and the relative distance between the robot and humans to form a danger criterion, and the danger criterion adopts an expression based on summation or multiplication. Step 3, reverse search stage: When the target position is in a dense obstacle area, calculate the target configuration through inverse kinematics according to the target workspace position and the desired end effector direction. Step 4, global forward search stage: Starting from the initial configuration, use the forward search method to generate a preliminary path. During this process, evaluate each configuration based on the dynamic danger criteria and preferentially select low-danger states; if during the current search process, the danger criterion reaches the preset low safety threshold, mark this path segment as an acceptable state. Step 5, path merging and attitude correction: Use an additional attitude potential energy function to seamlessly and smoothly connect the path obtained by forward search and the path obtained by reverse search to ensure that the two parts of the path are continuous and have the same attitude in the robot joint space. Step 6, real-time monitoring and safety control: Integrate a real-time monitoring module to continuously monitor the user feedback and environmental changes. Step 7, output and final safety path comparison and analysis: Output the final safety path after forward and reverse search fusion, attitude correction, and real-time monitoring adjustment to the robot control system for the robot to execute specific tasks.

3. The method for planning a safe interaction path of a robot based on a dynamic danger criterion and bidirectional search according to claim 2, wherein: In Step 1, receive and parse the command input by the user in natural language form, convert the command into information including the target position, action requirements, and relevant task constraints; at the same time, obtain the current initial configuration of the robot to provide basic data for subsequent path planning, safety assessment, and real-time control.

4. The robot safety interaction path planning method based on dynamic danger criterion and bidirectional search according to claim 2, wherein: In Step 2, calculate the inertia of the robot in each configuration, evaluate the relative distance between the robot and humans, and quantify the distance factor through a repulsive potential energy function. Combine the above factors to generate a danger index to evaluate the potential collision impact force and risk level in the current configuration, and use this danger index as the key input in the subsequent path planning stage to facilitate the selection of low-danger configurations and provide a basis for the two-way search strategy.

5. The robot safety interaction path planning method based on dynamic danger criterion and bidirectional search according to claim 4, characterized in that, The types of danger criteria include: 1) Summation type criterion: ； Among them, is an independent factor, representing the hazard index independently contributed by the inertia and distance terms in the additive criterion through linear superposition; is the current effective inertia of the robot; is the total mass of the robot; is the centroid distance between the robot and the human; is the inertia factor, is the centroid distance factor, with a weight ; is the centroid distance repulsive force function; 2) Product type criterion: ； Among them, is the dependence factor, representing the risk index contributed by the product coupling relationship between inertia and the human-machine distance; is the scaling factor to ensure that the risk index ranges from [0, 1]; is the preset maximum safe inertia threshold; is the maximum distance when the risk criterion takes effect. Beyond this distance, there is no risk contribution. is the minimum distance when the risk criterion takes effect.

6. The robot safety interaction path planning method based on dynamic danger criteria and bidirectional search according to claim 4, wherein, In Step 3, search backward from the target configuration for a low-potential energy path to determine the intermediate position for safe transition, providing a basis for the seamless connection of forward search.

7. The robot safety interaction path planning method based on dynamic danger criteria and bidirectional search according to claim 1, wherein In step 4, a preliminary path is generated from the initial configuration of the robot using forward search, and this path uses the key points that meet the task requirements as the segmentation boundary points; in each search step, a dynamic danger criterion is calculated based on the inertia of the robot and the relative centroid distance between the robot and the human, and low-danger configurations are preferentially selected; the search process is divided into multiple segments to quickly generate a continuous global path that meets safety constraints, providing a basis for subsequent local trajectory planning.

8. The robot safety interaction path planning method based on dynamic danger criteria and bidirectional search according to claim 1, wherein In step 5, the forward and backward paths are seamlessly merged using the attitude potential energy function, and attitude correction is performed to ensure that the robot path is continuously matched in the joint space. The potential energy of the attitude function: ； Among them, is the attitude potential energy function, is the joint angle in the search configuration, is the joint angle obtained from the end configuration of the reverse path, ensuring that the attitude of the end of the forward path is the same as that of the start of the reverse path.

9. The robot safety interaction path planning method based on dynamic danger criteria and bidirectional search according to claim 1, wherein, In step 6, the environment and user feedback are monitored in real time, the safety is dynamically evaluated, and the planned path is automatically adjusted to ensure that the robot is in a low-risk state.

10. The robot safety interaction path planning method based on dynamic danger criteria and bidirectional search according to claim 1, characterized in that, In step 7, the final path that has undergone safety optimization and continuity correction is output to the robot control system for task execution.

Images