Robot anti-collision control method, device, equipment, storage medium and product

Abstract

Embodiments of the present application provide a robot anti-collision control method, device, equipment, storage medium and product, and relate to the field of robot control. The method comprises: obtaining a preset car washing path and distance data between the robot and a car to be washed, to obtain a basic attractive force vector between the robot and a target surface of the car to be washed; based on the basic attractive force vector and the distance data, a combined repulsive force field vector is generated in combination with a preset constraint priority weight, and the constraint priority weight comprises an environmental constraint weight, a self-collision constraint weight and a cooperation constraint weight; the combined repulsive force field vector is subjected to tangential projection filtering to generate a safety force vector; the safety force vector is transposed to obtain a joint speed increment, and based on the joint speed increment, a control instruction of the robot is generated to enable the robot to perform real-time obstacle avoidance. The method of the present application significantly improves the obstacle avoidance capability of a dual-arm car washing robot in a complex dynamic scene through a multi-level artificial potential field model and a tangential projection filtering algorithm.

Description

Technical Field

[0001] This application relates to the field of robot control, and in particular to a robot collision avoidance control method, device, equipment, storage medium and product. Background Technology

[0002] In modern automotive maintenance services, dual-arm robots need to perform cleaning actions simultaneously within a confined space. During the cleaning process, the car body may experience slight displacement due to vibration or external disturbances. At the same time, the car wash operation requires extremely high response speed to the dual-arm robot's trajectory adjustments, and any obstacle avoidance maneuvers must prioritize ensuring the safety of the car body.

[0003] Existing technologies have significant shortcomings in cleaning path planning for complex vehicle body structures. Using offline fixed trajectories prevents real-time trajectory adjustments, leading to collisions or deadlocks of the double arms in the centerline area. Existing methods that treat multiple constraints equally lack a priority handling mechanism, easily triggering secondary collision risks and causing more serious safety accidents. Furthermore, existing methods may cause collisions with the vehicle body due to improper directional control when avoiding the double arms, resulting in damage to the customer's vehicle.

[0004] Therefore, existing dual-arm robots lack a multi-constraint priority processing mechanism in confined spaces, making it difficult to decouple the coupled constraints of environmental safety, self-collision avoidance, and dual-arm cooperative avoidance, which can easily lead to secondary collision risks. Summary of the Invention

[0005] This application provides a robot collision avoidance control method, device, equipment, storage medium, and product to solve the technical problem that existing dual-arm robots lack a multi-constraint priority processing mechanism in confined spaces, which makes it difficult to decouple the coupled constraints of environmental safety, self-collision avoidance, and dual-arm cooperative avoidance, and easily leads to secondary collision risks.

[0006] In a first aspect, this application provides a robot collision avoidance control method, comprising:

[0007] Obtain the preset car wash path and the distance data between the robot and the car to be cleaned, and obtain the basic gravitational vector between the robot and the target surface of the car to be cleaned.

[0008] Based on the basic gravity vector and distance data, and combined with the preset constraint priority weights, a synthetic repulsive field vector is generated. The constraint priority weights include environmental constraint weights, self-collision constraint weights, and cooperative constraint weights.

[0009] Tangential projection filtering is performed on the synthesized repulsive force field vector to generate a safety force vector;

[0010] The joint velocity increment is obtained by transposing the safety force vector. Based on the joint velocity increment, control commands for the robot are generated to enable the robot to perform obstacle avoidance in real time.

[0011] In one possible implementation, a synthetic repulsive field vector is generated based on the fundamental gravity vector and distance data, combined with preset constraint priority weights, including:

[0012] Based on a multi-level artificial potential field model, a repulsive field vector is generated. The multi-level artificial potential field model includes an environmental constraint repulsive field, a self-collision constraint repulsive field, and a cooperative constraint repulsive field.

[0013] Based on the preset constraint priority weights, the repulsive field vectors are weighted and synthesized to generate a composite repulsive field vector.

[0014] In one possible implementation, a repulsive field vector is generated based on a multi-level artificial potential field model, including:

[0015] Based on the environmental constraint distance, an environmental constraint repulsion field vector is generated. The environmental constraint distance is the shortest distance between the robot's end effector and the target surface of the car to be cleaned.

[0016] Based on the self-collision constraint distance, a self-collision constraint repulsion field vector is generated. The self-collision constraint distance is the shortest distance between non-adjacent links of the robot's arm.

[0017] Based on the cooperative constraint distance, a cooperative constraint repulsion field vector is generated. The cooperative constraint distance is the shortest distance between the robot's end effectors.

[0018] In one possible implementation, the repulsive field vector is weighted and synthesized according to a preset constraint priority weight to generate a synthesized repulsive field vector, including:

[0019] Multiply the environmental constraint repulsion field vector by the environmental constraint weight to obtain the weighted environmental constraint repulsion field vector;

[0020] Multiply the self-collision constraint repulsion field vector by the self-collision constraint weight to obtain the weighted self-collision constraint repulsion field vector;

[0021] Multiplying the cooperative constraint repulsion field vector by the cooperative constraint weights yields the weighted cooperative constraint repulsion field vector.

[0022] The weighted environmental constraint repulsion field vector, the weighted self-collision constraint repulsion field vector, and the weighted cooperative constraint repulsion field vector are added together to obtain the composite repulsion field vector.

[0023] In one possible implementation, the synthetic repulsive field vector is tangentially projected and filtered to generate a safety force vector, including:

[0024] Calculate the dot product between the composite repulsive field vector and the normal vector of the target surface of the car to be cleaned;

[0025] If the dot product is less than zero, the synthesized repulsive field vector is projected onto the tangent plane of the target surface of the car to be cleaned, generating a filtered repulsive field vector.

[0026] In one possible implementation, transposing the safety force vector to obtain the joint velocity increment includes:

[0027] The composite repulsive field vector is mapped to joint velocity increments by the Jacobi transpose, which includes matrix operations that convert the composite repulsive field vector into joint velocity increments.

[0028] Secondly, this application provides a robot collision avoidance control device, comprising:

[0029] The data acquisition module is used to acquire the preset car wash path and the distance data between the robot and the car to be cleaned, and to obtain the basic gravitational vector between the robot and the target surface of the car to be cleaned.

[0030] The repulsive field vector generation module is used to generate a synthetic repulsive field vector based on the basic gravity vector and distance data, combined with preset constraint priority weights. The constraint priority weights include environmental constraint weights, self-collision constraint weights, and cooperative constraint weights.

[0031] The tangential projection module is used to perform tangential projection filtering on the synthesized repulsive force field vector to generate a safe force vector;

[0032] The control command generation module is used to transpose the safety force vector to obtain the joint velocity increment, and generate control commands for the robot based on the joint velocity increment to enable the robot to perform obstacle avoidance in real time.

[0033] Thirdly, this application provides an electronic device, including: a processor and a memory communicatively connected to the processor;

[0034] The memory stores the instructions that the computer executes;

[0035] The processor executes computer-executable instructions stored in memory to implement any of the methods of the first aspect.

[0036] Fourthly, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the method of any one of the first aspects.

[0037] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the method of any one of the first aspects.

[0038] The robot collision avoidance control method, device, equipment, storage medium, and product provided in this application significantly improve the obstacle avoidance capability of a dual-arm car wash robot in complex dynamic scenarios through a multi-level artificial potential field model and tangential projection filtering algorithm. By modeling environmental constraints, self-collision constraints, and cooperative constraints as independent repulsive force fields, conflicts caused by constraint coupling in traditional single-layer potential field methods are avoided. By dynamically adjusting the constraint priority weights, vehicle safety is ensured to have the highest priority. By eliminating the normal component pointing into the target surface through mathematical projection operations, only the tangential component is retained, ensuring both the close-range requirements of the cleaning operation and zero collisions. In dynamic obstacle avoidance scenarios, multi-constraint priority management and absolute guarantee of vehicle safety are achieved, significantly improving the collaborative operation efficiency and safety of the dual-arm car wash robot in complex dynamic environments. Attached Figure Description

[0039] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0040] Figure 1 A schematic diagram illustrating a solution to the obstacle avoidance problem for a dual-arm robot;

[0041] Figure 2 A flowchart illustrating a robot collision avoidance control method provided in an embodiment of this application;

[0042] Figure 3 This is a schematic diagram of the structure of a robot collision avoidance control device provided in an embodiment of this application;

[0043] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0044] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0045] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in any of the foregoing embodiments.

[0046] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of the relevant data all comply with the relevant laws, regulations, and standards of the relevant countries and regions, have taken necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation access points for users to choose to authorize or refuse.

[0047] Furthermore, the technical solution involved in this application, which involves big data analysis of user information (including but not limited to personal biometrics, identity data, consumption data, asset data, electronic terminal operation data, etc.) and the use of artificial intelligence technology for automated decision-making, and makes decisions that have a significant impact on personal rights based on the results of automated decision-making, provides users with corresponding operation entry points for users to choose to agree to or reject the results of automated decision-making; if the user chooses to reject, the process will proceed to the expert decision-making process.

[0048] It should be noted that the robot collision avoidance control method, device, equipment, storage medium and product provided in this application can be used in the field of robot control, or in any field other than robot control. The application field of the robot collision avoidance control method, device, equipment, storage medium and product in this application is not limited.

[0049] This application is specifically applicable to close-range collaborative operation scenarios such as car washing and polishing, especially for dual-arm robots that need to simultaneously meet the complex requirements of environmental safety, self-collision avoidance and dual-arm collaborative avoidance in a confined space. During the car washing process, dual-arm robots need to perform cleaning actions simultaneously in a confined space. For example, one robotic arm is responsible for cleaning the outside of the car body, and the other robotic arm is responsible for cleaning the windows or tires. Such scenarios impose the following special requirements on the robot: (1) Dynamic obstacle avoidance requirements: During the cleaning process, the car body may undergo slight displacement due to vibration or external interference. The end effectors of the dual arms need to sense and adjust the trajectory in real time to avoid collisions. (2) Multi-constraint coupling scenario: Constraints need to be met simultaneously, including environmental constraints (minimum safe distance between the robotic arm and the car body), self-collision constraints (interference avoidance between the links of the robotic arm itself), and collaborative constraints (dynamic avoidance between the end effectors of the dual arms). (3) High real-time requirements: The response speed of the car washing operation to trajectory adjustment is required to be at the millisecond level. Traditional offline planning methods fail because they cannot dynamically compensate. (4) Safety priority: Any obstacle avoidance action must prioritize the safety of the car body, even if it sacrifices the collaborative efficiency of the dual arms or increases energy consumption.

[0050] Figure 1 A schematic diagram illustrating a solution to the obstacle avoidance problem for a dual-arm robot, as shown below. Figure 1As shown, existing technologies mainly solve the obstacle avoidance problem of dual-arm robots through the following three types of solutions: (1) Offline trajectory planning method: Based on teaching or simulation software, a fixed trajectory is generated, and the robotic arm strictly follows the preset path. Although this method can guarantee the global optimality of the trajectory, it lacks real-time adjustment capability and cannot cope with sudden situations such as micro-movement of the car body or dynamic interference of the dual arms during the car washing process. (2) Global planning method based on sampling: By randomly sampling, a collision-free path is searched, but the computational complexity is high, making it difficult to meet the millisecond-level real-time requirements in the car washing scenario, and dynamic obstacles (such as moving car body parts) cannot be effectively modeled. (3) Traditional artificial potential field method: The obstacle is modeled as a repulsive force source, and the target point is a gravitational force source. The robotic arm is guided to avoid obstacles through force field synthesis. However, the traditional artificial potential field method has significant drawbacks in multi-constraint coupled scenarios: when both arms face multiple constraints at the same time, the repulsive field synthesis may fall into a local optimum, causing the robotic arm to be unable to find a feasible path; the dynamic balance between repulsive and attractive forces can easily cause high-frequency vibrations at the end of the robotic arm, affecting the cleaning quality; directly mapping the repulsive force in Cartesian space to joint motion without considering the tangential sliding characteristics of the vehicle body surface may cause the robotic arm to intrude into the vehicle body's restricted area during emergency avoidance.

[0051] Therefore, the existing obstacle avoidance problem of dual-arm robots has the following shortcomings: (1) Insufficient real-time dynamic obstacle avoidance: The existing offline planning and sampling methods cannot adjust the trajectory in real time in the car wash scenario, resulting in collisions or deadlocks of the two arms in the centerline area. (2) Undecoupled multi-constraint conflicts: When environmental constraints (vehicle safety), self-collision constraints (interference of the robotic arm itself), and cooperative constraints (mutual collision of the two arms) act simultaneously, there is a lack of priority processing mechanism, which can easily lead to secondary collision risks (such as collisions caused by obstacle avoidance actions). (3) Coarse Cartesian-joint space mapping: The traditional artificial potential field method does not consider the tangential sliding characteristics of the vehicle surface. Direct mapping may cause the robotic arm to intrude into the vehicle's restricted area during emergency avoidance. (4) Lack of secondary risk control: The existing method may cause collisions with the vehicle body due to improper directional control when avoiding the two arms. Such risks need to be completely eliminated through algorithm design.

[0052] The robot collision avoidance control method, device, equipment, storage medium, and product provided in this application achieve online collision avoidance control of a dual-arm car wash robot in complex dynamic scenarios by constructing a multi-level artificial potential field model and combining a tangential projection filtering algorithm and a multi-constraint priority mapping mechanism. Environmental constraints, self-collision constraints, and cooperative constraints are modeled as repulsive force fields at different levels, and the intensity of each repulsive force is dynamically adjusted through priority weight factors. A mathematical projection algorithm eliminates the normal component pointing towards the vehicle body, retaining only the tangential sliding component, ensuring that obstacle avoidance actions always slide safely along the vehicle body surface. The safety force vector in Cartesian space is mapped to the velocity increment in joint space through Jacobian transpose, and constraint priorities are set differentially through a gain matrix, aiming to solve the aforementioned technical problems of existing technologies.

[0053] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0054] Figure 2 This is a flowchart illustrating a robot collision avoidance control method provided in an embodiment of this application, as shown below. Figure 2 As shown, the method includes:

[0055] S201. Obtain the preset car wash path and the distance data between the robot and the car to be cleaned, and obtain the basic gravitational vector between the robot and the target surface of the car to be cleaned.

[0056] S202. Based on the basic gravity vector and distance data, and combined with the preset constraint priority weights, a synthetic repulsive field vector is generated.

[0057] In this embodiment of the application, the constraint priority weights include environmental constraint weights, self-collision constraint weights, and cooperative constraint weights.

[0058] S203. Perform tangential projection filtering on the synthesized repulsive field vector to generate a safe force vector.

[0059] S204. Transpose the safety force vector to obtain the joint velocity increment, and generate control commands for the robot based on the joint velocity increment to enable the robot to perform obstacle avoidance in real time.

[0060] In one implementation scenario, a five-layer closed-loop control architecture is adopted: [Global Trajectory Injection Layer] → [Multi-Source Perception and Detection Layer] → [Repulsive Field Vector Synthesis Layer] → [Joint Space Mapping Layer] → [Execution and Feedback Layer].

[0061] Specifically, the global trajectory injection layer is used to read the preset car wash trajectory point sequence in real time. Calculate the fundamental gravitational vector pointing to the current target point. This drives the robotic arm to move towards the current target point.

[0062] A multi-source sensing and detection layer is used to monitor in real time using three types of distance sensors: external distance... This is the shortest distance between the end effectors and links of the robotic arm's two arms, used for detecting collisions between the two arms; environmental distance. This refers to the shortest distance between each link of the robotic arm and the point cloud of the car body to be cleaned, used for vehicle distance detection; internal distance. It represents the shortest distance between a single robotic arm and a non-adjacent link, used for self-collision detection.

[0063] The repulsive field vector synthesis layer is used to calculate three types of repulsive forces based on the aforementioned distance thresholds: a) double-arm repulsive force. When the external distance Less than the external distance safety threshold Activated at time, repulsive force of both arms Inversely proportional to external distance. b) Environmental repulsive force. When the environment is far Less than the environmental distance safety threshold Activated at time, environmental repulsion It has the highest gain coefficient. c) Self-collision repulsion. When the internal distance Less than the internal distance safety threshold Activated at time, self-collision repulsion force A damped potential field is generated in the joint space. A direction weighting factor α is introduced, with weighting coefficients between 0 and 1 to adjust the influence of different repulsive force sources, including: The repulsive force weight between the two arms (default 0.6); Set the environmental (vehicle body) repulsion weight (default 1.0, highest priority); This is the weight of the self-collision repulsion force (default 0.8).

[0064] When a single robotic arm avoids another robotic arm, it generates an environmental repulsive force ( When the resultant repulsive force is ), It may be pointing in the direction of the vehicle, posing a collision risk. Calculate the composite repulsive force vector. The dot product with the surface normal vector n of the vehicle body: If the resultant repulsive force vector dot product with the normal vector n of the vehicle body surface If the resultant repulsive force is directed towards the interior of the vehicle body, then it is considered a dangerous situation; if the resultant repulsive force vector... dot product with the normal vector n of the vehicle body surface If the resultant repulsive force points outwards from the vehicle body, it indicates a safe state. When a dangerous state is detected (the dot product of the resultant repulsive force field vector and the normal vector of the target surface of the vehicle to be cleaned is <0), the following formula is used: The composite repulsive vector The tangent plane projected onto the surface of the vehicle body, where, The filtered safety force vector retains only the tangential component parallel to the vehicle body surface, completely eliminating the normal component pointing towards the vehicle body.

[0065] Joint space mapping layer, via formula: The safety force vector of Cartesian space The Jacobian transpose is used to map the joint velocity increments, where... Let be the transpose of the Jacobian matrix of the robotic arm, and K be the diagonal gain matrix. Different gains are set according to constraint priorities, for example... (Environment) Maximum gain, forced priority response, (Self-collision) Medium gain, (Dual arms) Smaller gain.

[0066] The complete control law is achieved using the following formula: .

[0067] The execution and feedback layer is used to process the calculated joint velocity increments. The original trajectory speed command is superimposed on the servo motor and output in real time to form a closed-loop control; the cycle frequency is ≥100Hz to ensure dynamic response capability.

[0068] In another implementation scenario, robot collision avoidance control can be achieved through the following steps:

[0069] Step 1: Global trajectory injection and multi-source perception detection to acquire the preset car wash trajectory point sequence in real time. And real-time monitoring of the distance data between the robotic arm's dual end effectors and the vehicle body, including external distance. Environmental distance Internal distance The system reads the preset car wash trajectory point sequence P in real time and calculates the basic gravitational vector pointing to the current target point. This drives the robotic arm to move towards the target. Three types of distance sensors (external distance sensor, environmental distance sensor, and internal distance sensor) monitor the distance between the end effectors of the dual arms and the vehicle body in real time. Specifically, these include: external distance... : Shortest distance between the end effector and linkage of the dual-arm system (dual-arm collision detection); Environmental distance : Shortest distance between each link and the vehicle body point cloud (vehicle distance detection); Internal distance The shortest distance between non-adjacent links in a single arm (self-collision detection) generates the basic gravity vector. And real-time monitored distance data ( ).

[0070] The preset trajectory point sequence is a set of car wash path points pre-planned through teaching or simulation software, used to drive the robotic arm to perform basic movements. The basic gravity vector is a force vector calculated from the preset trajectory point sequence, pointing towards the target point, driving the robotic arm to move in the target direction. The external distance is the shortest distance between the end effectors and links of both arms, used to detect the risk of collision between the two arms. The environmental distance is the shortest distance between the robotic arm links and the vehicle body point cloud, used to detect the risk of distance between the robotic arm and the vehicle. The internal distance is the shortest distance between non-adjacent links of a single arm, used to detect the risk of self-collision.

[0071] A global trajectory injection layer ensures the robotic arm moves along the optimal path when there is no collision risk, while a multi-source perception and detection layer monitors dynamic environmental changes in real time through various distance sensors, providing a data foundation for subsequent repulsive field synthesis. This step solves the problem of offline trajectory planning not being able to be dynamically adjusted in existing technologies, and at the same time, it achieves comprehensive coverage of the risks of collision between the two arms, distance to the vehicle, and self-collision through multi-source perception.

[0072] Step 2: Repulsive field vector synthesis and tangential projection filtering to obtain the fundamental gravitational vector generated in Step 1. And real-time monitored distance data ( The following calculations are performed, including:

[0073] Basic repulsive force calculation: based on each distance threshold ( Calculate the three types of repulsive forces separately: repulsive force of both arms :when Activated by time, inversely proportional to distance; repulsive force from the environment. :when Time-activated, with the highest gain coefficient; self-collision repulsion. :when When activated, a damped potential field is generated in the joint space.

[0074] Directional weight factor adjustment: Introduce a weight coefficient between 0 and 1 (e.g., Adjusting the influence of different repulsive force sources.

[0075] Tangential projection filtering: When a synthetic repulsive vector is detected dot product with the normal vector n of the vehicle body surface In (dangerous situation), The tangent plane projected onto the vehicle body surface generates the safety force vector. .

[0076] Generate filtered safety force vector And the repulsive vector adjusted according to the weighting factor ( The repulsive field, modeled by a repulsive function, represents the obstacle repulsion force used to drive the robotic arm away from obstacles. The direction weight factor adjusts the coefficients of different repulsive force sources, with higher-priority constraints receiving higher weight values. Tangential projection eliminates the normal component pointing towards the vehicle body through mathematical projection operations, retaining only the tangential sliding component.

[0077] By employing a multi-level artificial potential field model and a tangential projection filtering algorithm, the robot arm's movement direction is dynamically adjusted to ensure that obstacle avoidance actions always glide safely along the vehicle's surface. The priority of different constraints (environmental constraints > self-collision constraints > cooperative constraints) is adjusted by a direction weight factor, completely eliminating the risk of secondary collisions. The tangential projection filtering algorithm solves the problem of intrusion into the vehicle's restricted area caused by coarse direction mapping in traditional artificial potential field methods, achieving a "vehicle-hugging" obstacle avoidance effect.

[0078] In another implementation scenario, a dynamic obstacle trajectory prediction module is introduced. This module predicts the vehicle's micro-motion trends using Kalman filtering or particle filtering algorithms and dynamically adjusts the repulsive force distribution of the artificial potential field. Specifically, this includes: dynamic obstacle modeling: modeling the vehicle's micro-motions as dynamic obstacles with velocity and acceleration, rather than static point clouds; and dynamic potential field adjustment: adjusting the parameters of the repulsive force function (such as repulsive force gain and radius of action) in real time based on the predicted obstacle trajectory, allowing the repulsive force field to dynamically evolve with the obstacle's motion state.

[0079] By introducing dynamic obstacle trajectory prediction, the obstacle avoidance lag problem caused by relying solely on static distance detection is solved. Dynamically adjusting the repulsive field parameters allows the robotic arm to avoid potential collision risks in advance, especially when the vehicle body undergoes continuous micro-movements due to vibration or external disturbances. This significantly improves the predictability and adaptability of obstacle avoidance actions and avoids robotic arm tremors or path interruptions caused by sudden trajectory changes.

[0080] In another implementation scenario, in a three-level priority weight model ( Based on this, an adaptive adjustment mechanism is introduced to dynamically adjust the weighting factors according to the real-time environment. Specifically, this includes: 1. Environmental risk assessment: Evaluating the current environmental risk level (e.g., changes in vehicle surface curvature, dual-arm collaboration density) through multi-source sensor fusion (e.g., LiDAR and visual data). 2. Dynamic mapping of weighting factors: Dynamically adjusting based on the risk level. , , The value, for example, is temporarily increased in areas of abrupt change in the curvature of the vehicle body. Or reduce in areas with high arm-to-arm coordination .

[0081] By using environmental risk assessment and dynamic mapping of weighting factors, the problem of fixed weighting factors being unable to adapt to complex dynamic scenarios is solved. The adaptive adjustment mechanism enables the system to automatically optimize obstacle avoidance strategies under different risk levels. For example, it prioritizes vehicle safety in areas with abrupt changes in vehicle curvature and appropriately relaxes cooperation constraints in areas with dense dual-arm cooperation, thereby achieving more flexible and safer obstacle avoidance control in complex environments.

[0082] In another implementation scenario, based on the existing tangential projection filtering algorithm, a vehicle surface curvature perception module is introduced. This module extracts vehicle surface curvature information from point cloud data and dynamically adjusts the mathematical model of the tangential projection. Specifically, this includes: 1. Curvature feature extraction: Calculating the local curvature of the vehicle surface at the current contact point using a surface fitting algorithm (such as the least squares method). 2. Tangential projection optimization: Adjusting the projection direction based on curvature information. For example, a more conservative projection direction is used in high-curvature areas (such as door recesses) to prevent the sliding component from pointing towards dangerous areas due to curvature changes.

[0083] By optimizing the tangential projection direction through curvature perception, the problem of unreasonable obstacle avoidance paths caused by neglecting the geometric characteristics of the vehicle body surface is solved. In high-curvature areas such as door recesses, the optimized projection direction allows the robotic arm to slide along a safe curvature path, avoiding the sliding component caused by abrupt curvature changes from intruding into the vehicle body's restricted areas, thus achieving a more fitting obstacle avoidance effect under complex vehicle body geometry.

[0084] In another implementation scenario, based on three types of distance sensors (external, environmental, and internal distance sensors), a multi-sensor fusion algorithm (such as weighted averaging) is introduced to improve obstacle avoidance reliability through redundant perception. Specifically, this includes: 1. Sensor data fusion: weighted fusion of detection results from LiDAR, visual sensors, and ultrasonic sensors to generate more robust distance data; 2. Anomaly detection and correction: identifying abnormal sensor data (such as false detections caused by occlusion) through statistical analysis or neural network models and dynamically correcting the distance threshold.

[0085] By integrating multi-sensor fusion and anomaly detection, the problem of misjudgment caused by the susceptibility of a single sensor to environmental interference is solved. For example, in scenarios involving vehicle reflections or obstruction by a robotic arm, redundant perception ensures the continuity and accuracy of distance detection, preventing obstacle avoidance strategy failure due to sensor malfunction, thereby achieving more stable obstacle avoidance control under complex lighting or occlusion conditions.

[0086] In another implementation scenario, a reinforcement learning framework (such as a deep Q-network) is introduced on top of the three-level priority model to dynamically optimize constraint priorities using historical obstacle avoidance data. Specifically, this includes: 1. State-Action Modeling: Modeling the current environmental state (such as distance data, vehicle curvature) and obstacle avoidance actions (such as repulsion weight adjustment) as a Markov decision process. 2. Priority Self-Optimization: Training the model using reinforcement learning algorithms allows the system to automatically learn the optimal constraint priority strategy through long-term interactions, such as dynamically increasing α_env or decreasing α_ext in specific scenarios.

[0087] By employing reinforcement learning to achieve self-optimization of constraint priorities, the problem of fixed-priority models being unable to adapt to long-term environmental changes is solved. After the car wash robot experiences multiple obstacle avoidance failures, the model can automatically adjust its priority strategy to avoid repeatedly falling into local optima, thereby achieving more efficient and stable obstacle avoidance control in complex dynamic environments.

[0088] The robot collision avoidance control method provided in this embodiment significantly improves the obstacle avoidance capability of a dual-arm car wash robot in complex dynamic scenarios through a multi-level artificial potential field model and a tangential projection filtering algorithm. By modeling environmental constraints, self-collision constraints, and cooperative constraints as independent repulsive force fields, conflicts caused by constraint coupling in traditional single-layer potential field methods are avoided. By dynamically adjusting the constraint priority weights, vehicle safety is ensured to have the highest priority. Mathematical projection operations eliminate the normal component pointing into the target surface, retaining only the tangential component, ensuring both the close-range requirements of the cleaning operation and zero collisions. In dynamic obstacle avoidance scenarios, multi-constraint priority management and absolute vehicle safety are achieved, significantly improving the collaborative operation efficiency and safety of the dual-arm car wash robot in complex dynamic environments.

[0089] Optionally, based on the basic gravity vector and distance data, and combined with preset constraint priority weights, a synthetic repulsive field vector is generated, including: generating a repulsive field vector based on a multi-level artificial potential field model, which includes an environmental constraint repulsive field, a self-collision constraint repulsive field, and a cooperative constraint repulsive field; and weighting and synthesizing the repulsive field vector according to preset constraint priority weights to generate a synthetic repulsive field vector.

[0090] In one example, a multi-level artificial potential field model is constructed, and environmental constraints (vehicle safety), self-collision constraints (robotic arm interference), and cooperative constraints (mutual collision between the two arms) are modeled as repulsive force fields at different levels. The intensity of each repulsive force is dynamically adjusted through a three-level priority weight model (α_env=1.0>α_int=0.8>α_ext=0.6) to ensure that vehicle safety has the highest priority. Combined with a tangential projection filtering algorithm, the normal component pointing to the vehicle body is eliminated, and only the tangential sliding component is retained to avoid collisions caused by obstacle avoidance actions.

[0091] By employing a multi-level artificial potential field model and a priority weighting mechanism, the local optimum trap and secondary risk problems of traditional artificial potential field methods in scenarios with multiple constraints are resolved. Environmental constraints, self-collision constraints, and cooperative constraints are modeled as independent repulsive force fields, avoiding conflicts caused by constraint coupling in traditional single-layer potential field methods. By dynamically adjusting the weighting factors, vehicle safety is ensured to always have the highest priority, and any obstacle avoidance action will not be directed towards the vehicle's interior. By eliminating the normal component through mathematical projection operations, the robotic arm "glides" along the vehicle surface rather than "collides" during emergency avoidance, completely eliminating the risk of secondary collisions.

[0092] Optionally, based on a multi-level artificial potential field model, a repulsive field vector is generated, including: generating an environmental constraint repulsive field vector based on the environmental constraint distance, where the environmental constraint distance is the shortest distance between the robot's end effector and the target surface of the car to be cleaned; generating a self-collision constraint repulsive field vector based on the self-collision constraint distance, where the self-collision constraint distance is the shortest distance between non-adjacent links of the robot's robotic arm; and generating a cooperative constraint repulsive field vector based on the cooperative constraint distance, where the cooperative constraint distance is the shortest distance between the robot's end effectors.

[0093] In one example, three types of distance sensors—environmental constraint, self-collision constraint, and cooperative constraint—calculate corresponding repulsive field vectors: 1. Environmental constraint repulsive field vector: When the distance between the robotic arm's end effector and the target surface is less than a safety threshold, the environmental constraint repulsive field vector is activated, driving the robotic arm away from the target surface; 2. Self-collision constraint repulsive field vector: When the distance between non-adjacent links of the robotic arm is less than a safety threshold, the self-collision constraint repulsive field vector is activated, driving the robotic arm to adjust its posture; 3. Cooperative constraint repulsive field vector: When the distance between the end effectors of both arms is less than a safety threshold, the cooperative constraint repulsive field vector is activated, driving the two arms to adjust their relative positions. The environmental constraint distance is the shortest distance between the robotic arm's end effector and the target surface (e.g., the car body), used to detect vehicle distance risk. For example, when the robotic arm is cleaning a car door, the environmental constraint distance is the minimum distance between the end effector and the car door surface. The self-collision constraint distance is the shortest distance between non-adjacent links of the robotic arm, used to detect self-collision risk. For example, when the upper and lower arms of the robotic arm are close together, the self-collision constraint distance is the minimum distance between them. The collaboration constraint distance is the shortest distance between the end effectors of the two arms, used to detect the risk of collision between the two arms. For example, when the two arms are working together in the centerline area, the collaboration constraint distance is the minimum distance between the two end effectors.

[0094] By refining the generation steps of the multi-level artificial potential field model, independent modeling and dynamic perception of three types of constraints (environmental constraints, self-collision constraints, and cooperative constraints) are achieved. The environmental constraint repulsive field vector prioritizes the response to vehicle safety requirements, the self-collision constraint repulsive field vector prevents interference from the robotic arm itself, and the cooperative constraint repulsive field vector adjusts the risk of collision between the two arms, thereby achieving more precise obstacle avoidance control in complex dynamic scenarios.

[0095] Optionally, the repulsion field vectors are weighted and synthesized according to preset constraint priority weights to generate a composite repulsion field vector, including: multiplying the environmental constraint repulsion field vector by the environmental constraint weight to obtain a weighted environmental constraint repulsion field vector; multiplying the self-collision constraint repulsion field vector by the self-collision constraint weight to obtain a weighted self-collision constraint repulsion field vector; multiplying the cooperative constraint repulsion field vector by the cooperative constraint weight to obtain a weighted cooperative constraint repulsion field vector; and adding the weighted environmental constraint repulsion field vector, the weighted self-collision constraint repulsion field vector, and the weighted cooperative constraint repulsion field vector together to obtain a composite repulsion field vector.

[0096] In one example, the three types of repulsive field vectors are weighted and synthesized by constraint priority weights: weighted environmental constraint repulsive field vector: the environmental constraint weight is the highest ( =1.0), ensuring vehicle safety priority response; weighted self-collision constraint repulsion field vector: self-collision constraint weight second highest ( =0.8), to prevent the robotic arm from interfering with itself; weighted cooperative constraint repulsive field vector: the cooperative constraint weight is the lowest ( =0.6), allowing the arms to move closer together when necessary; Vector addition: The three weighted repulsion field vectors are added together to generate the final composite repulsion field vector. The weighted environmental constraint repulsion field vector is the product of the environmental constraint repulsion field vector and the environmental constraint weight, reflecting the priority of the environmental constraint. For example, when the environmental constraint weight is 1.0, the weighted environmental constraint repulsion field vector is equal to the original repulsion field vector. The weighted self-collision constraint repulsion field vector is the product of the self-collision constraint repulsion field vector and the self-collision constraint weight, reflecting the priority of the self-collision constraint. For example, when the self-collision constraint weight is 0.8, the strength of the weighted self-collision constraint repulsion field vector is lower than the original repulsion field vector. The weighted cooperative constraint repulsion field vector is the product of the cooperative constraint repulsion field vector and the cooperative constraint weight, reflecting the priority of the cooperative constraint. For example, when the cooperative constraint weight is 0.6, the strength of the weighted cooperative constraint repulsion field vector is the lowest.

[0097] By refining the weighted synthesis steps of constraint priority, dynamic priority management of the three types of constraints is achieved. The environmental constraint repulsion field vector always has the highest priority, ensuring the vehicle body responds first for safety; the cooperative constraint repulsion field vector has the lowest priority, allowing the two arms to move closer appropriately when necessary, thus enabling a more flexible obstacle avoidance strategy in complex dynamic scenarios.

[0098] Optionally, the composite repulsive field vector is tangentially projected and filtered to generate a safety force vector, including: calculating the dot product of the composite repulsive field vector and the normal vector of the target surface of the car to be cleaned; if the dot product is less than zero, the composite repulsive field vector is projected onto the tangential plane of the target surface of the car to be cleaned to generate a filtered repulsive field vector.

[0099] In one example, the dot product is used to determine whether the resulting repulsive field vector points into the target surface: 1. Dot product calculation: Calculate the resulting repulsive field vector. dot product with the target surface normal vector n 2. Danger State Judgment: If the dot product is less than zero, it indicates that the repulsive field vector points into the target surface (danger state); 3. Tangential Projection: Project the synthesized repulsive field vector onto the tangential plane of the target surface to generate a filtered repulsive field vector. The target surface normal vector is the normal direction of the target surface (such as the car body) at the current contact point. For example, when the end effector of the robotic arm contacts the door surface, the target surface normal vector is the normal direction of the door surface at that point.

[0100] By refining the tangential projection filtering process, the components of dangerous directions are eliminated. For example, in the door recess area, the filtered repulsive field vector causes the robotic arm to slide along the vehicle surface instead of pointing inwards, thus ensuring the safety of obstacle avoidance maneuvers.

[0101] Optionally, transposing the safety force vector to obtain the joint velocity increment includes: mapping the composite repulsive field vector to the joint velocity increment through Jacobian transpose, wherein Jacobian transpose includes matrix operations to convert the composite repulsive field vector into the joint velocity increment.

[0102] In one example, the composite repulsive field vector in Cartesian space is transformed by the Jacobian transpose. velocity increments mapped to joint space 1. Matrix operations: using the transpose of the Jacobian matrix Cartesian space force vector Converted to joint velocity increment 2. Mapping result: velocity increment This is used to drive a robotic arm to perform obstacle avoidance maneuvers. The Jacobian transpose is a matrix operation that converts a Cartesian force vector into joint velocity increments. For example, if the force vector of the robotic arm's end effector is F, the Jacobian transpose J^T maps it to joint velocity increments. .

[0103] By introducing the Jacobian transpose mapping, an efficient conversion from the Cartesian space repulsion vector to the joint velocity increment is achieved, ensuring the feasibility of obstacle avoidance actions in the joint space, thereby improving the dynamic response capability of the dual-arm car wash robot.

[0104] Figure 3 This is a schematic diagram of the structure of a robot collision avoidance control device provided in an embodiment of this application, as shown below. Figure 3 As shown, the robot anti-collision control device 30 provided in this embodiment includes:

[0105] The data acquisition module 301 is used to acquire the preset car wash path and the distance data between the robot and the car to be cleaned, and to obtain the basic gravitational vector between the robot and the target surface of the car to be cleaned.

[0106] The repulsive field vector generation module 302 is used to generate a synthetic repulsive field vector based on the basic gravity vector and distance data, combined with preset constraint priority weights. The constraint priority weights include environmental constraint weights, self-collision constraint weights, and cooperative constraint weights.

[0107] The tangential projection module 303 is used to perform tangential projection filtering on the synthesized repulsive force field vector to generate a safety force vector;

[0108] The control command generation module 304 is used to transpose the safety force vector to obtain the joint velocity increment, and generate control commands for the robot based on the joint velocity increment so that the robot can perform obstacle avoidance in real time.

[0109] In one possible implementation, the repulsion field vector generation module 302 is specifically used to: generate a repulsion field vector based on a multi-level artificial potential field model, wherein the multi-level artificial potential field model includes an environmental constraint repulsion field, a self-collision constraint repulsion field, and a cooperative constraint repulsion field; and perform weighted synthesis of the repulsion field vector according to a preset constraint priority weight to generate a synthesized repulsion field vector.

[0110] In one possible implementation, the repulsion field vector generation module 302 is further specifically used to: generate an environmental constraint repulsion field vector based on the environmental constraint distance, where the environmental constraint distance is the shortest distance between the robot's end effector and the target surface of the car to be cleaned; generate a self-collision constraint repulsion field vector based on the self-collision constraint distance, where the self-collision constraint distance is the shortest distance between non-adjacent links of the robot's robotic arm; and generate a cooperative constraint repulsion field vector based on the cooperative constraint distance, where the cooperative constraint distance is the shortest distance between the robot's end effectors.

[0111] The robot collision avoidance control device is also specifically used for:

[0112] In one possible implementation, the repulsion field vector generation module 302 is further specifically used to: multiply the environmental constraint repulsion field vector by the environmental constraint weight to obtain a weighted environmental constraint repulsion field vector; multiply the self-collision constraint repulsion field vector by the self-collision constraint weight to obtain a weighted self-collision constraint repulsion field vector; multiply the cooperative constraint repulsion field vector by the cooperative constraint weight to obtain a weighted cooperative constraint repulsion field vector; and add the weighted environmental constraint repulsion field vector, the weighted self-collision constraint repulsion field vector, and the weighted cooperative constraint repulsion field vector together to obtain a composite repulsion field vector.

[0113] In one possible implementation, the tangential projection module 303 is specifically used to: calculate the dot product of the synthesized repulsive field vector and the normal vector of the target surface of the car to be cleaned; if the dot product is less than zero, the synthesized repulsive field vector is projected onto the tangential plane of the target surface of the car to be cleaned to generate a filtered repulsive field vector.

[0114] In one possible implementation, the control instruction generation module 304 is specifically used to: map the composite repulsive field vector into joint velocity increments through Jacobian transpose, wherein Jacobian transpose includes matrix operations to convert the composite repulsive field vector into joint velocity increments.

[0115] The robot anti-collision control device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0116] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 4 As shown, the electronic device 40 may include a memory 401 and a processor 402. Optionally, the electronic device may also include a transceiver 403, wherein the memory 401 and the processor 402 communicate with each other; for example, the memory 401, the processor 402 and the transceiver 403 may communicate via a communication bus 404, the memory 401 is used to store a computer program, and the processor 402 executes the computer program to implement the method of the above embodiments.

[0117] Optionally, the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps in the method embodiments disclosed in this application can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0118] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the methods in any of the above method embodiments.

[0119] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the methods in any of the above method embodiments.

[0120] All or part of the steps in the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a readable memory. When the program is executed, it performs the steps of the above method embodiments; and the aforementioned memory (storage medium) includes: read-only memory (ROM), RAM, flash memory, hard disk, solid-state drive, magnetic tape, floppy disk, optical disk, and any combination thereof.

[0121] This application describes embodiments with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processing unit of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processing unit of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0122] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0123] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0124] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of this application and its equivalents, this application also intends to include these modifications and variations.

[0125] In this application, the term "comprising" and its variations can refer to non-limiting inclusion; the term "or" and its variations can refer to "and / or". The terms "first", "second", etc., in this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. In this application, "multiple" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0126] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this application.

[0127] It should be further noted that although the steps in the flowchart are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowchart may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0128] It should be understood that the above-described device embodiments are merely illustrative, and the device of this application can also be implemented in other ways. For example, the division of units / modules in the above embodiments is only a logical functional division, and there may be other division methods in actual implementation. For example, multiple units, modules, or components may be combined, or integrated into another system, or some features may be ignored or not executed.

[0129] Furthermore, unless otherwise specified, the functional units / modules in the various embodiments of this application can be integrated into one unit / module, or each unit / module can exist physically separately, or two or more units / modules can be integrated together. The integrated units / modules described above can be implemented in hardware or as software program modules.

[0130] When integrated units / modules are implemented in hardware, the hardware can be digital circuits, analog circuits, etc. The physical implementation of the hardware structure includes, but is not limited to, transistors, memristors, etc. Unless otherwise specified, the processor can be any suitable hardware processor, such as a CPU, GPU, FPGA, DSP, and ASIC, etc. Unless otherwise specified, the storage unit can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc.

[0131] If the integrated unit / module is implemented as a software program module and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0132] In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments. The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0133] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein.

[0134] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

Claims

1. A robot collision avoidance control method, characterized in that, The method includes: Obtain the preset car wash path and the distance data between the robot and the car to be cleaned, and obtain the basic gravitational vector between the robot and the target surface of the car to be cleaned. Based on the basic gravity vector and the distance data, and combined with the preset constraint priority weights, a synthetic repulsive field vector is generated. The constraint priority weights include environmental constraint weights, self-collision constraint weights, and cooperative constraint weights. The synthesized repulsive field vector is tangentially projected and filtered to generate a safety force vector; The safety force vector is transposed to obtain the joint velocity increment. Based on the joint velocity increment, control commands for the robot are generated to enable the robot to perform real-time obstacle avoidance. The process of generating a synthetic repulsive field vector based on the fundamental gravity vector and the distance data, combined with preset constraint priority weights, includes: Based on a multi-level artificial potential field model, a repulsive field vector is generated. The multi-level artificial potential field model includes an environmental constraint repulsive field, a self-collision constraint repulsive field, and a cooperative constraint repulsive field. According to the preset constraint priority weight, the repulsive field vector is weighted and synthesized to generate the synthesized repulsive field vector; The step of performing tangential projection filtering on the synthesized repulsive field vector to generate a safety force vector includes: Calculate the dot product between the synthesized repulsive field vector and the normal vector of the target surface of the car to be cleaned; If the dot product is less than zero, the synthesized repulsive field vector is projected onto the tangent plane of the target surface of the car to be cleaned to generate a filtered repulsive field vector.

2. The method according to claim 1, characterized in that, The generation of the repulsive field vector based on the multi-level artificial potential field model includes: An environmental constraint repulsion field vector is generated based on the environmental constraint distance, where the environmental constraint distance is the shortest distance between the robot's end effector and the target surface of the car to be cleaned. Based on the self-collision constraint distance, a self-collision constraint repulsion field vector is generated, wherein the self-collision constraint distance is the shortest distance between non-adjacent links of the robot's robotic arm; Based on the cooperative constraint distance, a cooperative constraint repulsion field vector is generated, wherein the cooperative constraint distance is the shortest distance between the end effectors of the robot.

3. The method according to claim 2, characterized in that, The step of weighting and synthesizing the repulsive field vector according to the preset constraint priority weight to generate the synthesized repulsive field vector includes: Multiplying the environmental constraint repulsion field vector by the environmental constraint weights yields the weighted environmental constraint repulsion field vector. Multiply the self-collision constraint repulsion field vector by the self-collision constraint weight to obtain the weighted self-collision constraint repulsion field vector; Multiplying the cooperative constraint repulsion field vector by the cooperative constraint weight yields the weighted cooperative constraint repulsion field vector; The weighted environmental constraint repulsion field vector, the weighted self-collision constraint repulsion field vector, and the weighted cooperative constraint repulsion field vector are added together to obtain the composite repulsion field vector.

4. The method according to claim 1, characterized in that, The step of transposing the safety force vector to obtain the joint velocity increment includes: The composite repulsive field vector is mapped to the joint velocity increment by the Jacobian transpose, which includes a matrix operation that converts the composite repulsive field vector into the joint velocity increment.

5. A robot collision avoidance control device, characterized in that, The device includes: The data acquisition module is used to acquire the preset car wash path and the distance data between the robot and the car to be cleaned, and to obtain the basic gravitational vector between the robot and the target surface of the car to be cleaned. The repulsive field vector generation module is used to generate a synthetic repulsive field vector based on the basic gravitational vector and the distance data, combined with preset constraint priority weights. The constraint priority weights include environmental constraint weights, self-collision constraint weights, and cooperative constraint weights. The repulsive field vector generation module is specifically used to generate a repulsive field vector based on a multi-level artificial potential field model, wherein the multi-level artificial potential field model includes an environmental constraint repulsive field, a self-collision constraint repulsive field, and a cooperative constraint repulsive field; and to perform weighted synthesis of the repulsive field vector according to the preset constraint priority weights to generate the synthesized repulsive field vector. The tangential projection module is used to perform tangential projection filtering on the synthesized repulsive field vector to generate a safety force vector; The tangential projection module is specifically used to calculate the dot product of the synthesized repulsive field vector and the normal vector of the target surface of the car to be cleaned; if the dot product is less than zero, the synthesized repulsive field vector is projected onto the tangential plane of the target surface of the car to be cleaned to generate a filtered repulsive field vector. The control command generation module is used to transpose the safety force vector to obtain the joint velocity increment, and generate control commands for the robot based on the joint velocity increment, so that the robot can perform obstacle avoidance in real time.

6. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1 to 4.

8. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 4.

Images