Dynamic optimization method for flexible operation space of robot in complex scene

Abstract

The invention discloses a space dynamic optimization method for flexible operation of a robot in a complex scene, and the method comprises the steps: obtaining the point cloud information of a scene and an object through a laser radar, a depth camera and the like, carrying out the denoising and optimization processing through multi-algorithm fusion, precisely extracting the quality center of the object, and dynamically adjusting the pose. The flexible working space of the mechanical arm is calculated through the Monte Carlo algorithm, the operation boundary is finely described through a space envelope diagram, and the optimal operation path is achieved in combination with inverse kinematics solution. Through combination of a reverse pose adjustment strategy and real-time collision determination, high-degree-of-freedom safe operation in a complex environment is ensured. According to the method, the adaptability and the operation success rate of the robot in a complex stacking scene are remarkably improved, dependence on a high-cost sensor is reduced, and flexible and accurate grabbing is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of robot operation optimization and automatic control, and in particular to a method for dynamically optimizing a robot's flexible operation space in complex scenarios. Background Art

[0002] The ability of mobile robots to operate in complex scenarios is of great significance to industrial automation, warehousing and logistics, etc., especially when handling irregularly shaped and complexly stacked objects. Traditional static operation methods often fail to meet the requirements of efficient and precise operations. Usually, mobile robots perceive the environment through laser radar, visual sensors, etc., and use path planning algorithms to achieve positioning and grasping of the robot arm. However, in complex scenarios, the existing technology has the following problems: 1. Limitations of path planning and operating space: Traditional robot operations mostly rely on preset path planning and static workspaces. This approach lacks flexibility when facing scene changes, resulting in limited operating points of the robotic arm and inability to achieve optimal position grasping under multiple degrees of freedom. Especially when handling stacked, misplaced or irregular objects, the operation success rate is low.

[0003] 2. Point cloud data processing is complex and has limited accuracy: In existing technologies, point cloud data is usually collected through depth cameras, and point cloud information is optimized using RANSAC algorithms and filtering and denoising. However, this process is easily affected by ambient lighting and noise interference, resulting in insufficient accuracy in object recognition and positioning, making it difficult to operate efficiently in complex scenes.

[0004] 3. The contradiction between robot kinematics solution and collision avoidance: In complex scenarios, robot operation requires multi-step inverse kinematics solution and obstacle avoidance strategy. Currently, most common collision avoidance methods are static settings, which cannot dynamically adjust the posture and path of the robot in real time, resulting in low operation efficiency, inability to flexibly respond to environmental changes, and prone to collision risks.

[0005] 4. Existing optimization methods have poor adaptability and high cost: In order to cope with complex operations, the industry sometimes uses high-precision sensors or complex algorithms to improve recognition and operation accuracy. However, such methods not only increase the complexity and cost of the system, but also are difficult to popularize in large-scale practical applications.

[0006] Therefore, how to achieve dynamic optimization of the robot arm's operating space in complex scenarios, improve grasping accuracy and success rate, and reduce dependence on high-cost sensors has become a technical problem to be solved by the present invention. Summary of the invention

[0007] The technical problem solved by the present invention is to provide a method for dynamically optimizing the flexible operating space of a robot in complex scenarios in response to the defects existing in the above-mentioned prior art, so as to solve the problems of limited operating space, insufficient operating accuracy, poor obstacle avoidance ability and high system cost of the robotic arm in complex scenarios proposed in the above-mentioned background technology.

[0008] In order to solve the above technical problems, the technical solution adopted by the present invention is as follows: A method for dynamically optimizing a robot's flexible operating space in a complex scenario comprises the following steps: S1: Control the mobile robot to move to the position to be operated, the mobile robot includes the mobile chassis equipped with a laser radar, based on the laser radar navigation and the visual system of the depth camera, autonomously avoids obstacles through the ROS system and moves to a feature scene and stops; S2: The point cloud information of the feature scene and the item to be operated is obtained through the depth camera, and the object point cloud information is suppressed and denoised to remove the irrelevant scene point cloud to obtain high-quality point cloud information; S3: Construct the scene through the RANSAC algorithm and the minimum project square method, obtain the point cloud information of the scene, calculate the center of mass of the project point cloud and define it as the operation target; S4: The Monte Carlo algorithm is used to calculate the approach workspace of the manipulator, and the operating points of the flexible workspace are obtained through inverse kinematics solution. The flexible workspace is sampled and its spatial envelope is calculated; S5: Make the maximum intersection between the flexible workspace and the object point cloud, calculate the position transformation of the mobile robot according to the maximum intersection, and guide the point cloud of the object to be operated into the flexible workspace; S6: According to the optimal posture transformation information, the mobile structure is controlled to move to the optimal operation space position, so that the flexible workspace of the robot arm and the point cloud information of the feature scene are optimally coupled, thereby achieving a high success rate of operation.

[0009] As a further solution of the present invention, the point cloud optimization algorithm in step S2 includes suppression and noise removal processing to remove irrelevant point clouds, improve calculation efficiency, and generate a maximum spatial envelope diagram.

[0010] As a further solution of the present invention, the scene plane in step S3 includes obtaining a plane equation using a RANSAC algorithm, and performing vertical calculation on the algorithm support of the plane as a Z axis to segment the scene plane and the project point cloud.

[0011] As a further solution of the present invention, the flexible workspace calculation in step S4 includes: Monte Carlo sampling of the robot's neighboring workspace; Monte Carlo sampling is performed on the adjacent workspace of the robot arm. The position of the operating point is fixed, and it is rotated around the x, y, and z axes in steps of 30°. For each posture, it is verified whether the inverse kinematics solution exists. The solution is the operating point of the flexible workspace.

[0012] As a further solution of the present invention, the posture transformation in step S5 includes: Based on the calculation of the object point cloud centroid, guide the object point cloud centroid to move along the preset path to the center point of the robot arm base coordinate system; Adjust the project point cloud layout according to the optimal intersection, so that the project can fully enter the flexible workspace.

[0013] As a further solution of the present invention, step S5 also includes workspace adaptation adjustment. When the project point cloud cannot fully adapt to the flexible workspace, the posture of the mobile robot structure is adjusted to achieve partial intersection of the project point cloud and the robot arm workspace.

[0014] As a further solution of the present invention, step S5 further includes collision parameters and adjustments. When the robot arm moves to the collision parameter limit, it stops moving and adjusts its position to maximize the coverage of the workspace.

[0015] As a further solution of the present invention, the optimized operating space position in step S6 is obtained by inverse posture adjustment calculation, and the adjustment enables the robotic arm to maintain multi-degree-of-freedom flexibility in complex scenes, thereby achieving the best operating effect.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. Flexible workspace optimization based on spatial envelope diagram: The present invention specifically introduces the calculation and application of the spatial envelope diagram in the calculation of the flexible workspace of the robot arm. This method is not only limited to simple accessibility verification, but also describes the morphological boundary of the entire flexible workspace through the envelope diagram, providing more accurate data support for subsequent operation optimization. Compared with the traditional single workspace description method, the introduction of the spatial envelope diagram can intuitively reflect the operating boundary of the robot arm, making the adjustment of the flexible workspace more refined and dynamic, so as to complete tasks more flexibly in complex scenarios.

[0017] 2. Precise coordination of point cloud centroid guidance and robot arm posture transformation: The invention calculates the centroid of the object point cloud and uses the centroid posture to guide the operating position of the robot arm. This strategy is rare in the prior art. It particularly solves the problem of how to optimize the operating position when grasping irregular objects. By dynamically adjusting the center of mass, the robot arm's posture changes are guided so that the object enters the optimal grasping area, thereby significantly improving the success rate of grasping. This precise coordination mechanism avoids the limitations of traditional path planning when facing complex objects.

[0018] 3. Innovative design combining reverse posture adjustment strategy with real-time collision judgment: The combination of the reverse posture adjustment strategy and real-time detection of collision parameters in the present invention is a key innovation to ensure operational safety and flexibility. When adjusting the operation path of the robot arm, the operation posture is dynamically corrected through real-time collision judgment to avoid potential collisions between the robot arm and the environment. This combination is particularly advantageous in complex scenarios, and can complete refined operations with a high degree of freedom. It is a major breakthrough in static obstacle avoidance methods and ensures flexibility and safety.

[0019] 4. High efficiency of point cloud denoising and optimization processing in complex stacking scenes: For point cloud data in complex stacking scenes, the present invention obtains multi-view point cloud information through a depth camera, and combines it with a unique denoising processing technology to effectively remove irrelevant point clouds, thereby improving computing efficiency and data quality. Traditional point cloud processing usually faces large noise interference, while the denoising optimization processing of the present invention greatly reduces the impact of invalid data on operation path planning while ensuring the integrity of point cloud information, and enhances the system's adaptability in complex scenes.

[0020] 5. Multi-step joint optimization achieves high degree of freedom and multi-scenario adaptation: The present invention uses multi-step optimization design to cover the whole process optimization from initial positioning, workspace sampling, posture adjustment to final operation, so that the system can flexibly deal with various uncertainties in complex scenarios. Especially in the processing of objects of different sizes and shapes, the operation strategy can be dynamically adjusted to improve the success rate. Compared with the existing fixed process operation method, it has higher operation freedom and scene adaptability.

[0021] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0023] Figure 1 4 is a flowchart of a software recommendation method in an embodiment of the present invention.

[0024] Figure 2 Schematic diagram of the structure of the software recommendation system in the embodiment of the present invention.

[0025] In the figure: 1. LiDAR; 2. Mobile chassis; 3. Robotic arm; 4. Camera; 5. Gripper; 6. Feature scene; 7. Object to be operated. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0027] See also Figure 1 —2. In an embodiment of the present invention, a method for dynamically optimizing a robot's flexible operating space in a complex scenario includes the following steps: S1: Control the mobile robot to move to the position to be operated, the mobile robot includes the mobile chassis equipped with a laser radar, based on the laser radar navigation and the visual system of the depth camera, autonomously avoids obstacles through the ROS system and moves to a feature scene and stops; S2: The point cloud information of the feature scene and the item to be operated is obtained through the depth camera, and the object point cloud information is suppressed and denoised to remove the irrelevant scene point cloud to obtain high-quality point cloud information; S3: Construct the scene through the RANSAC algorithm and the minimum project square method, obtain the point cloud information of the scene, calculate the center of mass of the project point cloud and define it as the operation target; S4: The Monte Carlo algorithm is used to calculate the approach workspace of the manipulator, and the operating points of the flexible workspace are obtained through inverse kinematics solution. The flexible workspace is sampled and its spatial envelope is calculated; S5: Make the maximum intersection between the flexible workspace and the object point cloud, calculate the position transformation of the mobile robot according to the maximum intersection, and guide the point cloud of the object to be operated into the flexible workspace; S6: According to the optimal posture transformation information, the mobile structure is controlled to move to the optimal operation space position, so that the flexible workspace of the robot arm and the point cloud information of the feature scene are optimally coupled, thereby achieving a high success rate of operation.

[0028] As a further solution of the present invention, the point cloud optimization algorithm in step S2 includes suppression and noise removal processing to remove irrelevant point clouds, improve calculation efficiency, and generate a maximum spatial envelope diagram.

[0029] As a further solution of the present invention, the scene plane in step S3 includes obtaining a plane equation using a RANSAC algorithm, and performing vertical calculation on the algorithm support of the plane as a Z axis to segment the scene plane and the project point cloud.

[0030] As a further solution of the present invention, the flexible workspace calculation in step S4 includes: Monte Carlo sampling of the robot's neighboring workspace; Monte Carlo sampling is performed on the adjacent workspace of the robot arm. The position of the operating point is fixed, and it is rotated around the x, y, and z axes in steps of 30°. For each posture, it is verified whether the inverse kinematics solution exists. The solution is the operating point of the flexible workspace.

[0031] As a further solution of the present invention, the posture transformation in step S5 includes: Based on the calculation of the object point cloud centroid, guide the object point cloud centroid to move along the preset path to the center point of the robot arm base coordinate system; Adjust the project point cloud layout according to the optimal intersection, so that the project can fully enter the flexible workspace.

[0032] As a further solution of the present invention, step S5 also includes workspace adaptation adjustment. When the project point cloud cannot fully adapt to the flexible workspace, the posture of the mobile robot structure is adjusted to achieve partial intersection of the project point cloud and the robot arm workspace.

[0033] As a further solution of the present invention, step S5 further includes collision parameters and adjustments. When the robot arm moves to the collision parameter limit, it stops moving and adjusts its position to maximize the coverage of the workspace.

[0034] As a further solution of the present invention, the optimized operating space position in step S6 is obtained by inverse posture adjustment calculation, and the adjustment enables the robotic arm to maintain multi-degree-of-freedom flexibility in complex scenes, thereby achieving the best operating effect.

[0035] Embodiment 1: See also Figure 1 and Figure 2 , Figure 1 It is a flow chart of a method for optimizing the operating space of a mobile robot in a complex scene. Figure 2 middle, is the world coordinate system; is the camera coordinate system; is the coordinate system of the car; is the table screen characteristic coordinate system; is the centroid coordinate system of the object point cloud; the method includes: S1: Control the mobile robot to move to the position to be operated.

[0036] The mobile chassis 2 is equipped with a laser radar 1, which is used to operate the mobile robot for navigation through the ROS system, so that it controls the mobile chassis 2 to autonomously avoid obstacles and stop in front of the feature scene 6. In this process, its position has a certain deviation from the ideal operation position due to the accumulated error of laser navigation and the drift of the inertial navigation unit. Within this deviation, the field of view of the depth camera 4 at the end of the robot arm 3 can be guaranteed to include the feature scene 6 and the object 7 to be operated on it.

[0037] S2: Obtain the point cloud information of the operation plane and the objects to be operated in the scene.

[0038] The depth camera 4 is used to collect the point cloud information of some feature scenes 6 and the objects to be operated 7, and the point cloud information is filtered and denoised using the point cloud optimization algorithm to remove irrelevant scene point clouds, improve the calculation efficiency, obtain high-quality point cloud information, and make the maximum spatial envelope diagram of this point cloud information.

[0039] S3. Fit the scene plane and extract the point cloud information of objects above the scene.

[0040] Use RANSAC and the least squares method to fit the feature scene plane 6 information, obtain the fitted plane equation, find the normal vector perpendicular to this plane, define it as the z-axis, and segment the point cloud information on the scene plane 6 where z>0, leaving only the point cloud of the object 7 to be operated.

[0041] S4. Calculate the flexible workspace of the robot arm; First, the Monte Carlo algorithm is used to calculate the reachable workspace of the robot 3, and then the posture of the midpoint position of the reachable workspace is adjusted to verify whether its inverse kinematics has a solution. The solution is the operating point of the flexible workspace. After the flexible workspace is sampled, its spatial envelope diagram is obtained.

[0042] S5. Make the maximum intersection of the flexible workspace and the object point cloud, and obtain the position transformation of the mobile robot according to the maximum intersection; The point cloud of the object 7 to be operated is transformed by offsetting from the initial position toward the direction of the flexible working space of the robot arm 3, and when the point cloud just completely enters the flexible working space, it is defined as the optimal intersection.

[0043] Point cloud centroid calculation and posture guidance: Based on the point cloud data of the object to be grasped, the object's center of mass is calculated and moved along a predetermined path toward the center point of the robot's base coordinate system. During the movement, the pose of the object's point cloud is adjusted so that it gradually enters the robot's flexible workspace. When it just fits into the flexible workspace, its pose transformation is calculated, and this pose transformation is the optimal pose transformation.

[0044] Workspace adaptive adjustment: When the point cloud of the object to be grasped is too large to fit completely into the flexible workspace of the robot arm, the chassis position of the mobile robot is adjusted to the maximum extent possible. The point cloud of the object is partially brought into the workspace of the robot arm by moving the chassis of the mobile robot along a predetermined path.

[0045] Collision parameter limit and stop: When the chassis of the robot moves close to the preset collision parameter limit, the movement of the chassis stops. At this time, the intersection of the object point cloud and the robot workspace reaches the maximum, ensuring that the robot can flexibly grasp the object while avoiding collision; Assume that the point cloud pose is adjusted: Assume that the point cloud center of mass of the object to be grasped moves to the robot base coordinate system along a predetermined path, and gradually guide the point cloud into the flexible workspace of the robot.

[0046] Inverse pose adjustment: Based on the assumed point cloud pose adjustment results, the inverse pose transformation that the robot chassis needs to perform is calculated. Through reverse translation and rotation operations, the robot chassis is moved so that the robot arm can approach the object while maintaining flexible freedom.

[0047] Collision parameter limit determination: When performing reverse posture adjustment, the distance between the robot arm and the surrounding environment is detected in real time. When the mobile robot chassis is close to the preset limit value of the collision parameter, further movement is stopped. At this time, it is ensured that the robot arm can maximize the coverage of the object point cloud and achieve the grasping operation under the premise of avoiding collision.

[0048] S6. Control the mobile chassis 2 to move to the optimal operating space position according to the posture transformation information.

[0049] According to the inverse transformation posture information, the mobile chassis 2 moves to the optimal grasping candidate point, on which the flexible workspace of the robot arm 3 and the point cloud information of the feature scene 6 are optimally coupled, thereby realizing multi-degree-of-freedom and high-success rate operations.

[0050] Among them, the key points of the present invention are: S1: Control the mobile robot to move to the position to be operated.

[0051] The mobile chassis 2 is equipped with a laser radar 1, which is used to operate the mobile robot for navigation through the ROS system, so that it controls the mobile chassis 2 to autonomously avoid obstacles and stop in front of the feature scene 6. In this process, its position has a certain deviation from the ideal operation position due to the accumulated error of laser navigation and the drift of the inertial navigation unit. Within this deviation, the field of view of the depth camera 4 at the end of the robot arm 3 can be guaranteed to include the feature scene 6 and the object 7 to be operated on it.

[0052] In this step, a depth camera is introduced for target recognition and positioning to solve the cumulative error and inertial navigation unit drift problems in the mobile chassis navigation process, which improves the accuracy and stability of mobile chassis navigation, ensures that the robotic arm can accurately stop near the target position, and provides a reliable basis for subsequent operations.

[0053] S2: Obtain the point cloud information of the operation plane and the objects to be operated in the scene.

[0054] The depth camera 4 is used to collect the point cloud information of some feature scenes 6 and the objects to be operated 7, and the point cloud information is filtered and denoised using the point cloud optimization algorithm to remove irrelevant scene point clouds, improve the calculation efficiency, obtain high-quality point cloud information, and make the maximum spatial envelope diagram of this point cloud information.

[0055] By filtering and denoising the point cloud data, the scene features are effectively extracted, and making the maximum spatial envelope diagram also saves computing resources.

[0056] S3. Fit the scene plane and extract the point cloud information of objects above the scene.

[0057] Use RANSAC and the least squares method to fit the feature scene plane 6 information, obtain the fitted plane equation, find the normal vector perpendicular to this plane, define it as the z-axis, segment the point cloud information on the scene plane 6 where z>0, leaving only the point cloud of the object 7 to be operated, and use the point cloud processing algorithm to extract its center of mass.

[0058] S4. Computing the Flexible Workspace of the Robot Arm First, the Monte Carlo algorithm is used to calculate the reachable workspace of the robot 3, and then the posture of the midpoint position of the reachable workspace is adjusted to verify whether its inverse kinematics has a solution. The solution is the operating point of the flexible workspace. After the flexible workspace is sampled, its spatial envelope diagram is obtained.

[0059] S5. Make the maximum intersection between the flexible workspace and the object point cloud, and calculate the position transformation of the mobile robot based on the maximum intersection The point cloud of the object 7 to be operated is transformed by offsetting from the initial position toward the direction of the flexible working space of the robot arm 3, and when the point cloud just completely enters the flexible working space, it is defined as the optimal intersection.

[0060] Specifically in S4-S5: The first expression: Point cloud centroid calculation and posture guidance: Based on the point cloud data of the object to be grasped, the object's center of mass is calculated and moved along a predetermined path toward the center point of the robot's base coordinate system. During the movement, the pose of the object's point cloud is adjusted so that it gradually enters the robot's flexible workspace. When it just fits into the flexible workspace, its pose transformation is calculated, and this pose transformation is the optimal pose transformation.

[0061] Workspace adaptive adjustment: When the point cloud of the object to be grasped is too large to fit completely into the flexible workspace of the robot arm, the chassis position of the mobile robot is adjusted to the maximum extent possible. The point cloud of the object is partially brought into the workspace of the robot arm by moving the chassis of the mobile robot along a predetermined path.

[0062] Collision parameter limit and stop: When the chassis of the robot moves close to the preset collision parameter limit, the chassis movement operation is stopped. At this time, the intersection of the object point cloud and the robot workspace reaches the maximum, ensuring that the robot can flexibly grasp the object while avoiding collision. The second expression: Assume that the point cloud pose is adjusted: Assume that the point cloud center of mass of the object to be grasped moves to the robot base coordinate system along a predetermined path, and gradually guide the point cloud into the flexible workspace of the robot.

[0063] Inverse pose adjustment: Based on the assumed point cloud pose adjustment results, the inverse pose transformation that the robot chassis needs to perform is calculated. Through reverse translation and rotation operations, the robot chassis is moved so that the robot arm can approach the object while maintaining flexible freedom.

[0064] Collision parameter limit determination: When performing reverse posture adjustment, the distance between the robot arm and the surrounding environment is detected in real time. When the mobile robot chassis is close to the preset limit value of the collision parameter, further movement is stopped. At this time, it is ensured that the robot arm can maximize the coverage of the object point cloud and achieve the grasping operation under the premise of avoiding collision.

[0065] S6. Control the mobile chassis 2 to move to the optimal operating space position according to the posture transformation information.

[0066] According to the inverse transformation posture information, the mobile chassis 2 moves to the optimal grasping candidate point, on which the flexible workspace of the robot arm 3 and the point cloud information of the feature scene 6 are optimally coupled, thereby realizing multi-degree-of-freedom and high-success rate operations.

[0067] The present invention adopts the above-mentioned method for optimizing the operating space of mobile robots in complex scenes. By processing scene point cloud information, obtaining the flexible working space of the robotic arm, and then performing space optimization, it realizes the efficient positioning operation of the mobile robot in complex stacking scenes. It has the advantages of multi-degree-of-freedom operation, high-quality feature extraction, precise posture calculation and high success rate operation.

[0068] Embodiment 2: In the field of automated warehousing and logistics, mobile robots are often used for object grabbing and handling operations in complex scenarios. However, in the face of irregular stacking, dense or misplaced objects, traditional path planning and static operation methods cannot effectively cope with them, resulting in a high grab failure rate and low efficiency. This embodiment, combined with the above technical solution, specifically introduces how to achieve efficient and accurate object grabbing operations in such complex scenarios by dynamically optimizing the operating space.

[0069] In a typical automated warehouse environment, the robot needs to grab multiple randomly stacked packages of different sizes and in chaotic locations. Traditional robot arm operation methods are difficult to work effectively in such a complex scenario. In order to overcome these difficulties, this embodiment adopts a dynamic optimization method of flexible operation space, so that the robot arm can flexibly adjust in irregular stacking scenarios to achieve efficient grabbing.

[0070] First, the laser radar and depth camera on the mobile robot are used to obtain real-time point cloud data of the area to be operated. The data collected by the depth camera is filtered and denoised to remove irrelevant background interference. The optimized point cloud information clearly shows the outline and relative position of the stacked objects. This process not only improves the quality of the point cloud, but also significantly reduces the complexity of data processing, laying a solid foundation for subsequent operations.

[0071] Next, the scene is plane-fitted using the RANSAC algorithm and the least squares method to accurately identify the surface features of the manipulated object. The calculated center of mass of the object point cloud is used as the target position for grasping. Unlike the limitations of traditional methods that rely on fixed path grasping, this embodiment dynamically adjusts the position of the robot arm through a center of mass guidance strategy, so that the object point cloud gradually adjusts its position within the flexible workspace of the robot arm to ensure that the object is in the optimal grasping position. In this process, the Monte Carlo algorithm is used to sample the reachable workspace of the robot arm, and the reachability of each operating point is verified through inverse kinematics solution to ensure that the robot arm can smoothly reach the target in multiple degrees of freedom.

[0072] In order to avoid possible collision risks during the grasping process, this embodiment further introduces a real-time adjustment mechanism for collision parameters. During operation, the robot arm continuously detects its relative position with the surrounding environment and adjusts the grasping path and posture in real time. In particular, when the robot arm moves to the preset collision parameter limit, the system can immediately stop the adjustment and recalculate the optimal grasping path and posture to ensure the safety of the grasping process.

[0073] Through the above steps, this embodiment effectively overcomes the traditional problems of limited operating space, insufficient grasping accuracy and poor obstacle avoidance of the robot arm in complex scenes. In this automated warehousing environment, the operation success rate of the robot arm is significantly improved, avoiding operational failures caused by static path planning. At the same time, the innovative application of dynamically adjusting the workspace makes the system no longer rely on expensive high-precision sensors, and can achieve accuracy and efficiency that are difficult to achieve with traditional methods through intelligent algorithm optimization, reducing overall costs.

[0074] In summary, this embodiment demonstrates the superiority of the present invention in complex scenarios. By flexibly adjusting the workspace, optimizing the grasping path in real time, and using an intelligent obstacle avoidance mechanism, it achieves efficient and precise operations that are difficult to achieve with traditional technologies, and provides a new and significantly improved solution for application areas such as automated warehousing and logistics.

[0075] Embodiment 3: In industrial production lines, robots are often used for complex assembly and inspection tasks, especially in narrow, heavily obstructed or multi-object stacked working environments. Traditional operation methods cannot guarantee efficient and accurate operation results. In the prior art, robotic arms usually rely on teaching operations or static path planning, and cannot adaptively adjust the workspace, resulting in limited operational flexibility and success rate. To solve this problem, this embodiment proposes an operation space optimization method in complex scenarios, combining multi-sensors and dynamic posture adjustment strategies to achieve flexible and efficient industrial robot operation.

[0076] In the assembly area of ​​a production line, parts of different shapes and sizes need to be grasped, positioned, and assembled. In this scenario, parts are stacked in disorder and there are obstructions, which makes it difficult for traditional path planning and fixed operation strategies to cope with it. To this end, the robot is equipped with a lidar and a depth camera. By acquiring point cloud data of the working area in real time and combining it with an autonomous navigation system, the robot can accurately locate the position to be operated. The combination of lidar and depth camera enables the system to maintain high-precision scene perception under environmental interference, providing reliable data support for subsequent operations.

[0077] First, the robot collects point cloud information of the object to be operated through a depth camera, and uses a point cloud optimization algorithm to perform denoising and filtering to remove irrelevant scene data. The optimized high-quality point cloud information generates a maximum spatial envelope diagram, laying the foundation for subsequent workspace optimization. Subsequently, the RANSAC algorithm and the least squares method are used to fit the operation plane of the feature scene, and the center of mass information of the object is extracted. This center of mass information is used to guide the adjustment of the robot's operating space and achieve efficient matching between the object and the robot.

[0078] During the operation, this embodiment places special emphasis on the calculation and optimization of the flexible workspace. The reachable workspace of the robot is calculated through the Monte Carlo algorithm, and the sampling points are verified by inverse kinematics to determine the reachability and flexibility of each operation point. The workspace of the robot is maximized by the intersection with the object point cloud, and the position and posture are transformed according to the intersection result. By continuously adjusting the position and posture of the object point cloud, it is ensured that it gradually enters the optimal grasping area of ​​the robot, that is, the optimal intersection area, which significantly improves the success rate of the operation.

[0079] To prevent collision risks during operation, the system uses real-time collision parameter limitation and adjustment strategies. When the robot arm is moving, the system detects the distance between the robot arm and the surrounding environment, stops moving when it approaches the preset collision limit, and readjusts the grasping path and posture to ensure safe grasping. When facing larger parts, the robot chassis can achieve flexible chassis movement and posture correction through reverse posture adjustment to adapt to scene changes and object sizes to the greatest extent.

[0080] Through the optimization method of this embodiment, the robot's operating efficiency and grasping success rate in complex assembly scenarios are significantly improved. This method not only overcomes the limitations of traditional teaching methods, but also realizes adaptive operation of various forms of parts through the fusion and real-time adjustment of multiple algorithms. Especially when dealing with irregular, heavily obscured or complexly stacked objects, the robot can flexibly adjust the operating space to achieve fast and accurate grasping and assembly, significantly improving the overall efficiency and flexibility of industrial production lines.

[0081] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, the embodiments should be considered exemplary and non-restrictive in all respects, and the scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims be included in the present invention.

Claims

1. A method for dynamically optimizing the flexible operation space of a robot in a complex scene, which is characterized by the following steps: S1: Control the mobile robot to move to the position to be operated. The mobile robot includes a mobile chassis equipped with a laser radar, a visual system based on laser radar navigation and a depth camera, and autonomously avoids obstacles through the ROS system and moves to a feature scene and stops; S2: The point cloud information of the feature scene and the item to be operated is obtained through the depth camera, and the object point cloud information is suppressed and denoised to remove the irrelevant scene point cloud to obtain high-quality point cloud information; S3: Construct the scene through the RANSAC algorithm and the minimum project square method, obtain the point cloud information of the scene, calculate the center of mass of the project point cloud and define it as the operation target; S4: The Monte Carlo algorithm is used to calculate the approach workspace of the manipulator, and the operating points of the flexible workspace are obtained through inverse kinematics solution. The flexible workspace is sampled and its spatial envelope is calculated; S5: Make the maximum intersection between the flexible workspace and the object point cloud, calculate the position transformation of the mobile robot according to the maximum intersection, and guide the point cloud of the object to be operated into the flexible workspace; S6: According to the optimal posture transformation information, the mobile structure is controlled to move to the optimal operation space position, so that the flexible workspace of the robot arm and the point cloud information of the feature scene are optimally coupled, thereby achieving a high success rate of operation.

2. According to the method of dynamic optimization of robot flexible operation space in complex scenes in claim 1, it is characterized in that: The point cloud optimization algorithm in step S2 includes suppression and noise removal processing to remove irrelevant point clouds, improve computational efficiency, and generate a maximum spatial envelope diagram.

3. According to the method for dynamic optimization of robot flexible operation space in complex scenes in claim 1, it is characterized in that: The scene plane in step S3 includes obtaining a plane equation using a RANSAC algorithm, and performing a vertical calculation on the algorithm support of the plane as a Z axis to segment the scene plane and the project point cloud.

4. According to the method for dynamic optimization of robot flexible operation space in complex scenes in claim 1, it is characterized in that: The flexible workspace calculation in step S4 includes: Monte Carlo sampling of the robot's neighboring workspace; Monte Carlo sampling is performed on the adjacent workspace of the robot arm. The position of the operating point is fixed, and it is rotated around the x, y, and z axes in steps of 30°. For each posture, it is verified whether the inverse kinematics solution exists. The solution is the operating point of the flexible workspace.

5. The method for dynamic optimization of robot flexible operation space in complex scenarios according to claim 1 is characterized in that: The posture transformation in step S5 includes: Based on the calculation of the object point cloud centroid, guide the object point cloud centroid to move along the preset path to the center point of the robot arm base coordinate system; Adjust the project point cloud layout according to the optimal intersection, so that the project can fully enter the flexible workspace.

6. The method for dynamic optimization of robot flexible operation space in complex scenarios according to claim 1 is characterized in that: The step S5 also includes workspace adaptation adjustment. When the project point cloud cannot fully adapt to the flexible workspace, the position and posture of the mobile robot structure are adjusted to achieve partial intersection of the project point cloud and the workspace of the robot arm.

7. The method for dynamic optimization of robot flexible operation space in complex scenarios according to claim 1 is characterized in that: The step S5 further includes collision parameters and adjustments. When the robot arm moves to the collision parameter limit, it stops moving and adjusts its position to maximize the coverage of the workspace.

8. The method for dynamic optimization of robot flexible operation space in complex scenarios according to claim 1 is characterized in that: The optimal operating space position in step S6 is calculated by inverse posture adjustment, and the adjustment enables the robot arm to maintain multi-degree-of-freedom flexibility in complex scenes, thereby achieving the best operating effect.

Images