A method and system for obstacle avoidance path planning for a multi-degree-of-freedom robotic arm
By constructing a resultant force guiding field and a bidirectional random expansion tree, the problems of low efficiency and stability in obstacle avoidance path planning of robotic arms in complex 3D scenes are solved, generating smooth and safe obstacle avoidance motion paths suitable for tasks such as assembly, grinding, and handling.
Patent Information
- Application Number
- CN202610462497.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-09
- Publication Date
- 2026-06-26
AI Technical Summary
Existing obstacle avoidance path planning methods for robotic arms suffer from problems such as strong random sampling blindness, low search efficiency, uneven paths, and easy getting trapped in local minima in complex 3D scenes, making it difficult to balance planning efficiency, path safety, and motion executability.
Combining the artificial potential field method with a fast-expanding random tree, path planning is performed by constructing a resultant force guiding field. Orientation constraints are applied using the target's attraction and the obstacle's repulsion to generate a pre-path point sequence. This sequence is then mapped to the joint space via inverse kinematics. Finally, a bidirectional random expansion tree is used to repair the gaps to be repaired, generating the final obstacle avoidance path.
It improves path generation efficiency, enhances the adaptability and obstacle avoidance stability of the robotic arm in complex environments, and generates smooth and safe obstacle avoidance motion paths, which are suitable for complex tasks such as assembly, grinding, and handling.
Smart Images

Figure CN122274970A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of robot motion planning technology, and in particular to a method and system for obstacle avoidance path planning for a multi-degree-of-freedom robotic arm. Background Technology
[0002] Against the backdrop of rapid advancements in industry and intelligent manufacturing, digital twin technology provides crucial support for motion simulation and online control of robotic arms in complex 3D scenarios by constructing real-time mappings between physical devices and virtual spaces. As the types of obstacles and the complexity of tasks in work scenarios continue to increase, the obstacle avoidance path planning problem for robotic arms in virtual environments becomes increasingly prominent. Currently used obstacle avoidance planning methods for robotic arms often employ strategies such as pure fast expanding random trees, probabilistic route maps, or grid search. While these methods possess some global search capabilities, they generally suffer from strong random sampling blindness, low search efficiency in high-dimensional joint spaces, redundancy expansion in complex obstacle environments, and unsmooth path generation. Furthermore, using artificial potential fields alone for path planning can easily lead to getting trapped in local minima, causing the robotic arm to stagnate, oscillate, or even fail to reach the target position near obstacles. These shortcomings make it difficult for existing methods to balance planning efficiency, path safety, and motion executability in digital twin scenarios, hindering online path planning and virtual-physical synchronous control in complex tasks such as assembly, handling, and grinding. Summary of the Invention
[0003] To address the aforementioned shortcomings, the present invention aims to propose a multi-degree-of-freedom robotic arm obstacle avoidance path planning method and system. This method combines the environmental guidance capability of the artificial potential field method with the global search capability of a rapidly expanding random tree. It utilizes target attraction and obstacle repulsion to directionally constrain the sampling expansion process, reducing blind searching and improving path generation efficiency in complex spaces. Simultaneously, it enhances the robotic arm's adaptability and obstacle avoidance stability in scenarios with uneven obstacle distribution and narrow passages, thereby providing efficient and reliable path planning support for the safe movement of the robotic arm during assembly, grinding, and handling operations.
[0004] To achieve this objective, the present invention adopts the following technical solution: A method for obstacle avoidance path planning for a multi-degree-of-freedom robotic arm includes: Acquire target point and obstacle information in the robotic arm's working environment, and construct a resultant force guiding field composed of the superposition of attractive and repulsive potential fields; Along the gradient direction of the resultant force guiding field, an iterative search is performed from the starting point to the target point to generate a pre-path point sequence, and the generated pre-path point sequence is mapped to the joint space through inverse kinematics to obtain a candidate configuration sequence. The local path segments between adjacent configurations in the candidate configuration sequence are sequentially checked to see if they meet the collision constraints. If they do not meet the constraints, the adjacent configurations are determined to be gaps to be repaired. A bidirectional random expansion tree is established using the candidate configurations at both ends of the gap to be repaired as the starting and ending points, respectively, and the expansion step size is adjusted in real time according to the potential energy intensity of the repulsive potential field in the current region. The gaps to be repaired are repaired by connecting the bidirectional random expansion tree, and the repaired path segments are spliced with the path segments that satisfy the collision constraints to generate the final obstacle avoidance path.
[0005] Preferably, the construction of the resultant force guiding field formed by the superposition of the attractive potential field and the repulsive potential field includes: Calculate the attractive potential energy between the current position and the target point to obtain the attractive potential energy exerted by the target point on the end effector of the robotic arm. The following relation is satisfied: ; in, This indicates the current position coordinates of the robotic arm's end effector. Indicates the position coordinates of the target point. This indicates the preset attraction coefficient; The obstacles in the scene are fitted as cylindrical models, and the distance from the current position to the [missing value] is calculated. Minimum distance to obstacles The following relation is satisfied: ; in, Indicates the coordinates of the center of the base of the cylinder. This represents the unit vector pointing from the bottom surface of the cylinder to the top surface; Calculate the repulsive potential energy of a single obstacle based on the minimum distance. and the total repulsive potential energy of all obstacles The following relation is satisfied: ; ; in, Indicates the first The repulsive potential energy generated by the obstacle This represents the preset repulsion coefficient. Indicates the threshold value representing the effective range of an obstacle. Indicates by The total repulsive potential energy generated by all obstacles; The attractive force is obtained by calculating the gradients of the attractive potential energy and the total repulsive potential energy respectively. and repulsive force And the total guiding force is obtained by superposition. The following relation is satisfied: ; ; ; in, Represents the gradient operator. This represents the direction vector of the resultant force guiding field used to guide the movement of the robotic arm.
[0006] Preferably, the iterative search along the gradient direction of the resultant force guiding field from the starting point to the target point to generate a pre-path point sequence includes: Starting from the initial position, the path points are iteratively updated based on the direction vector of the resultant force guiding field, resulting in the updated path point coordinates. The following relation is satisfied: ; in, This represents the updated path point coordinates. This represents the coordinates of the path point in the current iteration step. Indicates the preset workspace step size. This represents the total guiding force vector at the current path point. This represents a preset minimum positive number used to prevent the denominator from being zero; The potential field state is monitored in real time during the iteration process. When it is determined that the system is trapped in a local minimum region, an obstacle avoidance strategy based on cylindrical obstacles is initiated, and the obstacle avoidance vector is calculated. The following relation is satisfied: ; in, Represents a symbolic function. Represents the position vector of the target point. This represents the current position vector of the robotic arm's end effector. The unit vector representing the axis of the cylindrical obstacle. This represents the radial unit vector pointing from the cylinder axis to the current point. Indicates the tangential direction of the cylindrical surface used to navigate around the obstacle; The obstacle avoidance vector is used to guide the end of the robotic arm to bypass the obstacle along the tangential direction to escape the local minimum and continue to advance towards the target point until the distance between the path point and the target point is less than the preset target tolerance, thus completing the generation of the pre-path point sequence.
[0007] Preferably, the step of mapping the generated pre-path point sequence to the joint space through inverse kinematics to obtain a candidate configuration sequence includes: A DH parameter table is established for the 6R robot, and the total transformation matrix is obtained by multiplying the transformation matrices of each joint. Transformation matrix of a single joint Satisfying the relation: ; in, Indicates the first The joint relative to the first Transformation matrix of each joint, Indicates the first The joint angles of each joint. Indicates the length of the link. Indicates the link twist angle. Indicates the link offset; Calculate the wrist center coordinates of the robotic arm The following relation is satisfied: ; in, This represents the vector indicating the center position of the wrist. Represents the end target position vector. Represents the end-target attitude matrix. This indicates the offset of the robotic arm's end effector; The joint angles of the first three joints are calculated based on the wrist center coordinates, where the first joint angle... Satisfying the relation: ; in, Indicates the first joint angle. This indicates the position coordinates of the wrist center in the base coordinate system; The second joint angle is calculated by using the projection distance of the wrist center onto the horizontal plane and the distance from the origin to the wrist center. With the third joint angle The following relation is satisfied: ; ; in, Indicates the center of the wrist in the base coordinate system Projected distance on the plane This represents the distance from the origin of the base coordinate system to the center of the wrist. This indicates the effective length of the first link of the robotic arm. This indicates the effective length of the second link of the robotic arm; Calculate the rotation matrices of the last three joints based on the target rotation matrix and the rotation matrices of the first three joints. The following relation is satisfied: ; in, This represents the rotation matrix of the last three joints relative to the ends of the first three joints. This represents the transpose of the total rotation matrix of the first three joints. Represents the target rotation matrix; The multiple joint space configurations calculated by the rotation matrix are traversed, and wide-phase collision detection and narrow-phase collision detection are performed in sequence. The joint space configurations that do not collide are determined as feasible candidate configurations.
[0008] Preferably, the step of sequentially detecting whether the local path segments between adjacent configurations in the candidate configuration sequence satisfy collision constraints, and determining the adjacent configurations as gaps to be repaired if they do not, includes: For two adjacent joint space configurations in the candidate configuration sequence, interpolation is performed between the two adjacent joint space configurations according to a preset step size to generate several intermediate joint configurations. Collision detection is performed on the intermediate joint configurations. If all the intermediate joint configurations satisfy the no-collision constraint, then it is determined that the two adjacent joint spatial configurations are directly connected. If any of the intermediate joint configurations does not meet the no-collision constraint, then it is determined that there is an obstacle interference between the two adjacent joint spatial configurations, and the two adjacent joint spatial configurations are determined to be gaps to be repaired.
[0009] Preferably, the step of establishing a bidirectional random expansion tree using the candidate configurations at both ends of the gap to be repaired as the start and end points, respectively, includes: The random sampling area is restricted, limiting the search range of the bidirectional random expansion tree to the local area to which the adjacent candidate configurations that generate the gap to be repaired belong; Using one candidate configuration at one end of the gap to be repaired as the starting point of the first search tree and the other candidate configuration as the ending point of the second search tree, the local obstacle avoidance path connecting the gap to be repaired is searched by expanding the two trees to random sampling points in turn and approaching each other.
[0010] Preferably, the step of adjusting the expansion step size in real time according to the potential energy intensity of the repulsive potential field in the current region includes: Calculate the first The magnitude of the repulsive force at each prepath point The following relation is satisfied: ; in, Indicates the first The magnitude of the repulsive force at each prepath point Indicates the first Prepath points The vector of the repulsive force acting at the location; For each of the two adjacent pre-path points corresponding to the gap to be repaired, the average value of their repulsive forces is calculated as the repulsive strength of the local segment, satisfying the following relationship: ; in, This represents the average repulsion strength between two adjacent prepath points. and These represent the magnitudes of the repulsive forces at two adjacent prepath points; The average repulsion strength is normalized to obtain the normalized repulsion strength. The following relation is satisfied: ; in, and These represent the maximum and minimum values of the repulsion strength among all local segments, respectively. This represents a preset minimum positive number used to prevent the denominator from being zero; The adaptive expansion step size is determined based on the normalized repulsion strength. The following relation is satisfied: ; in, This represents the adaptive expansion step size of the bidirectional random expansion tree in a local region. This indicates the preset maximum expansion step size. This indicates the preset minimum expansion step size.
[0011] A multi-degree-of-freedom robotic arm obstacle avoidance path planning system includes: The field construction module is used to acquire target point and obstacle information of the robotic arm's working environment and construct a resultant force guiding field composed of the superposition of attractive and repulsive potential fields. The path extraction module is used to iteratively search from the starting point to the target point along the gradient direction of the resultant force guiding field to generate a pre-path point sequence, and to map the generated pre-path point sequence to the joint space through inverse kinematics to obtain a candidate configuration sequence. The gap determination module is used to sequentially detect whether the local path segments between adjacent configurations in the candidate configuration sequence meet the collision constraints. If they do not meet the constraints, the adjacent configurations are determined to be gaps to be repaired. The local search module is used to establish a bidirectional random expansion tree with the candidate configurations at both ends of the gap to be repaired as the starting point and the ending point, respectively, and to adjust the expansion step size in real time according to the potential energy intensity of the repulsive potential field in the current region. The path splicing module is used to repair the gap to be repaired by connecting the bidirectional random expansion tree, and splice the repaired path segment with the path segment that meets the collision constraints to generate the final obstacle avoidance motion path.
[0012] One of the above technical solutions has the following advantages or beneficial effects: This invention constructs a resultant force guiding field composed of an attractive potential field and a repulsive potential field, enabling the robotic arm to obtain directional guidance pointing to the target point and avoiding obstacles, thus overcoming the blindness of traditional random sampling methods. By iteratively searching along the gradient direction of the resultant force guiding field to generate a pre-path point sequence, and using inverse kinematics to map it to the joint space to obtain a candidate configuration sequence, it effectively transfers environmental guidance information from the workspace to the high-dimensional joint space, avoiding the inefficiency caused by direct unguided search in the joint space. By sequentially detecting whether the local path segments between adjacent configurations in the candidate configuration sequence satisfy collision constraints and identifying gaps to be repaired, it accurately locates local areas requiring further fine-tuning, reducing the need for global unguided searches. The system performs efficient calculations. By establishing a bidirectional random expansion tree with candidate configurations at both ends of the gap to be repaired as the starting and ending points, respectively, and adjusting the expansion step size in real time according to the potential energy intensity of the repulsive potential field in the current region, it can automatically reduce the step size for fine exploration in areas with dense obstacles or narrow passages, and increase the step size to improve the search speed in open areas. This effectively solves the problem of difficulty in passing through narrow passages caused by fixed step size. At the same time, the bidirectional search strategy significantly improves the probability of finding feasible paths. Finally, the gap to be repaired is repaired by connecting the bidirectional random expansion trees, and the repaired path segments are spliced with the path segments that meet the collision constraints. This achieves an effective fusion of local fine search and global guidance, generating a complete obstacle avoidance path that is both smooth and safe. Attached Figure Description
[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0014] Figure 1 This is a flowchart of the obstacle avoidance path planning method for a multi-degree-of-freedom robotic arm provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the multi-degree-of-freedom robotic arm obstacle avoidance path planning system provided in an embodiment of the present invention. Detailed Implementation
[0015] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.
[0016] In this invention, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0017] A method for obstacle avoidance path planning for a multi-degree-of-freedom robotic arm, such as Figure 1 As shown, a preferred embodiment of the present invention includes the following steps: S1: Obtain target point and obstacle information in the robotic arm's working environment, and construct a resultant force guiding field composed of the superposition of attractive and repulsive potential fields; It should be noted that the target point refers to the preset position coordinates that the robotic arm's end effector needs to reach in the working environment, usually represented by coordinate values in three-dimensional space, such as a position vector in a base coordinate system. Obstacle information includes the geometry, position, orientation, and dimensional parameters of all obstacles in the working environment. In this scheme, obstacles are fitted as cylindrical models and mathematically represented by the coordinates of the base center and the direction vector of the axis. The attractive potential field refers to the virtual potential energy field constructed around the target point, whose potential energy value decreases as the distance between the robotic arm's end effector and the target point decreases, forming an attractive force pointing towards the target point. The repulsive potential field refers to the virtual potential energy field constructed around the obstacle, generating a repulsive effect when the robotic arm's end effector enters the obstacle's range of action; its potential energy value increases sharply as the distance from the obstacle decreases. The resultant force guiding field is the comprehensive potential field formed by vector superposition of the attractive and repulsive potential fields, used to provide directional guidance for the robotic arm's end effector that simultaneously considers target approach and obstacle avoidance.
[0018] Understandably, the purpose of constructing the combined force guiding field is to provide directional constraints at the environmental perception level for subsequent path search. By introducing the physical mechanism of target attraction and obstacle repulsion, the originally completely random search process is transformed into an exploration process with clear guidance. This step, by constructing a low-dimensional artificial potential field in the workspace (i.e., the three-dimensional physical space where the robotic arm's end effector actually moves), can effectively overcome the problem of low search efficiency caused by blindly sampling randomly in the joint space (i.e., the high-dimensional abstract space composed of the angles of each joint of the robotic arm). The attractive potential field ensures that the robotic arm is always guided by a force pointing towards the target, while the repulsive potential field ensures an avoidance tendency when approaching obstacles. The superposition of the two enables the robotic arm to obtain continuous path guidance information in complex obstacle environments, thereby reducing invalid search areas and improving the initial guidance accuracy of path planning.
[0019] S2: Along the gradient direction of the resultant force guiding field, iteratively search from the starting point to the target point to generate a pre-path point sequence, and map the generated pre-path point sequence to the joint space through inverse kinematics to obtain a candidate configuration sequence; It should be noted that the gradient direction refers to the direction in which the rate of change of the potential field function is greatest at a certain point in space. Moving along the gradient direction allows the robotic arm's end effector to approach the target point along the path of the fastest descent of the potential field, while avoiding obstacles. The pre-path point sequence refers to the set of path points generated in the workspace, consisting of discrete coordinate points. These points are distributed along the gradient direction of the resultant force guiding field, forming a preliminary path framework from the starting point to the target point. Inverse kinematics refers to the mathematical process of solving for the angle values of each joint of the robotic arm based on the target position and orientation of the end effector in the workspace, converting workspace coordinates into joint space coordinates. Joint space refers to the abstract multidimensional space composed of all joint angle variables of the robotic arm, with each dimension corresponding to the rotation angle or movement distance of a joint. The candidate configuration sequence refers to the set of several combinations of joint angles obtained through inverse kinematics. Each combination of joint angles is called a configuration, representing a possible orientation of the robotic arm.
[0020] Understandably, the purpose of iterative searching along the gradient direction of the resultant force guiding field is to transform continuous 3D environmental guidance information into a discrete sequence of path points. By performing low-dimensional path planning in the workspace, the computational complexity explosion problem caused by blindly searching directly in six-dimensional or higher-dimensional joint spaces is avoided. Mapping pre-path points to joint space through inverse kinematics aims to establish a mapping relationship between the workspace and joint space, enabling the path planning results in the virtual 3D scene to be transformed into joint angle commands required for the actual control of the robotic arm. This step, by first acquiring environmental guidance in the workspace and then converting it to the joint space to verify feasibility, effectively solves the problems of high computational cost of collision detection and difficulty in path visualization in high-dimensional joint spaces, while ensuring the physical executability of the generated path.
[0021] S3: Sequentially check whether the local path segments between adjacent configurations in the candidate configuration sequence meet the collision constraints. If they do not meet the constraints, then determine that the adjacent configurations are gaps to be repaired. It should be noted that adjacent configurations in the candidate configuration sequence refer to two joint space configurations that are adjacent in position within the sequence, corresponding to two adjacent pre-path points in the workspace. A local path segment refers to the path interval connecting two adjacent joint space configurations, typically represented by a series of intermediate configurations obtained through interpolation between these two configurations. Collision constraints refer to the safety conditions that prevent the robotic arm from physically interfering with any obstacles during its movement, including geometric intersection constraints between the robotic arm links and obstacles. A gap to be repaired refers to a situation in the candidate configuration sequence where at least one intermediate configuration in the local path segment between two adjacent configurations does not satisfy the collision constraint; that is, these two configurations cannot be directly connected by simple straight-line interpolation and require additional path search for repair. Interpolation processing refers to the process of generating intermediate transition configurations between two joint space configurations according to a preset step size, used to verify the feasibility of the path between the two points.
[0022] Understandably, the purpose of sequentially detecting local path segments between adjacent configurations is to accurately identify which adjacent points in the candidate configuration sequence are directly connected and which are interfered with by obstacles, thereby decomposing the global path planning problem into a feasibility verification problem of multiple local path segments. This step effectively narrows the scope of the local area requiring fine-grained search by transforming the originally complex global path search into a segment-by-segment verification of the pre-path point sequence, avoiding unnecessary computational overhead in obstacle-free areas. The role of determining the gaps to be repaired is to mark the breakpoint locations in the path, providing clear start and end point constraints for the subsequent bidirectional random expansion tree search, ensuring that the random search is focused only on local areas with interference from obstacles, thus significantly improving the computational efficiency of the overall path planning.
[0023] S4: Establish a bidirectional random expansion tree with the candidate configurations at both ends of the gap to be repaired as the starting point and the ending point, respectively, and adjust the expansion step size in real time according to the potential energy intensity of the repulsive potential field in the current region. It should be noted that bidirectional random expansion tree refers to simultaneously constructing two search trees in joint space from both ends of the gap to be repaired. One tree expands from the starting point towards a random sampling point, and the other expands from the ending point towards a random sampling point. Feasible paths are found by approximating each other between the two trees. The starting point and ending point refer to the candidate configurations at both ends of the gap to be repaired, with one end serving as the root node of the first tree and the other as the root node of the second tree. The expansion step size refers to the joint space distance extended from the existing tree node towards the random sampling point during the growth of the random expansion tree. The size of the step size directly affects the granularity of the search and computational efficiency. The potential energy intensity of the repulsive potential field refers to the magnitude of the repulsive force vector at the pre-path point, used to characterize the density of obstacles or the narrowness of the passage near that location. The larger the potential energy intensity, the stronger the environmental constraints. Adaptive adjustment refers to the mechanism of dynamically changing the expansion step size according to the complexity of the environment. In areas with dense obstacles, the step size is reduced for fine-grained searching, while in open areas, the step size is increased to improve search speed.
[0024] Understandably, the purpose of establishing a bidirectional random expansion tree is to restrict the global random search to a local area within the gap to be repaired. By having two trees simultaneously approach the middle from both ends, compared to a unidirectional search, it effectively increases the probability of finding a feasible path and reduces search time. Adjusting the expansion step size in real-time based on the repulsive potential energy intensity allows the random expansion process to adapt to the spatial characteristics of the local environment. Smaller step sizes are used in narrow areas with high potential energy to avoid collision risks caused by large step sizes, ensuring passage through narrow passages; larger step sizes are used in open areas with low potential energy to reduce unnecessary fine-grained searches and improve search efficiency. This step, by incorporating environmentally perceptual information (repulsive potential energy intensity) into the random sampling process, enables the algorithm to adapt to the environment, effectively solving the dilemma of fixed-step random expansion trees either being too slow in complex obstacle environments or unable to pass through narrow passages.
[0025] S5: Repair the gap to be repaired by connecting the bidirectional random expansion tree, and then splice the repaired path segment with the path segment that satisfies the collision constraint to generate the final obstacle avoidance path.
[0026] It should be noted that the interconnection of the bidirectional random expansion tree refers to the merging of the two trees during the expansion process when a new node of one tree enters the neighborhood of the other tree by attempting to directly connect the nodes of the two trees. Repairing the gap to be repaired refers to finding feasible local paths connecting the candidate configurations at both ends of the gap through a bidirectional random expansion tree search, filling the breakpoints in the original candidate configuration sequence. The repaired path segment refers to the collision-free local path connecting the two ends of the gap to be repaired, obtained through a bidirectional random expansion tree search. Path segments satisfying collision constraints refer to local paths that have passed verification in step S3 and can be directly connected between adjacent candidate configurations. Path splicing refers to sequentially combining all directly connected path segments satisfying collision constraints with the local path segments repaired by the bidirectional random expansion tree to form a complete and continuous path from the starting point to the target point. The final obstacle avoidance motion path refers to the motion trajectory generated through the above process, satisfying collision-free constraints, and available for actual execution by the robotic arm.
[0027] Understandably, the purpose of repairing gaps by connecting bidirectional random expansion trees is to leverage the flexibility of random search to address local obstacle interference problems. While maintaining the global path skeleton (pre-path point sequence) largely unchanged, a refined path search is performed on local areas with obstacle interference, ensuring the overall path's connectivity and feasibility. The purpose of splicing the repaired path segments with those satisfying collision constraints is to integrate local search results with global pre-planning results, generating a complete path that conforms to both environmental guidance (along the pre-path point distribution) and collision-free constraints. This step, combining local refined search with a global guiding skeleton, effectively balances global optimization and local feasibility in path planning, avoiding the problems of either low global search efficiency or getting trapped in local minima found in traditional methods. This ensures the generated path is both safe and executable in complex 3D environments.
[0028] Preferably, the construction of the resultant force guiding field formed by the superposition of the attractive potential field and the repulsive potential field includes: Calculate the attractive potential energy between the current position and the target point to obtain the attractive potential energy exerted by the target point on the end effector of the robotic arm. The following relation is satisfied: ; in, This indicates the current position coordinates of the robotic arm's end effector. Indicates the position coordinates of the target point. This indicates the preset attraction coefficient; The obstacles in the scene are fitted as cylindrical models, and the distance from the current position to the [missing value] is calculated. Minimum distance to obstacles The following relation is satisfied: ; in, Indicates the coordinates of the center of the base of the cylinder. This represents the unit vector pointing from the bottom surface of the cylinder to the top surface; Calculate the repulsive potential energy of a single obstacle based on the minimum distance. and the total repulsive potential energy of all obstacles The following relation is satisfied: ; ; in, Indicates the first The repulsive potential energy generated by the obstacle This represents the preset repulsion coefficient. Indicates the threshold value representing the effective range of an obstacle. Indicates by The total repulsive potential energy generated by all obstacles; The attractive force is obtained by calculating the gradients of the attractive potential energy and the total repulsive potential energy respectively. and repulsive force And the total guiding force is obtained by superposition. The following relation is satisfied: ; ; ; in, Represents the gradient operator. This represents the direction vector of the resultant force guiding field used to guide the movement of the robotic arm.
[0029] It should be noted that attractive potential energy refers to the virtual potential energy generated by the target point on the end effector of the robotic arm. Its magnitude is related to the spatial distance between the end effector and the target point; the greater the distance, the greater the potential energy, forming an attractive force pointing towards the target point. Repulsive potential energy refers to the virtual potential energy generated by an obstacle on the end effector of the robotic arm. When the end effector enters the range of the obstacle, a repulsive effect is generated; the closer to the obstacle, the greater the potential energy. The cylindrical model simplifies the actual irregular shape of the obstacle into a cylindrical geometry for mathematical description, defining the position and orientation of the cylinder in space through the coordinates of the center of the base and the direction vector of the axis. The minimum distance refers to the shortest distance from the current position of the end effector of the robotic arm to the surface of the cylindrical obstacle. Spatial distance is used to quantify the proximity of the robotic arm to obstacles; the effective range threshold refers to the preset radius of influence of the obstacle. When the distance between the robotic arm's end effector and the obstacle is less than this threshold, the repulsive potential energy begins to take effect, and when it is greater than this threshold, the repulsive potential energy is zero; the attraction coefficient and repulsion coefficient are proportional parameters used to adjust the potential field strength, determining the strength of the potential energy's influence on the robotic arm's movement; the gradient refers to the rate of change vector of the potential field function in space, pointing in the direction of the fastest increase in potential energy, obtained by calculating the partial derivative of the potential energy with respect to spatial coordinates; the total guiding force refers to the vector superposition of the attraction and repulsion forces, providing the robotic arm's end effector with comprehensive motion direction guidance that simultaneously considers target approach and obstacle avoidance.
[0030] Understandably, the purpose of calculating and superimposing the attractive and repulsive potential energies to obtain the total guiding force is to quantify the attraction of the target point and the repulsion of obstacles into a computable mathematical force field, enabling the robotic arm to obtain continuous and differentiable directional guidance in complex obstacle environments. By fitting the actual obstacles to a cylindrical model, the distance calculation problem for complex geometries is effectively simplified, and the calculation of the minimum distance and repulsive potential energy has analytical expressions, improving computational efficiency. By introducing an effective range threshold, unnecessary weak influences from distant obstacles on the robotic arm's movement are avoided, reducing computational overhead. By calculating the gradients of the attractive and repulsive potential energies separately, the scalar potential field is transformed into a vector force field, providing a directly usable direction vector for subsequent path point iteration updates. This environment modeling method based on artificial potential fields effectively overcomes the blindness of traditional path planning methods due to the lack of environmental guidance information, enabling the robotic arm to actively perceive the target position and obstacle distribution, thereby achieving directional path search in complex 3D scenes.
[0031] Preferably, the iterative search along the gradient direction of the resultant force guiding field from the starting point to the target point to generate a pre-path point sequence includes: Starting from the initial position, the path points are iteratively updated based on the direction vector of the resultant force guiding field, resulting in the updated path point coordinates. The following relation is satisfied: ; in, This represents the updated path point coordinates. This represents the coordinates of the path point in the current iteration step. Indicates the preset workspace step size. This represents the total guiding force vector at the current path point. This represents a preset minimum positive number used to prevent the denominator from being zero; The potential field state is monitored in real time during the iteration process. When it is determined that the system is trapped in a local minimum region, an obstacle avoidance strategy based on cylindrical obstacles is initiated, and the obstacle avoidance vector is calculated. The following relation is satisfied: ; in, Represents a symbolic function. Represents the position vector of the target point. This represents the current position vector of the robotic arm's end effector. The unit vector representing the axis of the cylindrical obstacle. This represents the radial unit vector pointing from the cylinder axis to the current point. Indicates the tangential direction of the cylindrical surface used to navigate around the obstacle; The obstacle avoidance vector is used to guide the end of the robotic arm to bypass the obstacle along the tangential direction to escape the local minimum and continue to advance towards the target point until the distance between the path point and the target point is less than the preset target tolerance, thus completing the generation of the pre-path point sequence.
[0032] It should be noted that the workspace step size refers to the distance the robotic arm's end effector moves in each iteration along the direction of the resultant force guiding field in three-dimensional physical space, determining the spatial interval between pre-path points; the minimum positive number refers to a constant close to zero introduced in mathematical calculations to prevent numerical errors caused by a zero denominator in division operations; the local minimum region refers to a non-target point location in the artificial potential field where the attractive and repulsive forces are balanced or canceled out, resulting in a zero or minimum resultant force, causing the robotic arm's end effector to stagnate and unable to continue moving towards the target point; obstacle avoidance strategy. This refers to a special motion pattern where, when a robotic arm is trapped in a local minimum, it changes its direction of motion to bypass the obstacle along the tangent direction of the obstacle. The obstacle-avoidance vector is a tangential unit vector calculated under local minimum conditions, used to guide the robotic arm to move along the tangent direction of the obstacle surface to escape the equilibrium state. The sign function is a mathematical function that returns 1, -1, or 0 based on the positive or negative value of the input value, used to determine the obstacle-avoidance direction. The tangential direction is a direction vector perpendicular to the radial direction of the cylinder and located on the surface of the cylinder; moving along this direction allows the end of the robotic arm to bypass the obstacle.
[0033] Understandably, the purpose of iterative search along the gradient direction of the resultant force-guided field is to transform continuous potential field information into a discrete sequence of path points. By performing low-dimensional path planning in the workspace, the computational complexity explosion problem caused by blindly searching in six-dimensional or higher-dimensional joint space is avoided. Iterative updates by setting a workspace step size enable the robotic arm's end effector to gradually approach the target point along the direction of the fastest potential field descent, while avoiding obstacles. Introducing a minimum positive number prevents the denominator from being zero, ensuring the stability of numerical calculations. Real-time monitoring of local minima and initiation of obstacle avoidance strategies effectively solves the inherent defect of traditional artificial potential field methods, which are prone to getting trapped in local minima and causing planning failures. By calculating the tangential obstacle avoidance vector based on the cylindrical geometry, the robotic arm can select the side more advantageous for approaching the target along the tangential direction of the obstacle surface to bypass the obstacle, continuously advancing towards the target point until reaching the vicinity of the target position and completing the generation of the pre-path point sequence.
[0034] Preferably, the step of mapping the generated pre-path point sequence to the joint space through inverse kinematics to obtain a candidate configuration sequence includes: A DH parameter table is established for the 6R robot, and the total transformation matrix is obtained by multiplying the transformation matrices of each joint. Transformation matrix of a single joint Satisfying the relation: ; in, Indicates the first The joint relative to the first Transformation matrix of each joint, Indicates the first The joint angles of each joint. Indicates the length of the link. Indicates the link twist angle. Indicates the link offset; Calculate the wrist center coordinates of the robotic arm The following relation is satisfied: ; in, This represents the vector indicating the center position of the wrist. Represents the end target position vector. Represents the end-target attitude matrix. This indicates the offset of the robotic arm's end effector; The joint angles of the first three joints are calculated based on the wrist center coordinates, where the first joint angle... Satisfying the relation: ; in, Indicates the first joint angle. This indicates the position coordinates of the wrist center in the base coordinate system; The second joint angle is calculated by using the projection distance of the wrist center onto the horizontal plane and the distance from the origin to the wrist center. With the third joint angle The following relation is satisfied: ; ; in, Indicates the center of the wrist in the base coordinate system Projected distance on the plane This represents the distance from the origin of the base coordinate system to the center of the wrist. This indicates the effective length of the first link of the robotic arm. This indicates the effective length of the second link of the robotic arm; Calculate the rotation matrices of the last three joints based on the target rotation matrix and the rotation matrices of the first three joints. The following relation is satisfied: ; in, This represents the rotation matrix of the last three joints relative to the ends of the first three joints. This represents the transpose of the total rotation matrix of the first three joints. Represents the target rotation matrix; The multiple joint space configurations calculated by the rotation matrix are traversed, and wide-phase collision detection and narrow-phase collision detection are performed in sequence. The joint space configurations that do not collide are determined as feasible candidate configurations.
[0035] It should be noted that the DH parameter table refers to the Denavit-Hartenberg parameter table, a standardized method for describing the relative position and attitude relationships between the coordinate systems of a robotic arm's links. It defines the transformation relationship between adjacent links using four parameters: link length, link torsion angle, link offset, and joint angle. The transformation matrix is a 4×4 homogeneous matrix describing the rotation and translation relationship between one coordinate system and another; cascaded transformations between coordinate systems can be achieved through matrix multiplication. The wrist center refers to the connection point between the first three and last three joints of a six-DOF robotic arm, serving as the rotation center when the end effector's attitude changes. Analytical solution refers to directly solving the analytical expression of a system of nonlinear equations using algebraic operations to obtain explicit solutions for the joint variables, unlike numerical iterative solutions. Wide-phase collision detection is a detection method that uses coarse geometry such as bounding boxes for rapid collision screening, used to quickly eliminate obviously non-collision cases. Narrow-phase collision detection is a detailed detection method that performs precise geometric intersection on candidate object pairs that pass wide-phase detection, used to accurately determine whether the robotic arm collides with an obstacle.
[0036] Understandably, the purpose of mapping the pre-path point sequence to joint space through inverse kinematics is to establish a mathematical mapping relationship between the workspace and joint space, transforming the path planning results in the three-dimensional physical space into high-dimensional joint angle commands required for the actual control of the robotic arm. Since the actual motion control of a multi-degree-of-freedom robotic arm needs to be based on joint space coordinates, it is necessary to inversely map the position coordinates in the workspace to the rotation angles of each joint. By establishing a DH parameter table and calculating the transformation matrix, the geometric relationships between the links of the robotic arm can be systematically described. By calculating the wrist center coordinates and solving them analytically, the complex inverse kinematics problem can be decomposed into two relatively independent sub-problems: solving the position of the first three joints and solving the posture of the last three joints, reducing the difficulty of the solution. Through a two-stage detection strategy of wide-phase collision detection and narrow-phase collision detection, feasible joint space configurations can be quickly screened, ensuring that the generated candidate configuration sequence not only meets the end-effector position requirements but also satisfies the collision-free constraint, thereby guaranteeing the executability and safety of subsequent path planning.
[0037] Specifically, the implementation of inverse kinematics mapping includes: first, establishing a DH parameter table for the six-DOF rotary joint manipulator, defining the link length, link twist angle, link offset, and joint angle parameters for each joint, and constructing a 4×4 homogeneous transformation matrix for a single joint, which contains a 3×3 rotation matrix and a 3×1 translation vector; multiplying the transformation matrices of each joint sequentially to obtain the total transformation matrix, which describes the complete pose of the manipulator's end effector relative to the base coordinate system; calculating the wrist center coordinates, and obtaining the three-dimensional coordinates of the wrist center in the base coordinate system through vector operations based on the end effector target position and pose matrix and the end effector offset; and calculating the first three joint angles based on the wrist center coordinates, with the first joint angle calculated using the projection coordinates of the wrist center onto the horizontal plane. The distances from the wrist center to the origin of the base coordinate system and the geometric lengths of each link are obtained, and the second and third joint angles are solved using trigonometric functions and the law of cosines. The relative rotation matrices of the last three joints are calculated based on the target posture matrix and the rotation matrices of the first three joints, and the last three joint angles are solved by comparing the matrix elements. All possible joint angle combinations are traversed. First, a wide-phase collision detection is performed to determine whether the bounding box of the robot arm overlaps with the bounding box of the obstacle. If an overlap occurs, the objects under its bounding box are selected as candidate object pairs. These object pairs are traversed and narrow-phase collision detection is performed to perform precise geometric intersection. If no collision detection occurs, the joint space configuration of the robot arm is a feasible candidate configuration. All feasible configurations are arranged in order to form a candidate configuration sequence.
[0038] Preferably, the step of sequentially detecting whether the local path segments between adjacent configurations in the candidate configuration sequence satisfy collision constraints, and determining the adjacent configurations as gaps to be repaired if they do not, includes: For two adjacent joint space configurations in the candidate configuration sequence, interpolation is performed between the two adjacent joint space configurations according to a preset step size to generate several intermediate joint configurations. Collision detection is performed on the intermediate joint configurations. If all the intermediate joint configurations satisfy the no-collision constraint, then it is determined that the two adjacent joint spatial configurations are directly connected. If any of the intermediate joint configurations does not meet the no-collision constraint, then it is determined that there is an obstacle interference between the two adjacent joint spatial configurations, and the two adjacent joint spatial configurations are determined to be gaps to be repaired.
[0039] It should be noted that interpolation refers to the process of generating an intermediate transition configuration between two joint spatial configurations according to a preset step size, used to verify the feasibility of the path between the two points; intermediate joint configuration refers to the transitional joint angle combination generated between two adjacent candidate configurations through the interpolation algorithm, representing the intermediate posture of the robotic arm when moving between two points; direct connectivity means that two adjacent joint spatial configurations can be connected without collision through a simple interpolation path, without the need for additional complex path search; obstacle interference refers to the situation where the robotic arm physically overlaps with an obstacle or the distance is less than a safety threshold during the movement, making the local path infeasible.
[0040] Understandably, the purpose of sequentially detecting local path segments between adjacent configurations in the candidate configuration sequence is to accurately identify which adjacent points in the pre-path point sequence are directly connected in joint space and which are interfered with by obstacles, thereby decomposing the global path planning problem into a feasibility verification problem of multiple local path segments. Interpolation processing generates intermediate joint configurations, which discretizes the continuous motion process, facilitating collision detection. By interpolating between adjacent configurations and detecting all intermediate configurations, the continuity and collision-free nature of the local path segment can be comprehensively verified. Determining the gaps to be repaired marks the breakpoints in the path, ensuring that subsequent random expansion tree searches only target local areas with obstacle interference, avoiding unnecessary computational overhead in obstacle-free areas. This concentrates computational resources on truly difficult areas requiring fine-grained searching, significantly improving the computational efficiency of the overall path planning.
[0041] Preferably, the step of establishing a bidirectional random expansion tree using the candidate configurations at both ends of the gap to be repaired as the start and end points, respectively, includes: The random sampling area is restricted, limiting the search range of the bidirectional random expansion tree to the local area to which the adjacent candidate configurations that generate the gap to be repaired belong; Using one candidate configuration at one end of the gap to be repaired as the starting point of the first search tree and the other candidate configuration as the ending point of the second search tree, the local obstacle avoidance path connecting the gap to be repaired is searched by expanding the two trees to random sampling points in turn and approaching each other.
[0042] It should be noted that bidirectional random expansion of the tree refers to simultaneously constructing two search trees in the joint space from both ends of the gap to be repaired. One tree expands from the starting point towards the random sampling point, and the other expands from the ending point towards the random sampling point. Feasible paths are found by the two trees approximating each other. The random sampling region refers to the joint space range in which random sampling is performed. Limiting this region can avoid invalid sampling in the global joint space. The local region refers to the joint space neighborhood range corresponding to the adjacent pre-path points that generate the gap to be repaired, which is usually much smaller than the global joint space. Alternating expansion of the two trees refers to the strategy of alternating node expansion of the first tree and the second tree, which improves search efficiency by approximating each other. Approximation of each other means that the two trees grow from both ends towards the middle, attempting to meet and connect at a certain point in the middle to form a complete path.
[0043] Understandably, the purpose of establishing a bidirectional random expansion tree with the candidate configurations at both ends of the gap to be repaired as the starting and ending points, respectively, is to restrict the global random search to a local region within the gap. By simultaneously approaching the middle from both ends, compared to a unidirectional search, this effectively increases the probability of finding a feasible path and reduces search time. By limiting the random sampling region to the local region of the adjacent candidate configurations that generated the gap, invalid sampling in open areas far from obstacles is avoided, concentrating random sampling within the neighborhood where feasible paths may appear, significantly improving sampling effectiveness and search efficiency. By having the two trees expand alternately and approach each other, they can meet in the middle region of the joint space. Once the distance between the nodes of the two trees is less than the connection threshold, a path can be connected. This bidirectional search strategy effectively solves the problem of unidirectional random expansion trees struggling to find feasible paths in narrow passages, while significantly reducing the number of nodes required for the search and the computation time.
[0044] Preferably, the step of adjusting the expansion step size in real time according to the potential energy intensity of the repulsive potential field in the current region includes: Calculate the first The magnitude of the repulsive force at each prepath point The following relation is satisfied: ; in, Indicates the first The magnitude of the repulsive force at each prepath point Indicates the first Prepath points The vector of the repulsive force acting at the location; For each of the two adjacent pre-path points corresponding to the gap to be repaired, the average value of their repulsive forces is calculated as the repulsive strength of the local segment, satisfying the following relationship: ; in, This represents the average repulsion strength between two adjacent prepath points. and These represent the magnitudes of the repulsive forces at two adjacent prepath points; The average repulsion strength is normalized to obtain the normalized repulsion strength. The following relation is satisfied: ; in, and These represent the maximum and minimum values of the repulsion strength among all local segments, respectively. This represents a preset minimum positive number used to prevent the denominator from being zero; The adaptive expansion step size is determined based on the normalized repulsion strength. The following relation is satisfied: ; in, This represents the adaptive expansion step size of the bidirectional random expansion tree in a local region. This indicates the preset maximum expansion step size. This indicates the preset minimum expansion step size.
[0045] It should be noted that the magnitude of the repulsion force refers to the modulus of the repulsion force vector at the pre-path point, used to quantify the density of obstacles or the narrowness of the passage near that location; the local segment refers to the path interval between two adjacent pre-path points corresponding to the gap to be repaired; the average repulsion intensity refers to the arithmetic mean of the repulsion force magnitudes of two adjacent pre-path points, used to characterize the overall environmental constraint strength of the local segment; normalization processing refers to the standardization process of mapping the repulsion intensity of different numerical ranges to a range between 0 and 1, eliminating the influence of differences in the numerical range of repulsion force under different scenarios on the step size adjustment; the adaptive expansion step size refers to the expansion step size that dynamically changes according to the complexity of the environment, automatically decreasing in areas with dense obstacles and automatically increasing in open areas.
[0046] Understandably, adjusting the expansion step size in real time based on the potential energy intensity of the repulsive potential field in the current region aims to enable the random expansion process to adaptively respond to the spatial characteristics of the local environment. The magnitude of the repulsive force at the pre-path point reflects the degree of environmental congestion, and the expansion step size is dynamically adjusted accordingly. In areas with high repulsive force, it indicates that the robotic arm is approaching an obstacle or is in a narrow passage. In this case, reducing the expansion step size for a finer search can effectively avoid the collision risk caused by large step sizes and ensure passage through narrow passages. In areas with low repulsive force, it indicates that the environment is relatively open. In this case, increasing the expansion step size can improve the search speed and reduce unnecessary fine searches. By calculating and normalizing the average repulsive intensity of local segments, the influence of differences in the range of repulsive force values under different scenarios on the step size adjustment effect is effectively eliminated, giving the step size adjustment cross-scenario versatility and stability. This adaptive mechanism effectively solves the dilemma of fixed-step random expansion trees in complex obstacle environments, where either the search is too slow or the passage through narrow passages cannot be completed.
[0047] A multi-degree-of-freedom robotic arm obstacle avoidance path planning system, such as Figure 2 As shown, it includes: The field construction module is used to acquire target point and obstacle information of the robotic arm's working environment and construct a resultant force guiding field composed of the superposition of attractive and repulsive potential fields. The path extraction module is used to iteratively search from the starting point to the target point along the gradient direction of the resultant force guiding field to generate a pre-path point sequence, and to map the generated pre-path point sequence to the joint space through inverse kinematics to obtain a candidate configuration sequence. The gap determination module is used to sequentially detect whether the local path segments between adjacent configurations in the candidate configuration sequence meet the collision constraints. If they do not meet the constraints, the adjacent configurations are determined to be gaps to be repaired. The local search module is used to establish a bidirectional random expansion tree with the candidate configurations at both ends of the gap to be repaired as the starting point and the ending point, respectively, and to adjust the expansion step size in real time according to the potential energy intensity of the repulsive potential field in the current region. The path splicing module is used to repair the gap to be repaired by connecting the bidirectional random expansion tree, and splice the repaired path segment with the path segment that meets the collision constraints to generate the final obstacle avoidance motion path.
[0048] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
[0049] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.
Claims
1. A method for obstacle avoidance path planning of a multi-degree-of-freedom robot arm, characterized in that, include: Acquire target point and obstacle information in the robotic arm's working environment, and construct a resultant force guiding field composed of the superposition of attractive and repulsive potential fields; Along the gradient direction of the resultant force guiding field, an iterative search is performed from the starting point to the target point to generate a pre-path point sequence, and the generated pre-path point sequence is mapped to the joint space through inverse kinematics to obtain a candidate configuration sequence. The local path segments between adjacent configurations in the candidate configuration sequence are sequentially checked to see if they meet the collision constraints. If they do not meet the constraints, the adjacent configurations are determined to be gaps to be repaired. A bidirectional random expansion tree is established using the candidate configurations at both ends of the gap to be repaired as the starting and ending points, respectively, and the expansion step size is adjusted in real time according to the potential energy intensity of the repulsive potential field in the current region. The gaps to be repaired are repaired by connecting the bidirectional random expansion tree, and the repaired path segments are spliced with the path segments that satisfy the collision constraints to generate the final obstacle avoidance path.
2. The method of claim 1, wherein, The construction of the resultant force guiding field, which is formed by the superposition of the attractive and repulsive potential fields, includes: Attracting potential energy between the current position and the target point is calculated to obtain attracting potential energy exerted by the target point on the end effector of the robot arm , the relationship is satisfied: ; wherein, represents the current position coordinates of the end effector of the robot arm, represents the position coordinates of the target point, represents a preset attractive coefficient; fitting obstacles in the scene as a cylinder model and calculating the minimum distance from the current position to the first obstacle satisfying the relationship: ; wherein, denotes the center coordinate of the bottom face of the cylinder, denotes the unit vector pointing from the bottom face to the top face of the cylinder; calculating a repulsive potential for a single obstacle based on the minimum distance and a total repulsive potential for all obstacles satisfying the relation: ; ; wherein, represents the repulsive potential generated by the th obstacle, represents a preset repulsive coefficient, represents a threshold of the action range of the obstacle, represents the total repulsive potential generated by the obstacles collectively; The attractive force is obtained by calculating the gradients of the attractive potential energy and the total repulsive potential energy respectively. and repulsive force And the total guiding force is obtained by superposition. The following relation is satisfied: ; ; ; in, Represents the gradient operator. This represents the direction vector of the resultant force guiding field used to guide the movement of the robotic arm.
3. The obstacle avoidance path planning method for a multi-degree-of-freedom robotic arm according to claim 2, characterized in that, The iterative search along the gradient direction of the resultant force guiding field from the starting point to the target point to generate a pre-path point sequence includes: Starting from the initial position, the path points are iteratively updated based on the direction vector of the resultant force guiding field, resulting in the updated path point coordinates. The following relation is satisfied: ; in, This represents the updated path point coordinates. This represents the coordinates of the path point in the current iteration step. Indicates the preset workspace step size. This represents the total guiding force vector at the current path point. This represents a preset minimum positive number used to prevent the denominator from being zero; The potential field state is monitored in real time during the iteration process. When it is determined that the system is trapped in a local minimum region, an obstacle avoidance strategy based on cylindrical obstacles is initiated, and the obstacle avoidance vector is calculated. The following relation is satisfied: ; in, Represents a symbolic function. Represents the position vector of the target point. This represents the current position vector of the robotic arm's end effector. The unit vector representing the axis of the cylindrical obstacle. This represents the radial unit vector pointing from the cylinder axis to the current point. Indicates the tangential direction of the cylindrical surface used to navigate around the obstacle; The obstacle avoidance vector is used to guide the end of the robotic arm to bypass the obstacle along the tangential direction to escape the local minimum and continue to advance towards the target point until the distance between the path point and the target point is less than the preset target tolerance, thus completing the generation of the pre-path point sequence.
4. The obstacle avoidance path planning method for a multi-degree-of-freedom robotic arm according to claim 1, characterized in that, The step of mapping the generated pre-path point sequence to joint space through inverse kinematics to obtain a candidate configuration sequence includes: A DH parameter table is established for the 6R robot, and the total transformation matrix is obtained by multiplying the transformation matrices of each joint. Transformation matrix of a single joint Satisfying the relation: ; in, Indicates the first The joint relative to the first Transformation matrix of each joint, Indicates the first The joint angles of each joint. Indicates the length of the link. Indicates the link twist angle. Indicates the link offset; Calculate the wrist center coordinates of the robotic arm The following relation is satisfied: ; in, This represents the vector indicating the center position of the wrist. Represents the end target position vector. Represents the end-target attitude matrix. This indicates the offset of the robotic arm's end effector; The joint angles of the first three joints are calculated based on the wrist center coordinates, where the first joint angle... Satisfying the relation: ; in, Indicates the first joint angle. This indicates the position coordinates of the wrist center in the base coordinate system; The second joint angle is calculated by using the projection distance of the wrist center onto the horizontal plane and the distance from the origin to the wrist center. With the third joint angle The following relation is satisfied: ; ; in, Indicates the center of the wrist in the base coordinate system Projected distance on the plane This represents the distance from the origin of the base coordinate system to the center of the wrist. This indicates the effective length of the first link of the robotic arm. This indicates the effective length of the second link of the robotic arm; Calculate the rotation matrices of the last three joints based on the target rotation matrix and the rotation matrices of the first three joints. The following relation is satisfied: ; in, This represents the rotation matrix of the last three joints relative to the ends of the first three joints. This represents the transpose of the total rotation matrix of the first three joints. Represents the target rotation matrix; The multiple joint space configurations calculated by the rotation matrix are traversed, and wide-phase collision detection and narrow-phase collision detection are performed in sequence. The joint space configurations that do not collide are determined as feasible candidate configurations.
5. The obstacle avoidance path planning method for a multi-degree-of-freedom robotic arm according to claim 1, characterized in that, The step of sequentially detecting whether the local path segments between adjacent configurations in the candidate configuration sequence satisfy collision constraints, and if not, determining that the adjacent configurations are gaps to be repaired, includes: For two adjacent joint space configurations in the candidate configuration sequence, interpolation is performed between the two adjacent joint space configurations according to a preset step size to generate several intermediate joint configurations. Collision detection is performed on the intermediate joint configurations. If all the intermediate joint configurations satisfy the no-collision constraint, then it is determined that the two adjacent joint spatial configurations are directly connected. If any of the intermediate joint configurations does not meet the no-collision constraint, then it is determined that there is an obstacle interference between the two adjacent joint spatial configurations, and the two adjacent joint spatial configurations are determined to be gaps to be repaired.
6. The obstacle avoidance path planning method for a multi-degree-of-freedom robotic arm according to claim 1, characterized in that, The step of establishing a bidirectional random expansion tree using the candidate configurations at both ends of the gap to be repaired as the starting and ending points, respectively, includes: The random sampling area is restricted, limiting the search range of the bidirectional random expansion tree to the local area to which the adjacent candidate configurations that generate the gap to be repaired belong; Using one candidate configuration at one end of the gap to be repaired as the starting point of the first search tree and the other candidate configuration as the ending point of the second search tree, the local obstacle avoidance path connecting the gap to be repaired is searched by expanding the two trees to random sampling points in turn and approaching each other.
7. The obstacle avoidance path planning method for a multi-degree-of-freedom robotic arm according to claim 1, characterized in that, The step of adjusting the expansion step size in real time based on the potential energy intensity of the repulsive potential field in the current region includes: Calculate the first The magnitude of the repulsive force at each prepath point The following relation is satisfied: ; in, Indicates the first The magnitude of the repulsive force at each prepath point Indicates the first Prepath points The vector of the repulsive force acting at the location; For each of the two adjacent pre-path points corresponding to the gap to be repaired, the average value of their repulsive forces is calculated as the repulsive strength of the local segment, satisfying the following relationship: ; in, This represents the average repulsion strength between two adjacent prepath points. and These represent the magnitudes of the repulsive forces at two adjacent prepath points; The average repulsion strength is normalized to obtain the normalized repulsion strength. The following relation is satisfied: ; in, and These represent the maximum and minimum values of the repulsion strength among all local segments, respectively. This represents a preset minimum positive number used to prevent the denominator from being zero; The adaptive expansion step size is determined based on the normalized repulsion strength. The following relation is satisfied: ; in, This represents the adaptive expansion step size of the bidirectional random expansion tree in a local region. This indicates the preset maximum expansion step size. This indicates the preset minimum expansion step size.
8. A multi-degree-of-freedom robotic arm obstacle avoidance path planning system, characterized in that, include: The field construction module is used to acquire target point and obstacle information of the robotic arm's working environment and construct a resultant force guiding field composed of the superposition of attractive and repulsive potential fields. The path extraction module is used to iteratively search from the starting point to the target point along the gradient direction of the resultant force guiding field to generate a pre-path point sequence, and to map the generated pre-path point sequence to the joint space through inverse kinematics to obtain a candidate configuration sequence. The gap determination module is used to sequentially detect whether the local path segments between adjacent configurations in the candidate configuration sequence meet the collision constraints. If they do not meet the constraints, the adjacent configurations are determined to be gaps to be repaired. The local search module is used to establish a bidirectional random expansion tree with the candidate configurations at both ends of the gap to be repaired as the starting point and the ending point, respectively, and to adjust the expansion step size in real time according to the potential energy intensity of the repulsive potential field in the current region. The path splicing module is used to repair the gap to be repaired by connecting the bidirectional random expansion tree, and splice the repaired path segment with the path segment that meets the collision constraints to generate the final obstacle avoidance motion path.
