CN-121989244-A - Joint angle inverse solution method, device, equipment and medium for mechanical arm

CN121989244ACN 121989244 ACN121989244 ACN 121989244ACN-121989244-A

Abstract

The application provides a joint angle inverse solution method, device and equipment of a mechanical arm and a medium, wherein the method comprises the steps of determining multiple groups of candidate joint position data on the mechanical arm from a preset joint position data set according to a target grabbing position, wherein the preset joint position data set stores multiple groups of joint position data on the mechanical arm, each group of joint position data comprises a tail end position and angles corresponding to multiple joints, respectively constructing local linear models for the multiple joints according to the multiple groups of candidate joint position data to obtain local linear models of the multiple joints, reversely predicting the joints according to the target grabbing position by adopting the local linear models of the joints, determining angle estimation values of the joints, obtaining tail end forward prediction positions of the mechanical arm according to the angle estimation values of the multiple joints, and determining the to-be-controlled angles of the target grabbing position corresponding to the joints according to the tail end forward prediction positions, the target grabbing position and the angle estimation values of the joints.

Inventors

TIAN ZHIWEI
ZHANG FACHAO
TANG JIAN
XIONG YOUJUN

Assignees

北京人形机器人创新中心有限公司

Dates

Publication Date: 20260508
Application Date: 20260304

Claims (10)

1. A method for joint angle inverse solution of a mechanical arm, the method comprising: Determining multiple groups of candidate joint position data on the mechanical arm from a preset joint position data set according to the target grabbing position, wherein the preset joint position data set stores multiple groups of joint position data on the mechanical arm, and each group of joint position data comprises an end position and angles corresponding to multiple joints; respectively constructing local linear models for a plurality of joints according to a plurality of groups of candidate joint position data to obtain local linear models of the joints; according to the target grabbing positions, adopting a local linear model of each joint to reversely predict each joint, and determining an angle estimated value of each joint; Acquiring forward prediction positions of the tail ends of the mechanical arms according to the angle estimation values of the joints; And determining the angle to be controlled of each joint corresponding to the target grabbing position according to the forward predicted position of the tail end, the target grabbing position and the angle estimated value of each joint.
2. The method of claim 1, wherein prior to determining the plurality of sets of candidate joint position data on the robotic arm from the set of preset joint position data based on the target capture position, the method further comprises: acquiring the tail end positions of the mechanical arm in a plurality of limit postures and angles corresponding to the joints; according to the tail end positions under the multiple limit postures and angles corresponding to the joints, respectively constructing global linear models for the multiple joints to obtain global linear models of the multiple joints; generating a plurality of point cloud data in a working space range of the mechanical arm by adopting a global linear model of a plurality of joints, wherein each point cloud data comprises one tail end position and angles of a plurality of corresponding joints; and constructing the preset joint position data set according to the plurality of point cloud data.
3. The method of claim 2, wherein the generating a plurality of point cloud data within the workspace of the robotic arm using a global linear model of a plurality of the joints comprises: Randomly generating a plurality of sampling positions in a working space range of the mechanical arm; According to each sampling position, carrying out inverse prediction on each joint by adopting a global linear model of a plurality of joints, and determining an angle estimated value of each joint corresponding to each sampling position; according to the angle estimated value of each joint corresponding to each sampling position, carrying out terminal forward position prediction to obtain the terminal position corresponding to each sampling position; And taking a data pair consisting of the tail end position and angle estimated values corresponding to a plurality of joints as one point cloud data.
4. The method of claim 1, wherein determining a plurality of sets of candidate joint position data on the robotic arm from a preset joint position dataset based on the target capture position comprises: acquiring the distance between the target grabbing position and each tail end position in the preset joint position data set; And determining a plurality of candidate tail end positions according to the distance between the target grabbing position and each tail end position and the preset quantity, and determining the joint position data corresponding to the candidate tail end positions as a plurality of groups of candidate joint position data.
5. The method of claim 1, wherein the determining the target capture position corresponding to the angle to be controlled for each of the joints based on the tip forward predicted position, the target capture position, and the angle estimate for each of the joints comprises: acquiring the position deviation of the forward predicted position of the tail end and the target grabbing position; and determining the angle to be controlled of each joint according to the position deviation and the angle estimated value of each joint.
6. The method of claim 5, wherein determining the angle to be controlled for each of the joints based on the positional deviation and the angle estimate for each of the joints comprises: if the position deviation is smaller than or equal to a preset deviation threshold value, determining an angle estimated value of each joint as the angle to be controlled of each joint.
7. The method of claim 5, wherein determining the angle to be controlled for each of the joints based on the positional deviation and the angle estimate for each of the joints comprises: If the position deviation is larger than a preset deviation threshold, determining a new target grabbing position according to the tail end forward predicted position and the target grabbing position; and re-determining the angle to be controlled of each joint corresponding to the new target grabbing position according to the new target grabbing position.
8. A joint angle inverse solution device for a mechanical arm, the device comprising: the determining module is used for determining a plurality of groups of candidate joint position data on the mechanical arm from a preset joint position data set according to the target grabbing position, wherein the preset joint position data set stores the plurality of groups of joint position data on the mechanical arm, and each group of joint position data comprises a tail end position and angles corresponding to a plurality of joints; The construction module is used for respectively constructing local linear models for a plurality of joints according to a plurality of groups of candidate joint position data to obtain local linear models of the joints; The prediction module is used for reversely predicting each joint by adopting a local linear model of each joint according to the target grabbing position to determine an angle estimated value of each joint; the acquisition module is used for acquiring forward prediction positions of the tail ends of the mechanical arms according to the angle estimation values of the joints; The determining module is further configured to determine, according to the forward predicted position of the end, the target capturing position, and the angle estimation value of each joint, a to-be-controlled angle of each joint corresponding to the target capturing position.
9. A control device comprising a processor, a storage medium and a bus, wherein the storage medium stores program instructions executable by the processor, the processor and the storage medium communicate via the bus when the control device is in operation, and the processor executes the program instructions to perform the steps of the joint angle inverse solution method of the mechanical arm according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that the storage medium has stored thereon a computer program which, when executed by a processor, performs the steps of the joint angle inverse solution method of a robotic arm as claimed in any one of claims 1 to 7.

Description

Joint angle inverse solution method, device, equipment and medium for mechanical arm Technical Field The invention relates to the technical field of mechanical arms, in particular to a joint angle inverse solution method, device, equipment and medium of a mechanical arm. Background Inverse kinematics (INVERSE KINEMATICS, IK) is a core technology in the fields of robot control, human motion simulation and the like, and the core aim is to determine angle parameters corresponding to each joint of the system when the end effector reaches the expected position and posture. In practical application, redundant degree-of-freedom systems such as a 7-degree-of-freedom mechanical arm and a human arm are very common, and the number of degrees of freedom of the system is larger than the degree of freedom of motion required by an end effector, so that for the same end expected position, infinite groups of joint angle combinations exist, namely, the problem of redundant solution can be realized. The existence of the redundant solution brings challenges to system control, and how to screen unique, natural and stable joint angles from infinite groups of solutions becomes a key pain point for solving redundant inverse kinematics, and the motion precision of an end effector, the system operation stability and the reasonability of motion gestures are directly influenced. At present, various solutions are proposed in the industry for solving the redundancy solution problem of redundancy inverse kinematics, but various methods have obvious limitations, and the requirements of efficient, automatic and high-quality solution in practical application are difficult to meet. The method is a method which is widely applied, typically comprises a jacobian pseudo-inverse method, a gradient projection method and the like, the method is used for screening redundant solutions by setting optimal targets such as minimum joint motions and the like, the solution results are solved and are seriously dependent on guesses of initial joint angles, the solution is easy to converge to an unnatural and unstable local optimal solution, and in the calculation process, singular problems are easy to occur, solving precision and efficiency are influenced, the analysis rule is used for realizing uniqueness of the solution by introducing additional constraint parameters such as arm angles and the like, the method is low in automation degree, selection of the constraint parameters lacks visual basis and is difficult to adapt to complex dynamic scenes, the sample library is used for solving the solution by matching samples based on the method of example/data driving, time and labor are wasted in the construction process of the sample library, the solution quality is completely dependent on the coverage and quality of the samples, the problem of the solution precision decline easily occurs in a sample sparse area, the global function approximation method is used for realizing the unique solution through the mapping relation between the end positions and the joint angles, the method requires training data, and has the defect of 'black box' is difficult to generate the consistency and nature style. In summary, the prior art has not formed a redundant inverse kinematics solving method which does not depend on huge data samples, can automatically ensure the uniqueness and naturalness of the solution, and has high-efficiency computing performance. Disclosure of Invention The invention aims to provide a joint angle inverse solution method, device, equipment and medium for a mechanical arm, aiming at the defects in the prior art, so as to realize high-precision online solution of redundant degree-of-freedom mechanical arm inverse kinematics by screening candidate data from a preset joint position data set, constructing a local linear model for each joint, inversely predicting joint angle estimation values and combining forward prediction position iterative optimization to determine a complete set of flow of angles to be controlled. In order to achieve the above purpose, the technical scheme adopted by the embodiment of the application is as follows: In a first aspect, an embodiment of the present application provides a joint angle inverse solution method of a mechanical arm, where the method includes: Determining multiple groups of candidate joint position data on the mechanical arm from a preset joint position data set according to the target grabbing position, wherein the preset joint position data set stores multiple groups of joint position data on the mechanical arm, and each group of joint position data comprises an end position and angles corresponding to multiple joints; respectively constructing local linear models for a plurality of joints according to a plurality of groups of candidate joint position data to obtain local linear models of the joints; according to the target grabbing positions, adopting a local linear model of each joint to reversely predict e