CN-122008218-A - Floating target control mechanical arm low-disturbance and noninductive grabbing method based on artificial potential field

CN122008218ACN 122008218 ACN122008218 ACN 122008218ACN-122008218-A

Abstract

The invention particularly relates to a low-disturbance and non-sense grabbing method for a floating target control mechanical arm based on an artificial potential field, and relates to the technical field of robot motion control and intelligent operation. The method comprises the steps of adopting a quintic polynomial to plan an initial track, constructing an artificial potential field comprising a radial potential field and a tangential potential field, generating virtual force, introducing an impedance model to dynamically correct the track, inhibiting contact impact, constructing a super-torsion sliding mode controller based on symmetrical preset performance to realize joint space track tracking, introducing a reinforcement learning algorithm to carry out iterative optimization on control parameters, setting a grabbing detection and re-planning mechanism, and automatically triggering track re-planning when grabbing fails. The invention can realize low contact force grabbing without depending on a force sensor, has the characteristics of high control precision, strong robustness, good autonomy and the like, and is suitable for mechanical arm grabbing tasks under complex environments such as floating targets and the like.

Inventors

MA ZHIQIANG
GAO HONGYI
ZHANG LE
Lang Siyi
DONG HANLIN

Assignees

西北工业大学深圳研究院

Dates

Publication Date: 20260512
Application Date: 20260312

Claims (8)

1. The utility model provides a floating target control mechanical arm low-disturbance noninductive grabbing method based on artificial potential field, which is characterized by comprising the following steps: acquiring an initial pose of an end effector of the mechanical arm and an expected grabbing pose of a target object, and generating an initial planning track of the end effector by adopting a five-time polynomial track planning method; Constructing an artificial potential field according to the relative spatial relation between the end effector and the target object, wherein the artificial potential field comprises a radial potential field and a tangential potential field, and respectively generating a radial potential field force and a tangential potential field force to jointly form an artificial potential field virtual force; Introducing the artificial potential field virtual force into an impedance control model, dynamically correcting an initial planning track, and generating an impedance corrected reference track; the super-torsion sliding mode controller based on the symmetrical preset performance is constructed and used for track tracking control of the joint space of the mechanical arm, so that the mechanical arm moves according to the reference track; Introducing a reinforcement learning algorithm, and performing iterative optimization on control parameters in the super-torsion sliding mode controller; after the execution of each section of planning track is completed, detecting the relative position relation between the end effector and the target object, and if the grabbing failure or the position deviation of the target object is detected, triggering a re-planning mechanism, regenerating the planning track and repeating the steps.
2. The artificial potential field-based low-disturbance and non-inductive grabbing method of the floating target manipulation mechanical arm, which is characterized in that the construction mode of the radial potential field force is as follows: Constructing a radial potential function: ; Wherein, the , In order to obtain the radius of the object block, Is a mechanical arm end the distance between the circle centers of the object blocks, , , , Is a constant defined by human beings; differentiating to obtain radial potential field force: 。
3. The artificial potential field-based low-disturbance and non-feel grabbing method of the floating target control mechanical arm is characterized in that the tangential potential field force is specifically constructed in the following manner: constructing a tangential potential function: ; differentiating to obtain tangential potential field force: ; Wherein, the Is the end of the mechanical arm and the spherical target position Is used for the radial error of (a), Is a constant defined by human beings.
4. The method for grasping the floating target manipulation mechanical arm with low disturbance and no sense based on the artificial potential field according to claim 1, wherein the constructing the super-torsion sliding mode controller based on the symmetrical preset performance comprises the following steps: Constructing a symmetrical preset performance function, and restraining the dynamic evolution process of the joint tracking error; the control law of the ultra-torsion sliding mode is constructed, and the robustness of the system under the condition of modeling uncertainty and external disturbance is improved; in combination with the dynamic compensation term, a complete joint space control law is formed.
5. The artificial potential field-based low-disturbance and noninductive grabbing method of a floating target manipulation mechanical arm, as set forth in claim 4, wherein the super-torsion sliding mode controller is expressed as: Wherein, the In order to be a dynamic compensation term, In order to control the maximum amplitude of the input, Is a space inertia matrix of the joint of the mechanical arm, The Super-Twisting sliding mode control law is constructed.
6. The artificial potential field-based low-disturbance-free grabbing method for the floating target control mechanical arm is characterized in that a Soft Actor-Critic algorithm is adopted in the reinforcement learning algorithm, and control parameters are subjected to online or iterative optimization by constructing a reward function based on track tracking errors, so that parameter self-adaptive adjustment is achieved.
7. The artificial potential field-based low-disturbance and non-perception grabbing method of a floating target manipulation manipulator according to claim 1, wherein the re-planning mechanism comprises: when the distance between the end effector and the target object exceeds a preset threshold value or the target object is pushed out of the original position or the grabbing fails, automatically triggering the track re-planning; and regenerating a planning track according to the current tail end state and the updated target object position, and re-executing the artificial potential field correction and control tracking process.
8. A low-disturbance non-inductive grabbing system of a floating target control mechanical arm based on an artificial potential field, which is characterized by comprising: the track planning module is used for generating an initial planning track according to the initial pose and the target pose; the artificial potential field construction module is used for generating radial potential field force and tangential potential field force according to the relative spatial relation between the tail end and the target to form artificial potential field virtual force; the impedance track correction module is used for introducing the artificial potential field virtual force into the impedance model and dynamically correcting the planned track; The joint space control module is used for constructing a super-torsion sliding mode controller based on symmetrical preset performance and realizing track tracking control of the mechanical arm; The control parameter learning module is used for carrying out iterative optimization on the control parameters through a reinforcement learning algorithm; and the grabbing detection and re-planning module is used for detecting grabbing results and judging whether a re-planning mechanism is triggered or not.

Description

Floating target control mechanical arm low-disturbance and noninductive grabbing method based on artificial potential field Technical Field The invention relates to the technical field of robot motion control and intelligent operation, in particular to a low-disturbance and non-sense grabbing method for a floating target control mechanical arm based on an artificial potential field, and belongs to the technical field of control combining robot compliance control, motion planning and learning optimization. Background Along with the wide application of industrial robots in assembly, sorting, grabbing and other operation scenes, the robot arm achieves stable and reliable target object grabbing in an unstructured or semi-structured environment, and becomes one of important research directions in the field of robot control. Especially in the application scene that the position of the target object is uncertain, the environment is disturbed or the requirement on contact safety is high, how to reduce unnecessary contact impact and force interference while ensuring the grabbing success rate is a key problem of long-term attention in the technical field. Existing robotic arm gripping systems typically consist of a trajectory planning layer and a motion control layer. At the track planning layer, a method such as polynomial track planning, spline curve planning and the like is often adopted to generate an expected motion track of the end effector from the initial pose to the target pose. The planning methods are mainly designed aiming at position, speed and acceleration continuity, so that the robot can smoothly move between given starting and stopping postures. In the motion control layer, the mechanical arm is usually used for tracking the planned track by a joint space or Cartesian space control method. In the task of making contact with a target object or environment, methods such as impedance control and compliance control are largely introduced into a robot arm control system in order to reduce contact shock and improve system compliance. The method enables the mechanical arm to show the characteristic similar to a spring-damping-mass system in the contact process by constructing a dynamic relation between force and displacement, thereby reducing the impact risk brought by rigidity control. In addition, the artificial potential field method is used as a classical motion planning and obstacle avoidance method, and by constructing attraction potential fields and repulsion potential fields, the mechanical arm is guided to move towards a target and away from an obstacle, and the method is widely applied to robot path planning and local motion correction. In the control strategy level, in order to cope with model uncertainty and external disturbance, robust control methods such as sliding mode control, self-adaptive control and the like are introduced in some researches so as to improve system stability and tracking precision. Meanwhile, with the improvement of computing power, data driving methods such as reinforcement learning and the like are gradually applied to the field of robot control and used for parameter setting, strategy optimization or decision control in complex environments. Although various methods for robotic arm gripping and contact control have been proposed in the prior art, the following deficiencies still exist in achieving zero contact force gripping, control robustness, parameter adaptation, and the like: (1) The existing grabbing track planning method is mainly based on position planning and lacks mechanical constraint on the contact process: In the prior art, the track planning is widely performed on the tail end of the mechanical arm by adopting methods such as polynomial tracks, spline curves and the like, the methods mainly focus on the continuity of position, speed and acceleration, and the default target object is fixed in position and can be accurately reached. When the tail end approaches to the target object, the method does not explicitly restrict the force change process before and after contact, so that impact force is easily generated at the moment of contact, and the grabbing requirement of strictly limiting or approaching zero of the contact force is difficult to meet. (2) The method based on the artificial potential field is mostly used for obstacle avoidance or path guidance, and is less directly used for contact force regulation: The artificial potential field method is mainly used for obstacle avoidance or path guidance in the prior art, and the potential field force is usually used as an auxiliary correction amount of a planning layer or a motion layer. In the existing method, more potential field force is not systematically introduced into a robot dynamics or impedance model, and a clear physical corresponding relation is difficult to establish between a potential field guiding effect and a contact force evolution process, so that the application of the method in a fine