Mechanical arm online motion planning method combining neural motion planning algorithm and artificial potential field method

An artificial potential field method and motion planning technology, applied in manipulators, program-controlled manipulators, claw arms, etc., can solve problems such as high precision requirements, complex local extremum construction of repulsive potential fields, and difficulty in training. The effect of improving training speed and success rate of motion planning

Active Publication Date: 2022-02-18
HARBIN INST OF TECH
View PDF7 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] Aiming at the problem that the neural motion planner based on reinforcement learning is difficult to train when the precision of the planning task is high, and the repulsive potential field of the artificial potential field method is complicated to construct and has local extremums, the present invention proposes a neural motion planner that combines neural motion On-line Motion Planning Method for Manipulator Based on Planning Algorithm and Artificial Potential Field Method

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Mechanical arm online motion planning method combining neural motion planning algorithm and artificial potential field method
  • Mechanical arm online motion planning method combining neural motion planning algorithm and artificial potential field method
  • Mechanical arm online motion planning method combining neural motion planning algorithm and artificial potential field method

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0042] A method for online motion planning of a manipulator combined with a neural programming algorithm and an artificial potential field method described in this embodiment includes the following steps:

[0043] Step 1: Establish an artificial potential field containing only the gravitational potential field in the working space of the manipulator. The establishment method of the gravitational potential field is shown in the following formula:

[0044]

[0045] In the formula, K a Indicates the gravitational coefficient, p end and at end respectively represent the position and attitude of the end coordinate system, p goal and at goal represent the target position and attitude, respectively, d p 、d a are the position distance and attitude distance, respectively. The formula for motion planning using an artificial potential field containing only a gravitational potential field is as follows:

[0046]

[0047] In the formula, J p (q t ) is the Jacobian matrix of ...

Embodiment

[0061] 1) Experimental tasks

[0062] The speed of direct training on the real manipulator is very slow, and it is easy to damage the manipulator, so the present invention first establishes the dynamics and kinematics model of the manipulator in the physical simulation engine MuJoCo, and performs simulation training to verify the effectiveness of the algorithm. Finally, the trained policy network is tested in the real environment. The mechanical arm adopted in the present invention is a jaco2 cooperative mechanical arm, and this mechanical arm has 7 joints. Such as figure 1 , the training task of manipulator reinforcement learning motion planning is to move the manipulator from an initial pose to a target pose in a desktop environment. The tabletop is the environmental obstacle of the robot arm, and the training round ends when the robot arm collides with the tabletop or between the links of the robot arm itself.

[0063] 2) Training parameters

[0064] The present inventi...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a mechanical arm online motion planning method combining a neural motion planning algorithm and an artificial potential field method, and belongs to the field of robot motion planning. The invention aims to solve the problem that a neural motion planner based on reinforcement learning is difficult to train when precision requirements of a planning task are high, and the problem that a repulsive force potential field of an artificial potential field method is complex to construct and has a local extreme value. According to the method, firstly, the artificial potential field method is simplified, and only a gravitational potential field borne by a mechanical arm is reserved, so that the problem of a local extreme value does not exist any more while construction of a complex repulsive force potential field is avoided; and secondly, a new thought of combining the artificial potential field method and reinforcement learning for planning is provided, a flexible switching mechanism is designed, the reinforcement learning is adopted for planning when a distance from a target is far, and an artificial potential field is adopted for planning when the distance is smaller than a threshold value, so that a training speed of the reinforcement learning and a motion planning success rate are improved. The effectiveness of the method is verified by training and testing planning tasks with different precisions in a simulation engine. The method is applied to the technical field of robot motion planning.

Description

technical field [0001] The invention relates to an online motion planning method of a mechanical arm in a dynamic environment, and belongs to the field of robot motion planning. Background technique [0002] The motion planning algorithm of the manipulator can be divided into two types: offline motion planning algorithm and online motion planning algorithm. The input of the offline motion planning algorithm is the planning target, environmental obstacle information and kinematic constraints, and the output is a complete trajectory. The online motion planning algorithm receives the planning target and state perception information to output the one-step joint motion of the robotic arm. After the robotic arm performs one-step motion, it feeds back the new state information to the motion planner to output the next action. Compared with the offline motion planning algorithm, the online motion planning algorithm has a closed-loop planning ability, and can realize motion planning i...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): B25J9/16B25J18/00
CPCB25J9/1664B25J9/1679B25J18/00
Inventor 白成超郭继峰张家维
Owner HARBIN INST OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap