Online self-learning multi-joint motion planning method based on neural network

Abstract

The invention discloses an online self-learning multi-joint motion planning method based on a neural network. According to the method, real-time self-learning and control of a nonlinear complex path are achieved by utilizing the global optimal approximation performance of the neural network. Due to the fact that trajectory planning control is synchronously carried out by multiple joints and has acomplex nonlinear relation with time, the online self-learning of the multiple joints is achieved by establishing a neural network model of a time correlation sequence of each joint position, and themethod is used for real-time control of an intelligent mechanical arm; and the difficulty of numerical solution is greatly reduced, the operation efficiency is improved, and the real-time self-learning capability is achieved.

Description

Technical field [0001] The invention relates to an online self-learning multi-joint motion planning method based on a neural network, which is used for online self-learning to realize the motion planning and control of a multi-joint mechanical arm, belongs to the field of intelligent robot trajectory planning and control, and is particularly suitable for realizing mechanical arm pairing Discovery, real-time self-learning and control of new trajectory paths. Background technique [0002] With the development of technology, intelligent robots have received extensive attention and applications. They include highly integrated equipment of multiple disciplines such as mechanical structure, drive, and control. The robotic arm of intelligent robots generally contains multiple drive joints, and its trajectory control determines The precision, service and application of the robotic arm. Robotic arm motion trajectory control is mostly used in three-dimensional space. At present, state spa...

AI Technical Summary

Problems solved by technology

However, the solution of the working space of the manipulator in the three-dimensional space involves a variety of sampling numerical methods, multi-joint synchronous trajectory planning control, and its complex nonlinear relationship with time, the solution process is cumbersome, difficult and inefficient, and has delays. characteristics, it is difficult to realize the self-learning real-time motion planning of the robotic arm

Images