Manipulator Motion Planning Based on Prior Knowledge Guidance in Complex Constrained Environment

Abstract

Manipulator motion planning in the complex constrained space is essential for realizing the universal service robot. However, the limited space makes it challenging for the manipulator to quickly plan a feasible path and easily fall into a local minimum. Moreover, the feasible path is tortuous and unsmooth due to complex space constraints. Therefore, an improved RRT*-Connect algorithm is proposed, which uses prior knowledge to guide the sampling process of two trees, aiming at fast convergence to a feasible path in various scenes. Then, constraints such as path length and joint rotation angle are introduced in this paper. Based on the above constraints, trajectory optimization is carried out to obtain the optimal path. The experimental results show that the non-uniform sampling method can quickly find a feasible path. After trajectory optimization, a smooth, efficient, and stable path is obtained. When the constrained space complexity reaches 95.56%, our proposed algorithm can guarantee a significantly high planning success rate.