A hybrid trajectory planning method for robot arms based on intelligent optimization algorithm and BP neural network

Abstract

With the rapid development of intelligent manufacturing, robotic arm trajectory planning has placed increasing demands on both time efficiency and online applicability. To address the time-minimization problem under kinematic constraints, this paper proposes a multi-strategy improved Walrus Optimizer (IMWO) and a BP-neural-network-assisted acceleration framework. In this framework, IMWO is employed to optimize the allocation of trajectory time, while the BP surrogate model is used to quickly predict key dynamic quantities. By replacing repeated discrete sampling during the iterative optimization process, the surrogate model reduces the computational burden of fitness evaluation and improves planning efficiency. The proposed method is validated through benchmark function tests, prediction accuracy analysis, and experiments on a 6-DOF industrial robotic arm. The results indicate that the proposed approach can effectively improve trajectory planning efficiency and has promising potential for online engineering applications.