基于积式决策的全方位移动双臂机器人连续轨迹任务多目标规划

Multi-objective Planning of Continuous Trajectory Task for an Omni-directional Mobile Dual-Arm Robot Based on Product Arbitration

OMDAR(全方位移动双臂机器人)的运动规划算法需要同时兼顾不同类型、量纲、幅值等多重优化目标.本文分析了受约束OMDAR系统的多目标运动规划任务的数学建模与求解方法.在积式决策多目标优化算法框架下,将OMDAR系统连续轨迹运动规划需求与相关约束建模为乘积形式单一优化目标函数,采用高斯巡游粒子群优化算法(GR-PSO),可靠有效地实现了问题的求解.同时,将GR-PSO算法与经典非线性优化SQP算法进行对比,以9重目标优化导引下的连续轨迹路径规划为实例,证明了GR-PSO算法的有效性与优势.

The motion planning algorithm of omni-directional mobile dual-arm robot （OMDAR） must cope with a group of optimization criteria simultaneously, such as different types, dimensions and ranges. The mathematical modeling method and resolving algorithm of the constrained OMDAR multi-objective motion planning mission are analyzed. Based on the product arbitration based multi-objective optimization algorithm, motion planning of the continuous trajectory of OMDAR is modeled as a single objective function which is consWucted with the product of multi-objective performance functions and constraints. The problem is successfully resolved with the Gaussian rovering particle swarm optimization （GR-PSO） algorithm. To compare the performance of the GR-PSO algorithm and the classical sequential quadratic optimization （SQP） algorithm, continuous trajectory planning governed by 9-criteria optimization functions is resolved. The effectiveness and advantage of the GR-PSO are verified.