一种工业机器人多目标轨迹优化算法

为解决工业机器人工作效率低、能耗损失严重和关节冲击磨损较大的问题，提出了一种基于布谷鸟搜索（cuckoo search， CS）算法和非支配排序遗传算法-II（non-dominated sorting genetic algorithm-II， NSGA-II）的混合算法（简称为CSNSGA-II），用于机器人的轨迹优化。采用5次非均匀有理B样条（non-uniform rational B-splines， NURBS）曲线作为工业机器人的轨迹规划曲线，同时以运动时间、能耗和冲击磨损为优化目标构建相应的多目标轨迹优化模型，并在速度、加速度和加加速度的约束下采用CSNSGA-II进行轨迹优化。CSNSGA-II以Tent混沌映射初始化时间序列，采用不可行度算法将解分为可行解与不可行解，并利用改进的CS算法对不可行解进行处理。利用MATLAB软件对6R勃朗特机器人进行建模仿真，并对得到的非支配解集和归一化加权迭代最优值进行对比分析。仿真结果表明，相比于NSGA-II、多目标粒子群优化（multi-objective particle swarm optimization， MOPSO）算法，所提出的CSNSGA-II可更有效地对6R勃朗特机器人的轨迹进行优化，所得非支配解集更加均匀且接近真实Pareto前沿，最终得到的轨迹曲线较为平滑，可同时满足6R勃朗特机器人的高效率、低能耗及少冲击磨损的要求。所提出的方法可为进一步推动工业机器人在生产中的广泛应用以及提高生产能力和效率提供指导。

A multi-objective trajectory optimization algorithm for industrial robot

In order to solve the problems of low work efficiency, serious energy loss and large joint impact wear of industrial robots, a hybrid algorithm (referred to as CSNSGA-II) based on cuckoo search (CS) algorithm and non-dominated sorting genetic algorithm-II (NSGA-II) was proposed for trajectory optimization of robots. The quintic non-uniform rational B-splines (NURBS) curve was used as the trajectory planning curve of the industrial robot. At the same time, the motion time, energy consumption and impact wear were taken as the optimization objectives and the corresponding multi-objective trajectory optimization model was constructed. Under the constraints of speed, acceleration and jerk, the CSNSGA-II was used to optimize trajectory. The CSNSGA-II initialized the time series with the Tent chaotic map, and used the infeasibility algorithm to divide the solutions into feasible solution and infeasible solution, and then the infeasible solution was processed by the improved CS algorithm. The 6R Bronte robot was modeled and simulated by using the MATLAB software, and the obtaind non-dominated solution set and the normalized weighted iterative optimal value were compared and analyzed. The simulation results showed that, compared with the NSGA-II and the multi-objective particle swarm optimization (MOPSO) algorithm, the proposed CSNSGA-II could optimize the trajectory of 6R Bronte robot more effectively, and the non-dominated solution set was more uniform and close to the real Pareto front, and the final trajectory curve was relatively smooth, which could meet the requirements of high efficiency, low energy consumption and less impact wear of 6R Bronte robot at the same time. The proposed method can provide guidance for further promoting the widespread application of industrial robots in production and improving production capacity and efficiency.