基于粒子群优化的时间最优机械臂轨迹规划算法

Time-Optimal Trajectory Planning Algorithm for Manipulator Based on PSO

根据机械臂运动学约束,提出了关节空间基于粒子群优化(PSO)的时间最优3-5-3多项式插值轨迹规划算法,解决了由于多项式插值轨迹规划具有阶次高、没有凸包性质等缺点,难以应用传统优化方法进行优化的问题.粒子群算法结构简单、参数易调整的特点弥补了多项式阶插值的缺点.直接在优化目标空间搜索,巧妙地避免了粒子群计算构造自变量和因变量的映射,降低了搜索维数,简化了计算.在优化过程中,采用两个适应度函数之间切换的开关控制,使各段插值尽快收敛于运动学约束内.通过与传统3-5-3多项式插值的运动位置、速度、加速度曲线对比,证明该方法运行时间更短,稳定性和流畅性更好.

According to the kinematic constraints of manipulator, a time-optimal 3-5-3 polynomial interpolation trajectory planning algorithm based on particle swarm optimization （PSO） is proposed, which solves the problem that polynomial interpolation based trajectory planning is hard to be optimized by traditional optimization methods for its shortcomings of high order and lack of convex hull property, etc. The characteristics of simple structure and easily adjusted parameters of PSO remedy the defects of polynomial optimization. Searching directly in the optimization space of the target skillfully avoids the mapping between independent and dependent variables, reduces the search dimension and simplifies the computation. In the optimization process, switch control is used between the two fitness functions to make interpolation converge quickly within the kinematic constraints. The comparison between the the algorithm and traditional 3-5-3 spline interpolation on the position, velocity and acceleration of the movement shows that the running time of the algorithm is shorter, and its stability and fluidity are better.