Task-level time-optimal collision avoidance trajectory planning for grinding manipulators. (April 2019)

• Record Type: Journal Article
• Authors: Diao, Shipu; Chen, Xindu; Wu, Lei; Zhong, Zhanqi; Lin, Zeqin
• Is Part Of: Proceedings of the Institution of Mechanical Engineers. Volume 233:Number 8(2019)
• Page Start: 2894
• Page End: 2908
• Publication Date: 2019-04
• Subjects: Grinding manipulators -- multiple discrete non-connected grinding areas -- simulated annealing -- task-level -- time-optimal collision avoidance trajectory -- combined stochastic perturbation method; Mechanical engineering -- Periodicals; 621.05
• DOI: 10.1177/0954406218793660
• Languages: English
• ISSNs: 0954-4062

Abstract

A computational framework that can plan the task-level time-optimal collision avoidance trajectory (TOCAT) of grinding manipulators is constructed based on the improved simulated annealing algorithm. When the workpiece surface has a plurality of discrete non-connected areas that need to be polished by grinding manipulators, the planning of TOCAT for a given grinding task is crucial, because it has a direct impact on the processing efficiency and intelligence of the automatic grinding system. Although many planning algorithms can be used to plan collision avoidance trajectories between any two points, the planning of the task-level TOCAT with multiple collision avoidance sub-trajectories is more difficult because it involves the permutation of the collision avoidance sub-trajectories for connecting several grinding areas. This paper proposes a task-level TOCAT planning framework based on the improved simulated annealing algorithm. Its key point is to plan the time-optimal sub-trajectory between any two points based on the trajectory evaluation mechanism. Its innovation lies in that the simulated annealing algorithm generates new solutions based on the combined stochastic perturbation method. The experimental results show that this framework can effectively solve the task-level TOCAT planning problem in multiple grinding areas, and the duration of the task-level collision avoidance trajectory is not only less discrete but also approximately globally optimal.