REAL-TIME SELF-COLLISION AVOIDANCE FOR HUMANOID ROBOT BASED ON DYNAMIC TASK-PRIOR SYSTEM

Yihong Chen and Qijun Chen

Keywords

Selfcollision, taskprior, freefloating, nonlinear, optimization, humanoid robot

Abstract

Self-collision between manipulator links is a physical constraint in designing and controlling the redundant manipulators. This paper proposes a task-prior algorithm to avoid self-collision during highly complex robotic whole-body motion. Based on 3D geometry information, a collision model with paired-up simple segments is constructed, in which collisions can be detected by calculating the distance between the pairs. To improve the real-time capability, nonlinear optimization method is employed to select potentially collusive pairs as candidates subjected to real-time monitoring. Dynamic task-prior structure is utilized not only to mitigate the conflict between self-collision avoidance and tracking the trajectory of the end-effector in the free-floating kinematic coordinates, but also to compare tracking accuracy under different task priorities. Experiments on both simulation and a full-body humanoid are designed to validate the task-prior algorithm. With the proposed method, the full-body humanoid robot succeeds in avoiding collision during horizontal and vertical movements of the arms.

Important Links:



Go Back