Enhancing Safety and Manipulability of Redundant Manipulators: Accelerated Motion Generation in Dynamic Environments
zongwu Xie, Mengfei Li, Wandong Sun, BAOSHI CAO, Yang Liu, Zhengpu Wang, Yiming Ji, Hong Liu, Boyu Ma, Zhihong Wu
AI summary
Problem
Redundant manipulators struggle with real-time, safe motion generation in dynamic environments due to high computational costs of non-convex optimization, singularity risks, and inadequate handling of moving obstacles.
Approach
The ESM framework integrates geometry-based dynamic obstacle avoidance, manipulability optimization, trajectory tracking, and joint limits at the joint-angle level, solved via a novel Accelerated Multi-agent recurrent Neural Network (AMNN) accelerated by GPU parallel computing.
Key results
- Unified joint-angle level framework integrating dynamic obstacle avoidance, manipulability optimization, tracking, and joint limits
- Reconstructed joint-velocity obstacle avoidance inequality incorporating obstacle speed for guaranteed collision-free motion
- AMNN solver using meta-heuristic activation functions to solve time-varying non-convex control problems
- GPU-based parallel acceleration method significantly reducing computation time for real-time deployment
Why it matters
Provides a computationally efficient and safe control solution for industrial and service robots operating in unpredictable, dynamic environments.
Abstract
Motion generation in dynamic environments is crucial for human-machine interaction with redundant manipulators. In this context, we propose the Enhancing Safety and Manipulability (ESM) scheme, which integrates geometry- based dynamic obstacle avoidance, manipulability optimization, trajectory tracking, and joint limit avoidance into a unified scheme operating at the joint-angle level. The introduction of a flexible collision library enables the scheme to locate critical points based on geometry, while the incorporated obstacle speed allows the scheme to effectively avoid dynamic obstacles. In the ESM, manipulability is naturally set as the non-convex goal. To solve the ESM, the Accelerated Multi-agent recurrent Neural Network (AMNN) is proposed, which uses a meta-heuristic approach to construct activation functions, endowing the neural network with non-convex control capabilities. Subsequently, a GPU-based parallel computing method is implemented, significantly reducing computing time. Detailed simulations, experiments, and comparisons demonstrate the framework's effectiveness and superiority.