Improved autonomous trajectory planning for dexterous arm motion: application to intelligent welding in confined spaces
Trajectory planning for robotic welding manipulators operating in confined spaces is difficult because traditional sampling-based algorithms often suffer from inefficient exploration, slow convergence, and unsmooth trajectories.This study develops an improved RRT*-Connect framework combined with artificial potential field guidance. A target attraction function is introduced during bidirectional tree expansion to guide node growth toward the goal and reduce unnecessary extensions. An improved repulsion function dynamically adjusts the expansion step size to enhance obstacle avoidance in densely obstructed regions. A redundant-node pruning strategy and third-order Bezier curves are then applied to smooth the initial path, after which quintic polynomial interpolation generates a welding trajectory satisfying kinematic continuity requirements. Simulation results show that the proposed algorithm reduces expanded nodes by 21.1% and path length by 19.7% relative to RRT in three-dimensional scenarios, while producing shorter paths than RRT-Connect across all tested environments. In confined-space welding simulations, the collision-free rate reaches 100%, weld seam tracking error is maintained within 0.3 mm, and trajectories remain smooth and continuous.These results indicate that the proposed method improves planning efficiency and path quality for welding manipulators operating in constrained environments and adapts effectively to non-standard workpieces.
Authors
- Yao Ranran
- Zhilong Wu (ORCID: https://orcid.org/0009-0006-9702-9732)
- Xiaolong Xu (ORCID: https://orcid.org/0009-0004-6336-1492)
- Xu Zhang (ORCID: https://orcid.org/0009-0005-9241-0019)
- Tiantian Yang
- Bao Zhao
- Fengqian Song
- Yangyang Zhang
Institutions
- Anhui University (CN)
Publication Details
- Journal
- Engineering Research Express
- Published
- 2026-11-18
- DOI
- https://doi.org/10.1088/2631-8695/aec413
- Primary Topic
- Robotic Path Planning Algorithms
- Type
- article
- Field-Weighted Citation Impact
- 0.00