Adaptive Dual-Template on Constraint Manifolds for Efficient Motion Planning of Robotic Arms in Narrow Passages
Year: 2026
Type: article
Abstract: Sampling-based methods are widely used for motion planning in high-dimensional configuration spaces of robotic arms. However, their performance significantly deteriorates in environments containing narrow passages or under manifold constraints. Current research has explored learning-based approaches to address these issues, but these methods lack general applicability to arbitrary scenarios and require extensive training. In this article, we propose a novel sampling-based planning framework that overcomes these challenges through adaptive exploration with a dual-layer template. Our method operates on constraint manifolds, performing biased sampling near obstacles, actively recognizing narrow passages, and adaptively adjusting the search radius. This approach not only ensures both planning efficiency and path quality, but also facilitates efficient planning in new environments by simply switching or reusing templates. Simulation results demonstrate that our method outperforms the widely used constrained motion planning method. Furthermore, experiments conducted with the Sawyer robotic arm in a narrow-passage environment validate the practical feasibility of the proposed method.
Authors: Jintao Ye, Lina Hao, Hongtai Cheng
Institution: Northeastern University
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $2,645
OpenAlex ID: W7225232152