Reproduce “Hierarchical Deformation Planning and Neural Tracking for DLOs in Constrained Environments” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yunxi Tang, Tianqi Yang, Jing Huang, Xiangyu Chu, Kwok Wai Samuel Au
Researchers from The Chinese University of Hong Kong developed a unified framework for manipulating deformable linear objects (DLOs) in cluttered environments, integrating a hierarchical deformation planner and a neural model predictive control (MPC) tracker based on a Transformer-based deformation model. This system enables precise and collision-free DLO manipulation, demonstrating robust performance across diverse cable types and complex real-world scenarios.
Key insights
- A hierarchical approach effectively decouples global DLO deformation planning from local real-time tracking, enabling efficient generation of complex trajectories while maintaining robust execution.
- Leveraging a Transformer-based architecture with self-attention and cross-attention mechanisms accurately models both internal DLO dynamics and long-range robot-DLO interaction effects.
- Efficient online adaptation of only the deformation model's decoder allows for robust generalization and deployment across various deformable linear objects in real-world scenarios.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.