GPD: Guided Polynomial Diffusion for Motion Planning
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909470102126592 |
|---|---|
| author | Srikanth, Ajit Mahanjan, Parth Saha, Kallol Mandadi, Vishal Paul, Pranjal Wadhwani, Pawan Bhowmick, Brojeshwar Singh, Arun Krishna, Madhava |
| author_facet | Srikanth, Ajit Mahanjan, Parth Saha, Kallol Mandadi, Vishal Paul, Pranjal Wadhwani, Pawan Bhowmick, Brojeshwar Singh, Arun Krishna, Madhava |
| contents | Diffusion-based motion planners are becoming popular due to their well-established performance improvements, stemming from sample diversity and the ease of incorporating new constraints directly during inference. However, a primary limitation of the diffusion process is the requirement for a substantial number of denoising steps, especially when the denoising process is coupled with gradient-based guidance. In this paper, we introduce, diffusion in the parametric space of trajectories, where the parameters are represented as Bernstein coefficients. We show that this representation greatly improves the effectiveness of the cost function guidance and the inference speed. We also introduce a novel stitching algorithm that leverages the diversity in diffusion-generated trajectories to produce collision-free trajectories with just a single cost function-guided model. We demonstrate that our approaches outperform current SOTA diffusion-based motion planners for manipulators and provide an ablation study on key components. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2501_18229 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | GPD: Guided Polynomial Diffusion for Motion Planning Srikanth, Ajit Mahanjan, Parth Saha, Kallol Mandadi, Vishal Paul, Pranjal Wadhwani, Pawan Bhowmick, Brojeshwar Singh, Arun Krishna, Madhava Robotics Diffusion-based motion planners are becoming popular due to their well-established performance improvements, stemming from sample diversity and the ease of incorporating new constraints directly during inference. However, a primary limitation of the diffusion process is the requirement for a substantial number of denoising steps, especially when the denoising process is coupled with gradient-based guidance. In this paper, we introduce, diffusion in the parametric space of trajectories, where the parameters are represented as Bernstein coefficients. We show that this representation greatly improves the effectiveness of the cost function guidance and the inference speed. We also introduce a novel stitching algorithm that leverages the diversity in diffusion-generated trajectories to produce collision-free trajectories with just a single cost function-guided model. We demonstrate that our approaches outperform current SOTA diffusion-based motion planners for manipulators and provide an ablation study on key components. |
| title | GPD: Guided Polynomial Diffusion for Motion Planning |
| topic | Robotics |
| url | https://arxiv.org/abs/2501.18229 |