Efficient Gradient-Based Inference for Manipulation Planning in Contact Factor Graphs
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866915188080377856 |
|---|---|
| author | Lee, Jeongmin Park, Sunkyung Lee, Minji Lee, Dongjun |
| author_facet | Lee, Jeongmin Park, Sunkyung Lee, Minji Lee, Dongjun |
| contents | This paper presents a framework designed to tackle a range of planning problems arise in manipulation, which typically involve complex geometric-physical reasoning related to contact and dynamic constraints. We introduce the Contact Factor Graph (CFG) to graphically model these diverse factors, enabling us to perform inference on the graphs to approximate the distribution and sample appropriate solutions. We propose a novel approach that can incorporate various phenomena of contact manipulation as differentiable factors, and develop an efficient inference algorithm for CFG that leverages this differentiability along with the conditional probabilities arising from the structured nature of contact. Our results demonstrate the capability of our framework in generating viable samples and approximating posterior distributions for various manipulation scenarios. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_06300 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Efficient Gradient-Based Inference for Manipulation Planning in Contact Factor Graphs Lee, Jeongmin Park, Sunkyung Lee, Minji Lee, Dongjun Robotics This paper presents a framework designed to tackle a range of planning problems arise in manipulation, which typically involve complex geometric-physical reasoning related to contact and dynamic constraints. We introduce the Contact Factor Graph (CFG) to graphically model these diverse factors, enabling us to perform inference on the graphs to approximate the distribution and sample appropriate solutions. We propose a novel approach that can incorporate various phenomena of contact manipulation as differentiable factors, and develop an efficient inference algorithm for CFG that leverages this differentiability along with the conditional probabilities arising from the structured nature of contact. Our results demonstrate the capability of our framework in generating viable samples and approximating posterior distributions for various manipulation scenarios. |
| title | Efficient Gradient-Based Inference for Manipulation Planning in Contact Factor Graphs |
| topic | Robotics |
| url | https://arxiv.org/abs/2503.06300 |