| Proj No. | A3216-251 |
| Title | Guiding Bimanual Manipulation toward Safety Through Objects Interaction Understanding |
| Summary | Bimanual motion planning in unstructured environments poses safety challenges. Existing diffusion-based approaches either ignore safety constraints or rely on complex geometric modeling. We propose a plug-and-play auxiliary motion planning module that leverages Vision-Language Models (VLM) to analyze motion patterns at both task planning and execution levels, identifying risks such as collisions or tearing. By incorporating semantic understanding, our method generates cost gradients to guide the diffusion model toward safer and more optimized trajectories. We plan to integrate our approach with RDT-1b and other diffusion-based strategies, demonstrating its effectiveness in semantic-aware and safety-driven motion planning. |
| Supervisor | Ast/P Wang Ziwei (Loc:S2 > S2 B2C > S2 B2C 83, Ext: +65 67906366) |
| Co-Supervisor | - |
| RI Co-Supervisor | - |
| Lab | Centre for Advanced Robotics Technology Innovation (CARTIN) (Loc: S2.1-B3-01) |
| Single/Group: | Single |
| Area: | Intelligent Systems and Control Engineering |
| ISP/RI/SMP/SCP?: |