Obstadiff improves robot motion planning in cluttered environments
ObstaDiff: Generalizable Diffusion Policy Learning via Obstacle-aware Representations
RoboticsMachine Learning
Summary
Robots often struggle moving safely around obstacles in messy real-world spaces. The authors created ObstaDiff, a new way to teach robots how to move by showing them how to see and understand objects and obstacles separately. This helps the robots plan their paths better when reaching for targets without bumping into things. Their tests on farm robots showed it worked better than earlier methods, with higher success and fewer collisions.
What this means in practice
- •For agricultural robot operators: Execute safer and more reliable robot motions around plants and structures in cluttered farm environments using obstacle-aware motion planning.
- •For warehouse automation teams: Enhance autonomous robot navigation and object handling in cluttered warehouse settings by integrating obstacle-aware diffusion policies.
Authors
Jiawen Wang, Kevin Yao, Khalid Jawed
Abstract
Imitation learning has achieved impressive results in robotic manipulation, yet most existing approaches assume clean backgrounds and lack explicit mechanisms for obstacle-aware motion generation. Extending such policies to cluttered, real-world scenes with unstructured obstacles remains a key generalization challenge. We present ObstaDiff, a decomposed diffusion-policy framework with a lightweight obstacle-aware visual encoder. ObstaDiff extracts a structured target-obstacle-background representation, enabling the downstream alignment policy to generate end-effector trajectories toward a target-centered bottleneck pose while reasoning about surrounding obstacles. We evaluate ObstaDiff on 61 real-robot greenhouse trials per method (366 executions in total). ObstaDiff achieves 75.41% average task success and 8.20% average obstacle collision rate, outperforming representative imitation-learning baselines and improving generalization in cluttered agricultural scenes.