Object interaction focus improves robot manipulation learning efficiency
FOCI Policy: Focus on Object-Centric Interactions for Relational Manipulation Policies
Robotics
Summary
Learning how robots manipulate objects can be tricky because actions depend on how objects move relative to each other. The authors noticed that many tasks involve brief moments where objects interact in specific ways. To take advantage of this, they developed a method called FOCI Policy that breaks down demonstrations into short interaction phases and represents skills based on how objects move relative to each other, rather than the robot's exact actions. This approach helps robots learn manipulation tasks more quickly and work well even with fewer training examples.
robot manipulationobject-centric policyinteraction phasesSE(3) motionrelational tasksrobot learninginductive biastraining efficiency
Authors
Ze Fu, Pinhao Song, Yutong Hu, Renaud Detry
Abstract
Object-centric manipulation policies improve generalization by modeling object motion instead of directly predicting robot actions. However, existing methods are often limited by representations which are either too simplistic to capture interaction dynamics or too dense to learn efficiently. We observe that many rigid relational manipulation tasks are governed by short interaction phases where the relative motion between task-relevant objects is tightly constrained. Based on this observation, we propose \textsc{Foci Policy}, an interaction-centric framework that achieves a two-fold abstraction: (1) temporally, by automatically extracting compact interaction segments from demonstrations;(2) spatially, by representing skills as relative $SE(3)$ motion between task-relevant objects, yielding invariance to scene configurations and robot embodiment. Experiments on RLBench, COLOSSEUM, and real-world tasks show that \textsc{Foci Policy} achieves strong performance with substantially less training data than prior object-centric and action-centric policies. These results suggest that modeling object-object interactions provides a simple and efficient inductive bias for rigid relational manipulation. Project page: \href{https://fitz0401.github.io/foci-page/}{fitz0401.github.io/foci-page/}.