Search papers, labs, and topics across Lattice.
This paper introduces SkillPlug, a framework that enhances existing visuomotor policies by integrating a skill-conditioning module that mines transferable skills from raw multi-task demonstrations. By employing self-supervised objectives, SkillPlug generates compact and reusable behavior-level primitives, allowing for efficient adaptation to new tasks with minimal data. The evaluation on simulation benchmarks and a real robot shows that SkillPlug significantly boosts multi-task performance and few-shot adaptation capabilities, demonstrating its effectiveness in robotic manipulation.
SkillPlug reveals that mining transferable skills can dramatically enhance few-shot adaptation in robotic manipulation, outperforming traditional end-to-end training methods.
Learning transferable visuomotor imitation policies that generalize across diverse manipulation tasks and adapt rapidly to new tasks from only a handful of demonstrations remains challenging. Most modern policies are trained end-to-end to map observations directly to low-level actions, offering little explicit structure for reusing and recombining behaviors across tasks and making transfer data-inefficient under limited supervision. We propose SkillPlug, a plug-in framework that augments an existing visuomotor policy with a skill-conditioning module and mines a shared, transferable skill library from raw multi-task demonstrations. SkillPlug learns skills via self-supervised objectives that promote compact, reusable, and non-redundant behavior-level primitives, forming a task-shared prior for compositional control. After skill mining, we keep the learned skills fixed and specialize to unseen tasks by fine-tuning only lightweight router and action head, enabling efficient adaptation without full end-to-end retraining. We evaluate SkillPlug on two simulation benchmarks and on a real robot, and observe that the mined transferable skills consistently improve both multi-task performance and few-shot adaptation. Overall, SkillPlug offers a scalable way to mine reusable skills that improve data-efficient generalization in robotic manipulation.
