Geometrically-Aware One-Shot Skill Transfer of Category-Level Objects
🤖 AI Summary
Robots face significant challenges in transferring manipulation skills to novel objects with unknown geometries in unseen environments, often requiring extensive training data or manual programming. This paper proposes a category-level skill transfer method that generalizes complex manipulation skills—including grasping, placing, and rotating—as well as task constraints from a single human demonstration, without additional training. The approach achieves generalization across geometrically diverse instances within the same object category. Its core innovation lies in the first integration of Functional Maps (FM) with Task Space Imitation Algorithms (TSIA), enabling the construction of object-centric geometric representations and interaction function mappings that preserve functional consistency across morphologically distinct instances. Experiments across multiple real-world scenarios demonstrate high success rates and strong geometric consistency, significantly improving both generalization capability and deployment efficiency of robotic manipulation skills.
Application Category
📝 Abstract
Robotic manipulation of unfamiliar objects in new environments is challenging and requires extensive training or laborious pre-programming. We propose a new skill transfer framework, which enables a robot to transfer complex object manipulation skills and constraints from a single human demonstration. Our approach addresses the challenge of skill acquisition and task execution by deriving geometric representations from demonstrations focusing on object-centric interactions. By leveraging the Functional Maps (FM) framework, we efficiently map interaction functions between objects and their environments, allowing the robot to replicate task operations across objects of similar topologies or categories, even when they have significantly different shapes. Additionally, our method incorporates a Task-Space Imitation Algorithm (TSIA) which generates smooth, geometrically-aware robot paths to ensure the transferred skills adhere to the demonstrated task constraints. We validate the effectiveness and adaptability of our approach through extensive experiments, demonstrating successful skill transfer and task execution in diverse real-world environments without requiring additional training.
Problem
Research questions and friction points this paper is trying to address.
Enables robots to transfer manipulation skills from single human demonstrations.
Uses geometric representations to handle object-centric interactions effectively.
Ensures skill transfer across objects with similar topologies but different shapes.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Skill transfer from single human demonstration
Geometric representations for object-centric interactions
Task-Space Imitation Algorithm for smooth robot paths
🔎 Similar Papers
No similar papers found.
