SkiP: When to Skip and When to Refine for Efficient Robot Manipulation

2026-05-15

AI Summary

Traditional imitation learning in robotic manipulation predicts actions frame-by-frame, failing to distinguish between redundant free-space motions and critical contact phases, which leads to suboptimal computational efficiency. This work proposes Skip Policy (SkiP), a novel approach that integrates unsupervised Motion Spectrum Keying (MSK) to automatically identify key action segments and introduces a pioneering action relabeling mechanism. Within a single end-to-end policy network, SkiP dynamically skips non-essential segments while finely predicting crucial ones—without requiring manual annotations or hierarchical architectures. Evaluated across 72 simulated tasks and three real-world robotic tasks, SkiP reduces execution steps by 15%–40% while maintaining or improving task success rates.

Abstract

Previous imitation learning policies predict future actions at every control step, whether in smooth motion phases or precise, contact-rich operation phases. This uniform treatment is wasteful: most steps in a manipulation trajectory traverse free space and carry little task-relevant information, while a small fraction of \emph{key} steps around contacts, grasps, and alignment demand dense, high-resolution prediction. We propose a novel \emph{action relabeling} mechanism: at each timestep in a skip segment, we replace the behavior cloning target with the action at the entrance of the next key segment, enabling the policy to leap over redundant steps in a single decision. The resulting \textbf{Skip Policy (SkiP)} dynamically leaps over skip segments and intensively refines actions in key segments, within a single unified network requiring no learned skip planner or hierarchical structure. To automatically partition demonstrations into key and skip segments without manual annotation, we introduce \emph{Motion Spectrum Keying} (MSK), a fast, task-agnostic procedure that detects local motion complexity from action signals. Extensive experiments across 72 simulated manipulation tasks and three real-robot tasks show that SkiP reduces executed steps by $15$--$40\%$ while matching or improving success rates across various policy backbones. Project page: \texttt{https://pgq18.github.io/SkiP-page/}.

Problem

• robot manipulation
• imitation learning
• action prediction
• efficiency
• key steps

Innovation

• action relabeling
• Skip Policy
• Motion Spectrum Keying
• imitation learning
• efficient robot manipulation