LiPo: A Lightweight Post-optimization Framework for Smoothing Action Chunks Generated by Learned Policies

2025-06-05

AI Summary

Imitation learning often yields discontinuous trajectories in unstructured environments for dynamic manipulation tasks (e.g., throwing, lifting), primarily due to discrete action chunking—compromising motion smoothness and system stability. This paper proposes a lightweight post-optimization framework comprising three key innovations: (1) inference-aware overlapping chunk scheduling to mitigate abrupt transitions between chunks; (2) a linear blending transition mechanism enabling smooth interpolation across action segments; and (3) jerk-minimizing trajectory optimization under bounded perturbation constraints. The method preserves real-time policy execution while significantly suppressing mechanical vibration and jitter, thereby enhancing trajectory smoothness and robotic robustness. Experimental validation on a physical robotic arm platform demonstrates both effectiveness in dynamic manipulation tasks and practical deployability.

Technology Category

• Intelligent Robots: Learning & Optimization for ROB
• Machine Learning: Imitation Learning & Inverse Reinforcement Learning
• Planning, Routing, and Scheduling: Replanning and Plan Repair

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Recent advances in imitation learning have enabled robots to perform increasingly complex manipulation tasks in unstructured environments. However, most learned policies rely on discrete action chunking, which introduces discontinuities at chunk boundaries. These discontinuities degrade motion quality and are particularly problematic in dynamic tasks such as throwing or lifting heavy objects, where smooth trajectories are critical for momentum transfer and system stability. In this work, we present a lightweight post-optimization framework for smoothing chunked action sequences. Our method combines three key components: (1) inference-aware chunk scheduling to proactively generate overlapping chunks and avoid pauses from inference delays; (2) linear blending in the overlap region to reduce abrupt transitions; and (3) jerk-minimizing trajectory optimization constrained within a bounded perturbation space. The proposed method was validated on a position-controlled robotic arm performing dynamic manipulation tasks. Experimental results demonstrate that our approach significantly reduces vibration and motion jitter, leading to smoother execution and improved mechanical robustness.

Problem

• Smoothing discontinuities in chunked action sequences
• Reducing vibration and motion jitter in robots
• Improving mechanical robustness in dynamic manipulation tasks

Innovation

• Inference-aware chunk scheduling for overlap
• Linear blending to reduce abrupt transitions
• Jerk-minimizing trajectory optimization in perturbation space