SAIL: Faster-than-Demonstration Execution of Imitation Learning Policies

Nadun Ranawaka Arachchige, Zhenyang Chen, Wonsuhk Jung, Woo Chul Shin, Rohan Bansal, Yu Hang He, Yingyan Celine Lin, Benjamin Joffe, Shreyas Kousik, Danfei Xu

Keywords: Visuomotor Policy, Adaptive Policy Execution

TL;DR: We identifies the challenges of speeding up the execution of learned visuomotor policies beyond the original demonstration speed, and propose set of design choices to address it.

Abstract: Offline Imitation Learning (IL) methods such as Behavior Cloning (BC) are a simple and effective way to acquire complex robotic manipulation skills. However, existing IL-trained policies are confined to execute the task at the same speed as shown in the demonstration. This limits the task throughput of a robotic system, a critical requirement for applications such as industrial automation. We propose SAIL (Speed-Adaptive Imitation Learning), a framework for enabling faster-than-demonstration execution of policies by addressing key technical challenges in robot dynamics and state-action distribution shifts. SAIL features four tightly-connected components (1) high-gain control to enable high-fidelity tracking of IL policy trajectories, (2) consistency-preserving trajectory generation to ensure smoother robot motion, (3) adaptive speed modulation that dynamically adjusts execution speed based on motion complexity, and (4) action scheduling to handle real-world system latencies. Experimental validation on six robotic manipulation tasks shows that SAIL achieves up to a 4$\times$ speedup over demonstration speed in simulation and up to 3.2$\times$ speedup on physical robots. Video results are shown in https://sail-robot.github.io/.

Submission Number: 14