Robots move faster with new SAIL system ^{CEFR B2}

Researchers at Georgia Tech introduced SAIL (Speed Adaptation for Imitation Learning) to address a practical limitation of imitation learning: robots trained on human demonstrations tend to be limited to the demonstrated speed. The researchers say the goal is a general-purpose robot that can match human hands, and that speed is essential to make such robots useful outside the lab.

SAIL follows a modular approach. Separate components work together to keep motions smooth at high speed, track movements accurately, adjust speed dynamically according to task complexity, and schedule actions to compensate for hardware delays. This design aims to let robots move quickly while remaining stable and coordinated, and to choose when speed helps versus when it could cause mistakes, as Benjamin Joffe explains.

The team tested SAIL in simulation and on two physical robot platforms across twelve tasks, including stacking cups, folding cloth, plating fruit, packing food items and wiping a whiteboard. In most cases SAIL-enabled robots completed tasks three to four times faster than standard imitation-learning systems without losing accuracy. The whiteboard-wiping task proved more difficult because maintaining contact made high-speed execution fragile.

While SAIL does not make robots universally adaptable, the work marks a meaningful step toward faster, more reliable imitation learning. The research was presented at the Conference on Robot Learning (CoRL) and funded by the State of Georgia and the Agricultural Technology Research Program at Georgia Tech.