Achieving Faster and More Accurate Operation of Deep Predictive Learning

TL;DR

This paper presents a motion generation model that enables robots to perform high-speed, high-precision tasks like sports stacking, achieving a 94% success rate through slow training and fast inference.

Contribution

The paper introduces a novel motion generation approach that improves inference speed and accuracy for robotic tasks, demonstrated on sports stacking.

Findings

Achieved 94% success rate in cup stacking with a real robot.

Demonstrated high-speed, high-precision performance in robotic tasks.

Validated the effectiveness of slow training combined with fast inference.

Abstract

Achieving both high speed and precision in robot operations is a significant challenge for social implementation. While factory robots excel at predefined tasks, they struggle with environment-specific actions like cleaning and cooking. Deep learning research aims to address this by enabling robots to autonomously execute behaviors through end-to-end learning with sensor data. RT-1 and ACT are notable examples that have expanded robots' capabilities. However, issues with model inference speed and hand position accuracy persist. High-quality training data and fast, stable inference mechanisms are essential to overcome these challenges. This paper proposes a motion generation model for high-speed, high-precision tasks, exemplified by the sports stacking task. By teaching motions slowly and inferring at high speeds, the model achieved a 94% success rate in stacking cups with a real robot.