Bi-LAT: Bilateral Control-Based Imitation Learning via Natural Language and Action Chunking with Transformers

Abstract

I. INTRODUCTION

II. RELATED WORKS

II-A. Bilateral Control-Based Imitation Learning

II-B. Imitation Learning Using Language Models

III. Bi-LAT Method

III-A. Overview

III-B. Data Collection

III-C. Language Encoder

III-D. Learning Model

III-E. Inference

IV. Unimanual Experiment

V. Bimanual Experiment

VI. CONCLUSIONS

References