ILBiT: Imitation Learning for Robot Using Position and Torque Information based on Bilateral Control with Transformer
Masato Kobayashi, Thanpimon Buamanee, Yuki Uranishi, Haruo Takemura
Abstract
Autonomous manipulation in robot arms is a complex and evolving field of study in robotics. This paper introduces an innovative approach to this challenge by focusing on imitation learning (IL). In contrast to traditional imitation methods, our approach uses IL based on bilateral control, allowing for more precise and adaptable robot movements. Conventional IL based on bilateral control method relies on Long Short-Term Memory (LSTM) networks. In this paper, we present the IL for robots using position and torque information based on Bilateral control with Transformer (ILBiT). This proposed method employs the Transformer model, known for its robust performance in handling diverse datasets and its capability to overcome LSTM, especially in tasks requiring detailed force adjustments. A highlighting feature of ILBiT is its high-frequency operation at 100Hz, which significantly improves the system's adaptability and response to varying environments and objects with different hardness levels. The effectiveness of the ILBiT method is demonstrated through comprehensive real-world experiments.