Fast Bilateral Teleoperation and Imitation Learning Using Sensorless Force Control via Accurate Dynamics Model
This addresses the challenge of collecting demonstration data for imitation learning with affordable hardware, particularly for fast or contact-rich tasks, though it appears incremental by building on existing bilateral control methods.
The paper tackled the problem of enabling fast teleoperation with force feedback using low-cost, force-sensorless manipulators by implementing 4-channel bilateral control based on an accurate dynamics model, achieving practical effectiveness for high-fidelity operations and data collection. It also showed that incorporating force information into imitation learning policies improves performance.
In recent years, the advancement of imitation learning has led to increased interest in teleoperating low-cost manipulators to collect demonstration data. However, most existing systems rely on unilateral control, which only transmits target position values. While this approach is easy to implement and suitable for slow, non-contact tasks, it struggles with fast or contact-rich operations due to the absence of force feedback. This work demonstrates that fast teleoperation with force feedback is feasible even with force-sensorless, low-cost manipulators by leveraging 4-channel bilateral control. Based on accurately identified manipulator dynamics, our method integrates nonlinear terms compensation, velocity and external force estimation, and variable gain corresponding to inertial variation. Furthermore, using data collected by 4-channel bilateral control, we show that incorporating force information into both the input and output of learned policies improves performance in imitation learning. These results highlight the practical effectiveness of our system for high-fidelity teleoperation and data collection on affordable hardware.