Fast Bilateral Teleoperation and Imitation Learning Using Sensorless Force Control via Accurate Dynamics Model

2025-07-08

AI Summary

To address the challenge of achieving high-fidelity teleoperation and effective imitation learning on low-cost robotic arms lacking force sensors, this paper proposes a sensorless force estimation and four-channel bilateral control framework grounded in precise dynamic modeling. Methodologically, it integrates nonlinear compensation, online velocity and external force estimation, inertia-adaptive gain tuning, system identification, and imitation learning algorithms. Notably, it achieves millisecond-level real-time force feedback teleoperation on affordable hardware without physical force sensors—demonstrating, for the first time, such capability on sensorless low-cost platforms. Experiments show substantial improvements in operational accuracy and stability during high-speed, contact-rich tasks; imitation learning policies trained on data augmented with estimated force signals achieve a 32% higher success rate and enhanced robustness. The core contribution is the establishment of the first integrated sensorless force-feedback teleoperation–imitation learning paradigm specifically designed for low-cost robotic systems.

Technology Category

• Machine Learning: Imitation Learning & Inverse Reinforcement Learning
• Intelligent Robots: Manipulation
• Search and Optimization: Learning to Search

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments

Abstract

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.

Problem

• Enabling fast teleoperation with force feedback using low-cost manipulators
• Improving imitation learning by incorporating force information in policies
• Achieving accurate dynamics modeling for sensorless force control

Innovation

• 4-channel bilateral control for force feedback
• Accurate dynamics model with nonlinear compensation
• Force-augmented imitation learning policies