Residual Reinforcement Learning for Robot Teleoperation under Stochastic Delays
The paper presents a novel approach to robot teleoperation that addresses the challenges posed by stochastic communication delays. By integrating a Long Short-Term Memory (LSTM) state estimator with a residual reinforcement learning policy, the method enhances control stability and performance. Experimental results indicate that this hybrid framework significantly outperforms existing methods, ensuring robust teleoperation even under variable delays.
- ▪Stochastic communication delays can disrupt control stability in robot teleoperation.
- ▪The proposed method combines LSTM state estimation with residual reinforcement learning to improve performance.
- ▪Experimental validation shows significant improvements over state-of-the-art baselines.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Robotics arXiv:2605.15480 (cs) [Submitted on 14 May 2026] Title:Residual Reinforcement Learning for Robot Teleoperation under Stochastic Delays Authors:Kaize Deng, Zewen Yang View a PDF of the paper titled Residual Reinforcement Learning for Robot Teleoperation under Stochastic Delays, by Kaize Deng and 1 other authors View PDF HTML (experimental) Abstract:Stochastic communication delays in teleoperation introduce signal discontinuities that undermine control stability and degrade control performance. Consequently, the conventional reinforcement learning (RL) methods struggle with the delayed observations due to the delay-induced observations, leading to high-frequency chattering.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
