Yinbei Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Interests
Yinbei Li has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Jiaqiang Yang
Zhejiang University
3 shared publications - Yasir Salam
College of Electrical Engineering, Zhejiang University, Hangzhou, China
2 shared publications - Jing Cao
College of Electrical Engineering, Zhejiang University, Hangzhou, China
1 shared publication - Qingyang Lyu
College of Design and Engineering, National University of Singapore, Singapore
1 shared publication - Rafik Ighil Nessouk
College of Electrical Engineering, Zhejiang University, Hangzhou, China
1 shared publication Wei Fan
College of Electrical Engineering, Zhejiang University, Hangzhou, China
1 shared publication
Latest Publications
- 6 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (10) , 10266-10273
A Hybrid Framework Using Diffusion Policy and Residual RL for Force-Sensitive Robotic Manipulation
Force-sensitive manipulation is essential for tasks such as cleaning, polishing, and surgical assistance, yet it remains challenging due to complex contact dynamics and the need for real-time adaptation. We propose DP-RRL, a hybrid learning framework that combines a diffusion policy (DP) for imitation learning (IL) with a residual reinforcement learning (RL) module to tackle these challenges. First, the DP leverages expert demonstrations to generate smooth, expert-like motion-force trajectories. Next, a residual RL component refines these trajectories online by adapting force commands to account for unmodeled contact dynamics. Our approach incorporates multimodal perception, including RGB point clouds, force feedback, and proprioceptive data, and achieves robust force-sensitive manipulation in unstructured environments. We validate DP-RRL on a 7-DOF robotic arm performing basin-cleaning tasks, where it demonstrates an average success rate of 88% across 12 unseen scenarios, outperforming state-of-the-art baselines.
- 6 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (10) , 10266-10273
A Hybrid Framework Using Diffusion Policy and Residual RL for Force-Sensitive Robotic Manipulation
Force-sensitive manipulation is essential for tasks such as cleaning, polishing, and surgical assistance, yet it remains challenging due to complex contact dynamics and the need for real-time adaptation. We propose DP-RRL, a hybrid learning framework that combines a diffusion policy (DP) for imitation learning (IL) with a residual reinforcement learning (RL) module to tackle these challenges. First, the DP leverages expert demonstrations to generate smooth, expert-like motion-force trajectories. Next, a residual RL component refines these trajectories online by adapting force commands to account for unmodeled contact dynamics. Our approach incorporates multimodal perception, including RGB point clouds, force feedback, and proprioceptive data, and achieves robust force-sensitive manipulation in unstructured environments. We validate DP-RRL on a 7-DOF robotic arm performing basin-cleaning tasks, where it demonstrates an average success rate of 88% across 12 unseen scenarios, outperforming state-of-the-art baselines.
- 1 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive and Developmental Systems
- Vol. 18 (4) , 1-14
FIT: Future-aware Imitation Transformer for Long-Horizon Robotic Manipulation
- 1 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive and Developmental Systems
- Vol. 18 (4) , 1-14
FIT: Future-aware Imitation Transformer for Long-Horizon Robotic Manipulation
- Open Access
- 20 September 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 108688-108696
Torque-Type Backlash Inverse Control for an Electromechanical Actuator With Active Disturbance Rejection
Backlash inverse (BI) is a useful backlash compensator in electromechanical actuators (EMA). However, conventional BI is only relevant when the system damping is strong and the load torque is zero. When confronted with complex conditions, it is frequently ineffective, resulting in a decline in servo system performance. To address this issue, this work proposes a torque-type BI controller for the EMA with active disturbance rejection. To begin modeling EMA, a two-mass system with a deadzone and rigid link is built. The ineffective scenarios of conventional BI are then assessed using this paradigm. Following that, Torque-type BI is developed, which is distinguished by the design of a switching function based on motor current and tracking error. It is also used in conjunction with an active disturbance rejection controller. Finally, Experiments show that a torque-type BI controller may significantly decrease steady-state error and improve the dynamic performance of the EMA in complex conditions, therefore expanding the application scope of conventional BI.
- 20 September 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 108688-108696
Open AccessTorque-Type Backlash Inverse Control for an Electromechanical Actuator With Active Disturbance Rejection
Backlash inverse (BI) is a useful backlash compensator in electromechanical actuators (EMA). However, conventional BI is only relevant when the system damping is strong and the load torque is zero. When confronted with complex conditions, it is frequently ineffective, resulting in a decline in servo system performance. To address this issue, this work proposes a torque-type BI controller for the EMA with active disturbance rejection. To begin modeling EMA, a two-mass system with a deadzone and rigid link is built. The ineffective scenarios of conventional BI are then assessed using this paradigm. Following that, Torque-type BI is developed, which is distinguished by the design of a switching function based on motor current and tracking error. It is also used in conjunction with an active disturbance rejection controller. Finally, Experiments show that a torque-type BI controller may significantly decrease steady-state error and improve the dynamic performance of the EMA in complex conditions, therefore expanding the application scope of conventional BI.