Publications

Deep reinforcement learning-based variable impedance control for grinding workpieces with complex geometry

Robotic Intelligence and Automation

This paper presents a DRL-based variable impedance control policy for robot force tracking in complex environments, emphasizing stability and interpretability.

Yanghong Li , Yahao Wang , Li Zheng , Yingxiang Lv , Jin Chai , Erbao Dong^✉

Deep reinforcement learning-based variable impedance control for grinding workpieces with complex geometry

Impedance learning-based adaptive force tracking for robot on unknown terrains copy

IEEE Transactions on Robotics

We propose a novel adaptive variable impedance control policy based on deep reinforcement learning (DRL) to address the robust force tracking challenge for robots in uncertain environments.

Yanghong Li , Li Zheng , Yahao Wang , Shiwu Zhang , Erbao Dong^✉

Impedance learning-based adaptive force tracking for robot on unknown terrains copy

Uncertainty in Bayesian Reinforcement Learning for Robot Manipulation Tasks with Sparse Rewards

2023 IEEE International Conference on Robotics and Biomimetics (ROBIO)

We propose a Bayesian deep reinforcement learning (BDRL) framework to address uncertainty in robot manipulation tasks with sparse rewards, significantly improving convergence speed.

Li Zheng , Yanghong Li , Yahao Wang , Guangrui Bai , Haiyang He , Erbao Dong^✉

Uncertainty in Bayesian Reinforcement Learning for Robot Manipulation Tasks with Sparse Rewards

A novel planner framework compatible with various end-effector constraints

Robotic Intelligence and Automation

This paper presents the Transformation Cross-sampling Framework (TC-Framework), a novel planner framework that enhances the adaptability of robot motion planners to various end-effector constraints.

Yahao Wang , Yanghong Li , Li Zheng , Haiyang He , Sheng Chen , Erbao Dong^✉

A novel planner framework compatible with various end-effector constraints

Deep visual-guided and deep reinforcement learning algorithm based for multip-peg-in-hole assembly task of power distribution live-line operation robot

Journal of Intelligent & Robotic Systems

We propose a framework for a Power Distribution Network Live-line Operation Robot (PDLOR) with autonomous tool assembly capabilities, addressing challenges in dynamic environments through deep visual-guided localization and a novel assembly algorithm.

Li Zheng , Jiajun Ai , Yahao Wang , Xuming Tang , Shaolei Wu , Sheng Cheng , Rui Guo , Erbao Dong^✉

Deep visual-guided and deep reinforcement learning algorithm based for multip-peg-in-hole assembly task of power distribution live-line operation robot

An efficient constraint method for solving planning problems under end-effector constraints

Industrial Robot

This paper presents the Transformation Cross-sampling Framework (TC-Framework), a novel planner framework that enhances the adaptability of robot motion planners to various end-effector constraints.

Yahao Wang , Yanghong Li , Li Zheng , Haiyang He , Sheng Chen , Erbao Dong^✉

An efficient constraint method for solving planning problems under end-effector constraints

An efficiently convergent deep reinforcement learning-based trajectory planning method for manipulators in dynamic environments

Journal of Intelligent & Robotic Systems

We address the inefficient convergence problem in DRL-based trajectory planning for manipulators in dynamic environments by proposing a dynamic action selection strategy and combinatorial reward function.

Li Zheng , Yahao Wang , Run Yang , Shaolei Wu , Rui Guo , Erbao Dong^✉

An efficiently convergent deep reinforcement learning-based trajectory planning method for manipulators in dynamic environments

A Multi-Scale Grasp Detector Based on Fully Matching Model

Computer Modeling in Engineering & Sciences

This paper presents a multi-scale grasp detector that predicts grasp rectangles of varying sizes in real-time for hand-eye cameras and parallel-plate grippers.

Xinheng Yuan , Hao Yu , Houlin Zhang , Li Zheng , Erbao Dong , Heng'an Wu^✉

A Multi-Scale Grasp Detector Based on Fully Matching Model