Xiangyu Kong
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Xiangyu Kong has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
- Yuanqing Xia
Zhongyuan University of Technology
3 shared publications - Zhongqi Sun
Beijing Institute of Technology
3 shared publications - Yunshan Deng
School of Automation, Beijing Institute of Technology, Beijing, China
2 shared publications - Di-Hua Zhai
Beijing Institute of Technology
1 shared publication - Huahui Xie
School of Automation, Beijing Institute of Technology, Beijing, China
1 shared publication - Jinxian Wu
School of Automation, Beijing Institute of Technology, Beijing, China
1 shared publication
Latest Publications
- 26 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (8) , 8340-8347
Convex MPC With Unreachable Setpoint for a Class of Affine System
We propose a convex model predictive control (MPC) scheme for a class of affine input systems to reduce the dependence on terminal components and improve real-time control capability. Artificial reference variables are introduced to handle unreachable references, and the terminal set constraint is replaced with an equality constraint. Additionally, the original non-convex problem is replaced by a second-order-cone programming problem, while considering the linearization errors. A tube is constructed to ensure that the predicted states strictly satisfy the state constraints. Moreover, we identified two types of deadlock phenomena in this scheme: one caused by the non-convex characteristics, and the other caused by a zero radius tube. These deadlock are resolved by an adding constraints, and the proposed method is applied to the setpoint tracking problem of wheeled robots. By adjusting the corresponding parameters, a safety region, known as soft obstacle-avoidance constraint, is introduced to the state constraint, which differs from traditional constraint relaxation. Simulation results validate the effectiveness of the proposed method, and the influence of parameters on closed-loop trajectories is analyzed.
- 26 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (8) , 8340-8347
Convex MPC With Unreachable Setpoint for a Class of Affine System
We propose a convex model predictive control (MPC) scheme for a class of affine input systems to reduce the dependence on terminal components and improve real-time control capability. Artificial reference variables are introduced to handle unreachable references, and the terminal set constraint is replaced with an equality constraint. Additionally, the original non-convex problem is replaced by a second-order-cone programming problem, while considering the linearization errors. A tube is constructed to ensure that the predicted states strictly satisfy the state constraints. Moreover, we identified two types of deadlock phenomena in this scheme: one caused by the non-convex characteristics, and the other caused by a zero radius tube. These deadlock are resolved by an adding constraints, and the proposed method is applied to the setpoint tracking problem of wheeled robots. By adjusting the corresponding parameters, a safety region, known as soft obstacle-avoidance constraint, is introduced to the state constraint, which differs from traditional constraint relaxation. Simulation results validate the effectiveness of the proposed method, and the influence of parameters on closed-loop trajectories is analyzed.
- 30 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (7) , 7524-7531
Differential High Order Control Barrier Function-Based Safe Reinforcement Learning
Safe reinforcement learning (RL) aims to learn policy while also ensuring the safety constraints. An increasingly common approach is to design a safety filter based on control barrier function (CBF) or high order control barrier function (HOCBF) for the RL policy. A quadratic programming (QP) is then formulated and solved to modify the RL policy, enabling safe exploration. However, directly integrating the safety filter with RL presents two challenges: (1) the conservativeness of safe policy, and (2) potential infeasibility of the QP under bounded input constraints. These issues limit the performance of safe RL. In this letter, we introduce a differential HOCBF constraint by incorporating neural network-based penalty functions into HOCBF. Furthermore, we propose a differential HOCBF-based safe RL framework in which the penalty functions and RL policy are trained concurrently. To address conservativeness, we train penalty functions to maximize long-term rewards while preventing abrupt changes in safe action, thereby achieving ideal performance. To ensure the feasibility of the formulated QP under bounded input constraints, we calculate a set for penalty functions and prove that the feasibility is guaranteed if the learned penalty functions remain within the set. Finally, we verify the effectiveness of the proposed framework on the wheeled mobile robot navigation and obstacle avoidance task.
- 30 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (7) , 7524-7531
Differential High Order Control Barrier Function-Based Safe Reinforcement Learning
Safe reinforcement learning (RL) aims to learn policy while also ensuring the safety constraints. An increasingly common approach is to design a safety filter based on control barrier function (CBF) or high order control barrier function (HOCBF) for the RL policy. A quadratic programming (QP) is then formulated and solved to modify the RL policy, enabling safe exploration. However, directly integrating the safety filter with RL presents two challenges: (1) the conservativeness of safe policy, and (2) potential infeasibility of the QP under bounded input constraints. These issues limit the performance of safe RL. In this letter, we introduce a differential HOCBF constraint by incorporating neural network-based penalty functions into HOCBF. Furthermore, we propose a differential HOCBF-based safe RL framework in which the penalty functions and RL policy are trained concurrently. To address conservativeness, we train penalty functions to maximize long-term rewards while preventing abrupt changes in safe action, thereby achieving ideal performance. To ensure the feasibility of the formulated QP under bounded input constraints, we calculate a set for penalty functions and prove that the feasibility is guaranteed if the learned penalty functions remain within the set. Finally, we verify the effectiveness of the proposed framework on the wheeled mobile robot navigation and obstacle avoidance task.
- 8 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (7) , 6099-6106
Adaptive High-Order Control Barrier Function-Based Iterative LQR for Real Time Safety-Critical Motion Planning
This letter proposes an adaptive high-order control barrier function-based iterative linear quadratic regulator (AHOCBF-ILQR) algorithm for real time safety-critical motion planning. Firstly, we propose a HOCBF-ILQR method, where a HOCBF-based controller is designed as a safety filter of ILQR to guarantee safety. Then, to address the potential infeasibility issue in HOCBF-ILQR, AHOCBF-ILQR is proposed by introducing an auxiliary variable. In AHOCBF-ILQR, only an unconstrained optimization problem and a quadratic programming need to be solved within a sampling interval. The low computation burden significantly enhances the time efficiency of AHOCBF-ILQR. Furthermore, we provide theoretical proof of the safety and feasibility of AHOCBF-ILQR. The algorithms are tested by motion planning experiments for a wheeled mobile robot, where the robot is required to navigate around static or moving obstacles that are unknown in advance. The experimental results show that AHOCBF-ILQR can solve the control inputs within $\text{0.05}\,s$ , and ensure safety.
- 8 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (7) , 6099-6106
Adaptive High-Order Control Barrier Function-Based Iterative LQR for Real Time Safety-Critical Motion Planning
This letter proposes an adaptive high-order control barrier function-based iterative linear quadratic regulator (AHOCBF-ILQR) algorithm for real time safety-critical motion planning. Firstly, we propose a HOCBF-ILQR method, where a HOCBF-based controller is designed as a safety filter of ILQR to guarantee safety. Then, to address the potential infeasibility issue in HOCBF-ILQR, AHOCBF-ILQR is proposed by introducing an auxiliary variable. In AHOCBF-ILQR, only an unconstrained optimization problem and a quadratic programming need to be solved within a sampling interval. The low computation burden significantly enhances the time efficiency of AHOCBF-ILQR. Furthermore, we provide theoretical proof of the safety and feasibility of AHOCBF-ILQR. The algorithms are tested by motion planning experiments for a wheeled mobile robot, where the robot is required to navigate around static or moving obstacles that are unknown in advance. The experimental results show that AHOCBF-ILQR can solve the control inputs within $\text{0.05}\,s$ , and ensure safety.