Fengrui Zhang

Mr. Fengrui Zhang

orcid

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 January 2025 in Electronics
Verified
Journal Article
Open access
PDF file
HCHaochong Chen
Haochong Chen
FZFengrui Zhang
Fengrui Zhang
Bilin Aksun GuvencBilin Aksun Guvenc
Bilin Aksun Guvenc
Collision avoidance and path planning are critical topics in autonomous vehicle development. This paper presents the progressive development of an optimization-based controller for autonomous vehicles using the Control Lyapunov Function–Control Barrier Function–Quadratic Programming (CLF-CBF-QP) approach. This framework enables a vehicle to navigate to its destination while avoiding obstacles. A unicycle model is utilized to incorporate vehicle dynamics. A series of simulations were conducted, starting with basic model-in-the-loop (MIL) non-real-time simulations, followed by real-time simulations. Multiple scenarios with different controller configurations and obstacle setups were tested, demonstrating the effectiveness of the proposed controllers in avoiding collisions. Real-time simulations in Simulink were used to demonstrate that the proposed controller could compute control actions for each state within a very short timestep, highlighting its computational efficiency. This efficiency underscores the potential for deploying the controller in real-world vehicle autonomous driving systems. Furthermore, we explored the feasibility of a hierarchical control framework comprising deep reinforcement learning (DRL), specifically a Deep Q-Network (DQN)-based high-level controller and a CLF-CBF-QP-based low-level controller. Simulation results show that the vehicle could effectively respond to obstacles and generate a successful trajectory towards its goal.

20 December 2024 in Preprints.org
Verified
Preprint
Open access
PDF file
HCHaochong Chen
Haochong Chen
FZFengrui Zhang
Fengrui Zhang
Bilin Aksun GuvencBilin Aksun Guvenc
Bilin Aksun Guvenc
Collision avoidance and path planning are critical topics in autonomous vehicle development. This paper presents the progressive development of an optimization-based controller for autonomous vehicles using the Control Lyapunov Function - Control Barrier Function - Quadratic Programming (CLF – CBF - QP) approach. This framework enables the vehicle to navigate to its destination while avoiding obstacles. A unicycle model is utilized to incorporate vehicle dynamics. A series of simulations were conducted, starting with basic model-in-the-loop (MIL) non-real-time simulations, followed by real-time simulations. Multiple scenarios with different controller configurations and obstacle setups were tested, demonstrating the effectiveness of the proposed controllers in avoiding collisions. Real-time simulations in Simulink were used to demonstrate that the proposed controller could compute control actions for each state within a very short timestep, highlighting its computational efficiency. This efficiency underscores the potential for deploying the controller in real-world vehicle autonomous driving systems. Furthermore, we explored the feasibility of a hierarchical control framework comprising a Deep Reinforcement Learning (DRL) specifically Deep Q-Network (DQN) based high-level controller and a CLF-CBF-QP-based low-level controller. Simulation results show that the vehicle can effectively respond to obstacles and generate a successful trajectory towards its goal.