Kunkun Wang

Kunkun Wang

Auto-generated profile, based on publicly available information

Kunkun Wang

College of Information Engineering, Zhejiang University of Technology, Hangzhou, China

Research Focus & Metrics

Academic Career Length

1Year (2026-Present)

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

Representative Publications

View All Publications
8 May 2026 in IEEE Transactions on Automation Science and Engineering
Verified
Journal Article
Closed access
JWJunxiao Wang
Junxiao Wang
HCHaochi Chen
Haochi Chen
KWKunkun Wang
Kunkun Wang
LGLE Gao
LE Gao
JYJun Yang
Jun Yang
In this paper, a novel deep reinforcement learning based navigation policy for mobile robot is proposed to address the challenges in multi-dynamic obstacles environments. Firstly, this method designs a risk perception function to evaluate the collision probability (CP) between robot and dynamic obstacles. Then, the observation space with risk perception is designed for the robot with a sense of the danger level between dynamic obstacles. In order to guide the robot to actively avoid such high-risk obstacles, the velocity obstacles(VO)-based reward is used to find desired direction angle only considering these critical obstacles. In addition, a novel safety constraint is formulated based on control barrier function (CBF) theory, it could adaptively adjust the safety distance according to the risk level of dynamic obstacles. The policy is then optimized within a CBF-guided training framework to enhance safety and adaptability in multi-dynamic obstacles environments. A series of simulation and real-world experiments demonstrate that the proposed policy achieves superior performance in both safety and efficiency compared with state-of-the-art methods, and it is well-suited for safety-critical navigation tasks in dense, dynamic environments such as indoor service and warehouse logistics scenarios.