Jiaxuan Wei
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Jiaxuan Wei has a total of 6 co-authors. Here are the top co-authors by number of shared publications:
- Jahid Shahidul Islam
School of Information and Software Engineering, University of Electronic Science and Technology of China, Chengdu, Sichuan, China
1 shared publication - Jiahao Wang
School of Software Engineering, University of Electronic Science and Technology of China (UESTC), Chengdu 611731, China, China
1 shared publication - John Idriss Lahai
University of New England
1 shared publication Ryvel Timothy Stamber
School of Software Engineering, University of Electronic Science and Technology of China (UESTC), Chengdu, P.R.China
1 shared publication- Saio Alusine Marrah
University of Electronic Science and Technology of China, Sichuan P.R., 611731 Chengdu, China
1 shared publication Wu Jinming
School of Software Engineering, University of Electronic Science and Technology of China (UESTC), Chengdu, P.R.China
1 shared publication
Latest Publications
- Open Access
- 8 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. PP, 1
Adaptive Sensor Fusion-Enhanced Deep Reinforcement Learning for Fault-Tolerant Autonomous Control
This study presents a deep reinforcement learning–based control architecture with an adaptive sensor fusion module for complex, sensor-driven autonomous systems. The agent, trained in simulation using PPO with domain randomization, learns robust policies from multi-modal inputs (LiDAR, IMU, camera). Rigorous evaluation (10 random seeds) against PID and MPC baselines under nominal and fault-injection scenarios demonstrates statistically significant improvements in tracking accuracy, rise time, and fault recovery. The adaptive fusion mechanism, formally defined using residual-based attention, reduces RMSE by 56% and enables recovery from sensor faults in 0.9s. Results validate the feasibility of DRL in safety-critical, real-time embedded environments.
- 8 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. PP, 1
Open AccessAdaptive Sensor Fusion-Enhanced Deep Reinforcement Learning for Fault-Tolerant Autonomous Control
This study presents a deep reinforcement learning–based control architecture with an adaptive sensor fusion module for complex, sensor-driven autonomous systems. The agent, trained in simulation using PPO with domain randomization, learns robust policies from multi-modal inputs (LiDAR, IMU, camera). Rigorous evaluation (10 random seeds) against PID and MPC baselines under nominal and fault-injection scenarios demonstrates statistically significant improvements in tracking accuracy, rise time, and fault recovery. The adaptive fusion mechanism, formally defined using residual-based attention, reduces RMSE by 56% and enables recovery from sensor faults in 0.9s. Results validate the feasibility of DRL in safety-critical, real-time embedded environments.