Xu Wang

Xu Wang

Auto-generated profile, based on publicly available information

Xu Wang

Xian Key Laboratory of Big Data and Intelligent Vision, Xidian University, Xian, China

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

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

Representative Publications

View All Publications
28 September 2022 in IEEE Transactions on Neural Networks and Learning Systems
Verified
Journal Article
Closed access
XWXu Wang
Xu Wang
SWSen Wang
Sen Wang
XLXingxing Liang
Xingxing Liang
DZDawei Zhao
Dawei Zhao
JHJincai Huang
Jincai Huang
Qiguang MiaoQiguang Miao
Qiguang Miao
Deep reinforcement learning (DRL) integrates the feature representation ability of deep learning with the decision-making ability of reinforcement learning so that it can achieve powerful end-to-end learning control capabilities. In the past decade, DRL has made substantial advances in many tasks that require perceiving high-dimensional input and making optimal or near-optimal decisions. However, there are still many challenging problems in the theory and applications of DRL, especially in learning control tasks with limited samples, sparse rewards, and multiple agents. Researchers have proposed various solutions and new theories to solve these problems and promote the development of DRL. In addition, deep learning has stimulated the further development of many subfields of reinforcement learning, such as hierarchical reinforcement learning (HRL), multiagent reinforcement learning, and imitation learning. This article gives a comprehensive overview of the fundamental theories, key algorithms, and primary research domains of DRL. In addition to value-based and policy-based DRL algorithms, the advances in maximum entropy-based DRL are summarized. The future research topics of DRL are also analyzed and discussed.

27 December 2023 in IEEE Transactions on Intelligent Vehicles
Verified
Journal Article
Closed access
XWXu Wang
Xu Wang
ESErke Shang
Erke Shang
BDBin Dai
Bin Dai
Qiguang MiaoQiguang Miao
Qiguang Miao
YNYiming Nie
Yiming Nie
Path planning is a critical aspect of autonomous driving. Traditional methods have excelled in urban environments but struggle off-road due to unpredictable conditions. This paper introduces a Deep Reinforcement Learning-based off-road path planning approach called PPOAC by training the agent in a low-dimensional simulator built by Occupancy Grid Maps (OGMs). A self-supervised auxiliary task is proposed to help the agent to learn environmental knowledge. Moreover, an adaptive curriculum learning method is introduced to enhance the agent's training efficiency and generalization ability. The agent is trained and evaluated in real-world data, and experimental results demonstrate that it can obtain traversable and safe paths while generalizing effectively to unknown environments.

20 February 2024 in IEEE Transactions on Intelligent Vehicles
Verified
Journal Article
Closed access
XWXu Wang
Xu Wang
BDBin Dai
Bin Dai
Qiguang MiaoQiguang Miao
Qiguang Miao
YNYiming Nie
Yiming Nie
ESErke Shang
Erke Shang
This paper introduces a data augmentation method based on control theory and rule constraints, aiming to address the insufficient coverage issue in existing vehicle trajectory datasets. Grounded in control theory, the method combines the motion and safety rules of driving vehicles for data augmentation. It exhibits high interpretability, credibility, a minimal and controllable set of parameters, while maintaining physical feasibility. Additionally, a one-step Markov Decision Process is introduced to model the car following task as a short-term decision problem, adapting to unpredictable variations in the generated trajectories. Experimental results demonstrate that the proposed rule-constrained data augmentation method significantly enhances the model's performance in addressing car following issues, showcasing strong adaptability and generalization capabilities.