Gaoxiang Peng

Gaoxiang Peng

Auto-generated profile, based on publicly available information

Gaoxiang Peng

National Key Laboratory of Multispectral Information Intelligent Processing Technology, School of Artificial Intelligence and Automation, Huazhong University of Science and Technology, Wuhan, China

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

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

Representative Publications

View All Publications
29 August 2023 in Aerospace Science and Technology
Verified
Journal Article
Closed access
GPGaoxiang Peng
Gaoxiang Peng
BWBo Wang
Bo Wang
LLLei Liu
Lei Liu
HFHuijin Fan
Huijin Fan
ZCZhongtao Cheng
Zhongtao Cheng

15 August 2024 in IEEE Transactions on Aerospace and Electronic Systems
Verified
Journal Article
Closed access
GPGaoxiang Peng
Gaoxiang Peng
BWBo Wang
Bo Wang
LLLei Liu
Lei Liu
HFHuijin Fan
Huijin Fan
In this paper, by employing a modular strategy and hybrid action reinforcement learning (RL), an intelligent three-dimensional (3D) entry trajectory planning approach is proposed to achieve online autonomous trajectory generation for any initial location and target within a large reachable area. To reduce training difficulty and enhance task adaptability, the modular design approach decomposes the entry trajectory problem into the bank and angle of attack (AOA) modules. The environment, with complete 3D state space and entry dynamics, is established to train both modules, in which additional performance requirements are incorporated in the reward design, such as suppressing long-period oscillations and reducing reversal numbers. In the bank module, discrete-continuous hybrid action space is adopted to formulate the bank angle decision problem, which determines the sign and amplitude of the bank angle simultaneously, effectively avoiding the trade-off between RL exploration difficulties and trajectory accuracy. Specifically, Dueling Hybrid TD3 (DHTD3) is proposed to address the hybrid action problem and hence to train the 3D bank module. Subsequently, the AOA module is trained to adjust the AOA profile to meet flight capability requirements. The simulations highlight the ability of our proposed algorithm to achieve autonomous robust trajectory planning in expansive reachable areas.

3 July 2025 in IEEE Transactions on Aerospace and Electronic Systems
Verified
Journal Article
Closed access
GPGaoxiang Peng
Gaoxiang Peng
BWBo Wang
Bo Wang
LLLei Liu
Lei Liu
HFHuijin Fan
Huijin Fan
Normal reinforcement learning (RL) methods for entry guidance face challenges in environmental design and generalization due to the uncertainty of geographic constraint types and distributions. This paper presents an adaptive guidance algorithm that integrates modular RL with model-based rules to address entry guidance problems subject to complex geographical constraints such as no-fly zones and waypoints. Firstly, a normal trajectory module is developed based on modular RL to generate reference commands online. Then, focusing on geographical constraints, model-based rules are designed by employing both the virtual target method and the geometric trajectory planning. The virtual target method guides the normal trajectory module to meet the requirements for maneuverability of geographical constraints and terminal target. The geometric trajectory planning designs arc trajectories to ensure precise navigation to waypoints and avoidance of no-fly zones. Moreover, a maximum lateral module is designed to constrain the control commands derived from arc trajectories, ensuring their feasibility, such as satisfaction of path constraints. By integrating model-based rules with RL modules, the proposed algorithm does not require training for specific geographic constraint scenarios while avoiding computationally intensive iterations, enabling direct, real-time adaptation to varying geographical constraints. Extensive simulations demonstrate the adaptability and robustness of the proposed algorithm under various geographical constraints with different types and distributions.