Research
My work blends multi-agent reinforcement learning, distributed robotics, and embedded autonomy, focusing on reliable coordination strategies for autonomous robot teams.
Research Assistant (RA) · HKUST(GZ)Jul. – Oct. 2026
Multi-agent Reinforcement Learning with Communication
Research assistant at the Hong Kong University of Science and Technology (Guangzhou) after completing the master's degree, focusing on multi-agent reinforcement learning with communication.
Methods & contributions
- Research focus: the integration of inter-agent communication and multi-agent policy learning for cooperative decision-making.
MARL / Communication
Core Researcher2024 – 2026
Intelligent Swarm Pursuit-Evasion under Incomplete Information
Studied MARL policies for aerial swarms that adapt between local exploration and burst communication under incomplete information.
Methods & contributions
Prof. Yuanshi Zheng · National Natural Science Foundation of China
- Raised pursuit success to 91% in the NP-1E pursuit-evasion environment.
- Filed multiple published invention patents on UAV coordination and sparse-reward exploration.
MARL / UAV
Research Engineer2025
Autonomous UAV–Ground Vehicle Navigation
Developing navigation stack combining RGB, LiDAR, odometry for indoor logistics.
Methods & contributions
Meituan Academy of Robotics Shenzhen · Industry Collaboration
- Implemented dual control modes with constraint-aware planning.
- Achieved 10 cm arrival accuracy in dynamic test arenas.
Navigation / ROS
Graduate Researcher · Xidian University2025
Multi-UAV Formation Control and Simulation
Studied multi-UAV formation control during the master's programme, implementing square, circular and flocking formations with collision avoidance in MATLAB and Python simulations.
Methods & contributions
- Implemented 2D and 3D square formations using artificial potential fields and a leader–follower structure, and ring-topology formations using distributed consensus control.
- Reproduced and extended an existing flocking implementation with target tracking, obstacle avoidance and trajectory visualisation, building on MSN-Flocking-Formation-Control.
- Analysed simulated trajectories, velocity evolution, connectivity and centroid motion; published code and visualisation tools in three GitHub repositories.
Formation Control / MATLAB / Python / Simulation
Team Lead2022 – 2023
Visual-Inertial SLAM for Autonomous UAVs
Integrated IMU and vision for warehouse-scale SLAM and obstacle-aware planning.
Methods & contributions
Assoc. Prof. Li Jun · Undergraduate Thesis
- Maintained <0.3 m drift over 10-minute flights.
- Automated waypoint navigation through cluttered aisles.
SLAM / UAV / Navigation
Research Assistant (RA) · Beihang UniversityJun. – Sep. 2022
TOTAL: Multi-Corner Timing Optimisation with Transfer and Active Learning
Contributed to TOTAL, a machine-learning framework for multi-corner static timing analysis and optimisation of integrated circuits, with work on data preprocessing, model validation and benchmark experiments.
Methods & contributions
Advisor: Wei W. Xing
- Supported Gaussian-process-based timing prediction using transfer and active learning, including experimental data preparation and model validation for the 15 nm technology node.
- Ran experiments on the b17, b18 and b19 benchmarks, collected results and analysed critical-path prediction across timing optimisation iterations.
- Contributed experimental data and validation results to “TOTAL: Multi-Corners Timing Optimization Based on Transfer and Active Learning”, published at ACM/IEEE DAC 2023.
EDA / Gaussian Processes / Transfer Learning / Active Learning