CV

Download CV (PDF)

Education

• Washington University in St. Louis, St. Louis, MO, USA
  Thesis-Based M.S. in Electrical Engineering (Honors Program), 08/2024 – Present
• Hong Kong University of Science and Technology, Hong Kong
  B.Eng. in Computer Engineering, 08/2020 – 08/2023
• University of California, San Diego, San Diego, CA, USA
  Exchange Student, 09/2022 – 12/2022
• Washington University in St. Louis, St. Louis, MO, USA
  HKUST–WashU Joint Education Program, 08/2023 – 05/2024

Research experience

• Integrated Embodied Multi-Agent Task and Motion Planning Framework, Riverside, CA, USA (05/2025 – Present)
  Supervisor: Prof. Mingyu Cai
  Outcome: Deadlock-Free Hybrid RL-MAPF Framework for Zero-Shot Multi-Robot Navigation (L4DC 2026 submission; under review)
  • Developed an integrated multi-agent framework combining LLM-based task decomposition, RL navigation, and MAPF fallback for deadlock resolution.
  • Designed deadlock detection and an interface between decentralized RL control and a local MAPF solver to ensure feasible solutions in extreme cases.
  • Validated the framework in a near photo-realistic interactive simulator across diverse multi-agent task sequences.
• Human Behavior-Informed Framework for Safe Robot Motion Planning, St. Louis, MO, USA (09/2024 – 04/2025)
  Supervisors: Prof. Bruno Sinopoli, Prof. Yiannis Kantaros
  • Built a human action prediction system combining active probing of human characteristics with reachability analysis.
  • Designed an MPC planner that integrates inferred human type and reachable sets as constraints for uncertainty-aware safety.
  • Evaluated the system in driving simulation environments and compared against baseline planners.
• Human Trajectory Prediction in the Interaction between Human and Unmanned Vehicles, St. Louis, MO, USA (01/2024 – 05/2024)
  Supervisors: Prof. Bruno Sinopoli, Prof. Yiannis Kantaros
  • Developed deep learning-based human trajectory prediction and integrated predictions with robot path planning.
• Transfer Learning in Deep Reinforcement Learning: Self-Supervised Learning for Policy Adaptation, St. Louis, MO, USA (08/2023 – 04/2024)
  Supervisor: Prof. Bruno Sinopoli
  • Studied DQN, DDPG, and PPO and improved adaptation via self-supervised auxiliary tasks during deployment.
  • Validated on the DeepMind Control suite with improved generalization under visual modifications.

Internship and engineering experience

• Hong Kong Center for Construction Robotics, Hong Kong (06/2022 – 12/2022)
  Start-up firm co-founder

• Solar Panel Cleaning Robot System, Hong Kong (02/2023 – 08/2023)
  Supervisor: Prof. Ling Shi

• Development of Network Communication System for Locomotion Robots, Hong Kong (06/2021 – 08/2022)
  Supervisors: Prof. Ling Shi, Prof. Man Wong

Skills

• Programming: C, C++, Python, MATLAB
• Software & Tools: PyTorch, OpenAI Gym, Simulink, Solidworks, Keil, 3D Printing
• Language: Mandarin (native), English

Publications

Deadlock-Free Hybrid RL-MAPF Framework for Zero-Shot Multi-Robot Navigation Permalink

Haoyi Wang, Licheng Luo, Yiannis Kantaros, Bruno Sinopoli, Mingyu Cai. (2025). "Deadlock-Free Hybrid RL-MAPF Framework for Zero-Shot Multi-Robot Navigation." arXiv preprint arXiv:2511.22685. Submitted to L4DC 2026.

Talks

No talks listed yet.

Teaching

No teaching listed yet.

Service and leadership

• WashU Electrical & Systems Engineering Student Advisory Board (2023 – 2025) — International Student Ambassador
• HKUST RoboMaster Robotics Competition Team (2020 – 2021) — R&D member