Guanya Shi

About

Guanya Shi is a robotics and control researcher whose work spans aerial robotics, humanoid robot control, and the theoretical foundations of learning-based adaptive systems. He is best known for **Neural-Fly** (2022), a landmark contribution that combines meta-learning with adaptive control to enable agile UAV flight in strong, unpredictable winds — a paper that has accumulated over 230 citations and represents a significant step toward reliable autonomous aerial vehicles in real-world conditions. His research elegantly bridges classical control theory and modern machine learning: his **Online Meta-Adaptive Control** framework provides rigorous theoretical guarantees for adaptive nonlinear control under adversarial disturbances, while his work on **Deep Model Predictive Optimization** seeks to unify the flexibility of reinforcement learning with the structure of model-based planning. More recently, Shi has turned his attention to humanoid robotics, developing influential frameworks including **H2O** for real-time whole-body teleoperation and **HOVER** for versatile neural locomotion and manipulation control. His work on uncertainty quantification for pose estimation further reflects a consistent commitment to safe, deployable autonomy. Across these diverse areas, Shi's research is united by a compelling vision: building robots that learn rapidly, adapt robustly, and operate safely in the complexity of the real world.

Research Focus

Key Achievements

Top Papers

1. 1
   Neural-Fly enables rapid learning for agile flight in strong winds

   232 citations · 2022
2. 2
   Learning Human-to-Humanoid Real-Time Whole-Body Teleoperation

   68 citations · 2024
3. 3
   Fast Uncertainty Quantification for Deep Object Pose Estimation

   31 citations · 2021
4. 4
   The jumping mechanism of flea beetles (Coleoptera, Chrysomelidae, Alticini), its application to bionics and preliminary design for a robotic jumping leg

   22 citations · 2020
5. 5
   Meta-Adaptive Nonlinear Control: Theory and Algorithms

   19 citations · 2021
6. 6
   Full-Order Sampling-Based MPC for Torque-Level Locomotion Control via Diffusion-Style Annealing

   12 citations · 2025
7. 7
   HOVER: Versatile Neural Whole-Body Controller for Humanoid Robots

   12 citations · 2025
8. 8
   Safe Deep Policy Adaptation

   8 citations · 2024
9. 9
   Deep Model Predictive Optimization

   7 citations · 2024
10. 10
    Humanoid Locomotion and Manipulation: Current Progress and Challenges in Control, Planning, and Learning

    7 citations · 2025

Key Collaborators

Contact & Links