My areas of interest include robotics, reinforcement learning, and learning-based control, with a particular focus on legged robotic systems, including quadrupeds and humanoids. My research has primarily focused on energy-efficient locomotion, adaptive gait control, fall recovery, and robust sim-to-real transfer.
I have experience with physics simulation environments such as Isaac Gym, MuJoCo, and PyBullet, along with ROS 2 for robotic software development, simulation, and system integration. I am currently expanding my simulation and learning-based control workflows toward NVIDIA Isaac Lab and Isaac Sim, particularly for humanoid robotics and sim-to-real applications.
My broader research interests include:
- Model-based reinforcement learning
- Learned world models
- Uncertainty-aware control
- Imitation learning
- Generative action policies, including diffusion and flow matching
- Vision-Language-Action (VLA) models
- Multimodal foundation models
I am particularly interested in predictive and adaptive control frameworks that use learned dynamics and uncertainty estimation to anticipate instability, detect unfamiliar operating conditions, and enable proactive adaptation or recovery.