Research Statement

I develop trustworthy and reliable AI algorithms for robotics applications, focusing on the intersection of computer vision, machine learning, and autonomous systems. My work aims to create failure-resistant robotic systems that can operate safely in real-world environments.

Research Themes

3D Scene Understanding

Developing robust object tracking and scene reasoning algorithms for long-term autonomy. I work on contextual object association and tracking in household environments, addressing challenges in real-time classification, segmentation, and localization. Collaboration with Amazon Lab126.

Object Tracking Scene Reasoning Real-time Perception

Risk-Aware Navigation

Creating safe and reliable navigation systems that leverage rich visual information for autonomous planning. I focus on handling environmental symmetries, motion blur, and high-frequency demands through observability analysis and constraint-based control methods for ground and aerial robots.

SLAM Safety Constraints Observability Analysis

Safe Uncertainty-Aware Robot Learning

Bridging the gap between reinforcement learning and control theory to develop generalizable decision-making systems. I design reward functions that guarantee safe agent actions while exploring sim-to-real transfer capabilities for robust real-world deployment.

Safe RL Sim-to-Real Control Theory

View Publications on Google Scholar