Overview
I develop algorithms for human-robot collaboration and shared autonomy, where robots intelligently interact with humans to accomplish complex tasks. My work sits at the intersection of robotics, control theory, planning, and machine learning.
The central question driving my research is: How can we enable robots to work safely and intuitively alongside humans in dynamic, unstructured environments?
Research Areas
Human-Robot Collaboration
Developing algorithms that enable robots to anticipate human actions, adapt to human preferences, and coordinate seamlessly with human partners in collaborative tasks. This involves understanding human motion patterns, predicting intentions, and generating robot behaviors that complement human actions.
Shared Autonomy
Creating systems where control is dynamically shared between humans and robots. The robot provides assistance when needed while respecting human agency and expertise. This requires real-time assessment of task difficulty, human performance, and optimal assistance levels.
Motion Prediction & Planning
Building models that predict human motion in real-time to enable proactive robot behavior. This involves combining classical planning approaches with modern machine learning techniques to handle uncertainty and variability in human behavior.
Teleoperation & Control
Designing intuitive teleoperation interfaces and control schemes for complex robotic systems, particularly humanoid robots. Focus on reducing operator cognitive load while maintaining fine control over robot actions.
Safety in Human-Robot Interaction
Ensuring safe operation in shared spaces through predictive safety mechanisms, collision avoidance, and ergonomic considerations. Developing formal methods to verify safety properties in collaborative scenarios.
Current Research
At the People, AI, & Robots (PAIR) Lab directed by Animesh Garg and the Robot Vision and Learning (RVL) Lab directed by Florian Shkurti, I am working on:
- • Learning-based approaches for human motion prediction in collaborative tasks
- • Safe shared autonomy frameworks with formal guarantees
- • Adaptive robot behaviors that respond to human skill levels and preferences
- • Integration of classical control with modern learning-based methods