Advancing fundamental machine learning algorithms and methodologies that drive innovation across robotics and artificial intelligence
Novel reinforcement learning algorithms that improve sample efficiency, stability, and performance across complex robotic tasks. Our research focuses on developing robust policy optimization methods that can handle high-dimensional state and action spaces.
Multi-level learning frameworks that decompose complex tasks into manageable subtasks. Our hierarchical approaches enable robots to learn long-horizon behaviors more efficiently and transfer skills across different domains.
Advanced models that integrate visual and linguistic information for enhanced robotic understanding and interaction. Our research enables robots to interpret complex visual scenes and follow natural language instructions.
Intelligent algorithms that combine information from multiple sensors including vision, tactile, proprioceptive, and auditory systems. Our fusion methods enable robust perception and decision-making in complex environments.
Methods for quantifying and managing uncertainty in machine learning models. Our research focuses on developing reliable AI systems that can assess their own confidence and make safe decisions in uncertain environments.
Developing transparent and interpretable AI systems that can explain their decisions and reasoning processes. Our work enables better human-AI collaboration and builds trust in autonomous systems.