Vision: Today’s robots are mainly confined to structured and predictable environments, such as carefully engineered factory floors. I envision a future where artificially intelligent robots, whether manipulators or mobile platforms, operate seamlessly in everyday settings like homes and offices. This demands innovations in both the theoretical foundations and practical systems throughout the robot technology stack.

Focus: My research advances robotics in unstructured environments by focusing on two essential aspects of autonomy: the ability to see (robot perception) and the ability to act (motion generation), along with their interaction. Robot perception seeks to extract meaningful information from sensor data, while motion generation leverages these representations to enable robots to interact naturally with their surroundings. 

Robotics is Interdisciplinary: To function effectively in the diverse and unpredictable real world, robots must learn from data, interpret their surroundings, reason about potential outcomes, and adapt their behaviour accordingly. My work develops robust machine learning methods for perception and motion generation while incorporating structure to reason about the system’s confidence and theoretical guarantees. I draw on principles from probabilistic modelling to quantify uncertainty, control theory to understand dynamic systems, and 3D vision to maintain geometric consistency in representations.