Robot Embodiment

Robot embodiment research investigates how a robot's physical form influences its capabilities and interactions. Current work focuses on developing control policies that generalize across diverse robot morphologies, employing architectures like transformers and leveraging large datasets to train single policies capable of controlling multiple robot types for various tasks (manipulation, locomotion, etc.). This research is crucial for advancing robust and adaptable robots, impacting fields ranging from manufacturing and logistics to human-robot collaboration and assistive technologies by enabling the creation of more versatile and efficient robotic systems.