We build unmanned vehicles that can fly and drive without maps or GPS.
Our research goals are to build unmanned aerial vehicles (drones) that can fly without GPS through unmapped indoor environments, robots that can drive through unmapped cities, and to build social robots that can quickly learn what people want without being annoying or intrusive.
Such robots must be able to perform effectively with uncertain and limited knowledge of the world, be easily deployed in new environments and immediately start autonomous operations with no prior information. This engineering challenge requires algorithmic advances in decision-theoretic planning, statistical inference, and artificial intelligence. We focus on problems of planning and control in domains with uncertain models, using optimization, statistical estimation and machine learning to learn good plans and policies from experience
MIT CSAIL’s frugal deep-learning model infers the hidden physical properties of objects, then adapts to find the most stable grasps for robots in unstructured environments like homes and fulfillment centers.
