A Control Framework for Autonomous Physical Systems: Observation, Modeling, Prediction, and Planning

This research aims to record the past experience of automated systems (robots)in the form of generative models that support state predictions into thefuture. We argue that background knowledge can be used to: compensate for missing observations; direct active information gathering; monitor future exposure to risk, avoid actions (and states) that lead to unrecoverable future states; and support a large variety of learning and planning algorithms. Our goal is to create new technologies that respond actively to uncertainty that threatens safety and performance on mission. The research will be conducted by the Laboratory of Perceptual Robotics at the University of Massachusetts Amherst---a leader in the hierarchical skills for autonomous robots, constraint satisfaction in systems that learn by exploration, and the representation of control knowledge. Our group has access to several state of the art bimanual mobile manipulators that we plan to use in an experimental evaluation of learning and planning in this representation in the context of multi-objective activites involving autonomous robot grasping and manipulation with a bimanual mobile manipulator.

We argue that background knowledge can be used to: compensate for missing observations; direct active information gathering; monitor future exposure to risk, avoid actions (and states) that lead to unrecoverable future states; and support a large variety of learning and planning algorithms. Our goal is to create new technologies that respond actively to uncertainty that threatens safety and performance on mission.

This is early/mid-stage (TRL 4) — the most realistic path in is NASA SBIR/STTR, which funds small businesses and research institutions to develop technology aligned with NASA's needs (equity-free, phased funding). Check whether a current SBIR/STTR solicitation topic overlaps with this project's technology area, or contact the project directly (above) to ask.

None of these are guaranteed paths for this specific project — TechPort itself doesn't have an "apply" button. Reaching out to the contact(s) above with a specific question is usually the fastest way to find out what's actually open.