Robotics and Autonomy
Our team is pioneering high-risk, high-reward research in revolutionary robotic systems, robotic intuition, robust robotic super agility, and teaming and trust.
Program Manager: Reed Young
Revolutionary Robotic Systems
Advancing contributing capabilities to significantly improve robot system capabilities
The capability of a robotic system is very much a function of the independent capabilities of its component capabilities, including mobility, perception, intelligence, awareness, communications, and state space characterization. Research here centers on significantly advancing component capabilities to overcome inherent shortcomings that up to now have limited functionality in complex environments.
Enabling robots with human-quality decision-making
Robotic systems must be able to perform many tasks in highly dynamic and unstructured environments. This capability can be enabled by developing algorithms that allow the robot to continually learn, adapt, predict, and reason about how its actions will affect the physical world. In addition, robots must be able to learn generalized concepts that can transfer across tasks and environments. This intuition is what will enable them to choose reasonable actions even in situations with high uncertainty.
Robust Robotic Super Agility
Enabling air and ground platforms with an ability to maneuver aggressively through complex control
Today’s robots fall far short of the speed and robust agility required for battlefield dominance. This is not because they are physically incapable of such performance but because their control systems cannot reason intelligently about these complex dynamic regimes where underactuation dominates. We believe advances in three areas—algorithmic control theory, learning for control, and output feedback—are required to achieve robust robotic super agility.
Teaming and Trust
Optimizing human-machine teaming, including assurance considerations
A new paradigm is necessary that results in an adequate level of confidence and assurance that a system is behaving in an acceptable and expected way. Further, any human associated with the system (e.g., a teammate) will have to be able to attain a psychologically based level of trust enabled through an understanding of the underlying capabilities, experience operating with the autonomous system, and real-time state awareness.
- Loyal, Unafraid and Unmanned (Armada International, December 14, 2020)
- [Video] Scientists at Work: Teaching Robots to Think (PEWtrusts.org, April 12, 2019)
- Plays Well with Humans (Johns Hopkins Magazine, Winter 2019)