Welcome to the Robust Autonomy and Decisions Group website
The Robust Autonomy and Decisions group at the Institute of Perception, Action and Behaviour is directed by Dr Subramanian Ramamoorthy.
Our research is centred on the problem of autonomous learning and decision making over time and under uncertainty by single or multiple long-lived agents interacting with a continually changing world, typically involving large state/action spaces, strategic objectives and sophisticated task specifications/constraints.
This work is motivated by applications involving autonomous robotics (coordinated motion strategies for humanoid robots in autonomous or mixed-initiative scenarios; robotic soccer, through our RoboCup team - Edinferno) and autonomous agents in virtual environments (information aggregation, risk analysis and decision making in electronic markets).
A key conceptual issue that underlies all of these problems is that of constructively developing representations and task semantics, and understanding their relationships to techniques for decision making and learning. With this in mind, methodologically, we draw on various mathematical techniques (including from control, dynamical systems and game theory, geometry, topology and probabilistic methods) to model problem domains, understand representations, mechanisms for representation refinement or discovery and various properties of the resulting problem and solution space. Within these spaces, we explore online and incremental learning algorithms for decision making.