Research Interests

My research interests are information processing in distributed and neural networks. To optimise information processing in large networks, local learning methods can help to improve performance in a self-organised way. This self-organisation may be guided by, for example, information-theoretic principles.

I am also interested in the representation of sensory information, and the emergence of coding for both technical and biological systems, and the application of bio-inspired approaches of coding to representation and communication.

My current work involves the development of new architectures and learning algorithms, as well as their applications to real world problems, for example in fault detection in distributed systems, such as smart electrical grids or sensor networks. Wider research interests include representation, machine learning, autonomous development, and planning in physical multiagent systems.

In the past, I have made contributions to RoboCup, an international research and education initiative with the goal is to foster artificial intelligence and robotics research. In particular, I have been lead initiator of the 3D Soccer Simulation league, and one of the main developers of the simulation system, SimSpark. SimSpark facilitates researchers in MAS to create different kinds of multiagent simulations. It is designed to create agents with different morphologies, actuators and sensors in various environments. The system is currently being used and extended (by others) to run the annual RoboCup Soccer Simulation world championship.

Other contributions in this domain include plan- and also behavior-based approaches to multi-agent control. One of these, Inverse Steering Behaviors, is being used in a commercial application for emergency egress simulation.