Director of study: Prof. Jacques Penders
Additional information: Open to all Students; Self funded

The aim is to develop group control by taking advantage of the interaction and cohesion within the group of robots. Distributed group control, as referred to here, is not a fully understood concept but nevertheless promises important future applications. It has inherent advantages such as parallel execution of multiple tasks, reliability and tolerance to single-point-of-failure (including failure of single robots).

All group members having the same sensor systems and thus the same information about the environment, and the same operating programs. These groups are called congenial [Penders et al., 1994]. The robots perform obstacle avoidance and goal finding, and these two actions generate group behaviours. Among themselves the robots create a dynamic environment, and we study the impact of the particular sensor systems or avoidance rules on the emerging behaviours. The algorithms applied by the robots can be based on artificial potential fields. This approach is relatively simple and therefore attractive when aiming for group modelling. However other approaches are useful as well.

Our robots are non-communicative. The argument for considering non-communicative groups is that the basic aim is to understand why and how interaction patterns develop. Many real-life situations in human society do not require explicit communication: in shopping centres people pushing their shopping trolleys generally pass each other without communicating; also commonly pedestrians do not communicate verbally when passing each other, although pedestrians do interact by emitting signals to indicate their purposes.

Haptic Human robot interaction, make a robot communicate with a person using the human's tactile and haptic senses.

For further information, please contact Prof. Jacques Penders