Current Projects

The objective of the FN project "A Methodology for Collective Robotics Design", a collaboration between LAMI (EPF, Lausanne) and IDSIA (Lugano), is to devise a methodology to design teams of robots in view of preestablished tasks. In the previous part of the project, we have studied the team behaviour synthesis problem both through explicit programming and learning. The programming approach is motivated by the observation that collective intelligence in animal societies emerges from the interaction of individuals whose behaviour seems to be innate, whereas learning enables the adaptation of behaviour to changing task conditions. We believe that the integration of programming techniques with learning methods is highly desirable and functional to the goal of achieving real, self-programming robot teams. Our research work includes an important experimental verification part. Specific tools necessary for the these experiments, in particular for those which require learning, have been developed and currently tested at LAMI.

Keywords: Collective Robotics, Robot Programming, Robot Learning, Modelling.

The objective of this thesis is to study the collective locomotion of a group of autonomous mobile robots, without external supervision, in a complex, dynamically changing environment. Different strategies, inspired from biological models, will be studied. We will propose algorithms and hardware implementations which will allow to optimise locomotion.

The first part of this research will be conducted using a multiple robot simulation program developed for the purpose. The second part will involve groups of real robots.

This research is supported by the FN.

Keywords: Collective Robotics, Group Locomotion, Navigation, Self-Optimisation.

Some missions given to robot teams include several activities to be carried on concurrently. The teams would gain in efficiency if they could decide by themselves how to allocate their manpower between the different activities, and adjust this allocation according to the mission progress. How far can this be done without central control? Ants, for instance, have found amazing ways to dynamically distribute their workers according to the changing needs of their colony, without central control. The question is studied both with a computer simulation and with a team of Khepera robots.

Keywords: Collective Robotics, Task Allocation, Division of Labor, Self-Optimisation.

All the experiments are inspired by the behaviour of social insects and are carried out with a group of Khepera robots (developed at the LAMI and supported now by K-Team SA) as basic platform.