Experimental Evaluation of Teamwork in Many-Robot Systems (Demonstration)

The experimental evaluation of methods and techniquesfor teamwork in multi robot systems (MRS) is challenging.Experiments with multiple robots are very dicult to man-age [4] and thus the proposed approaches are seldom evalu-ated on real multi robot systems composed by several robots.Teamwork in MRS, especially when aiming at massiveexperiments, is often evaluated using abstract simulators,which typically focus on the communication model, but makevery rough assumptions on the behavior of the robots in theoperational environment. In these cases, it may happen thatthe simulation model is too abstract to provide convincingevidence that the results obtained in simulation, apply alsoto the real case. Obviously, the more complex each individ-ual robot is, the larger the distance between the simulationand the real case. Indeed, we have experienced that the per-formance of teamwork in MRS is deeply inuenced by theperformance of the robotic platform in the operational envi-ronment. Consequently, in order to bridge the existing gapwith real robots, we have focussed on simulators that areoriginally designed for robotic systems and provide a moreaccurate model of the performance of the robots. This ap-proach is challenging for a number of reasons. First of all,simulation tools are sometimes embedded in a software de-velopment framework, like for example Microsoft RoboticsCite as: