Embodied evolution for collective indoor surveillance and location

Abstract

This work is devoted with the application of a canonical Embodied Evolution algorithm in a collective task in which a fleet of Micro Aerial Vehicles (MAVs) have to survey an indoor scenario. The MAVs need to locate themselves to keep track of their trajectories and to share this information with other robots. This localization is performed using the IMU, artificial landmarks that can be sensed using the onboard camera and the position of other MAVs in sight. The accuracy in the decentralized location of each MAV has been included as a part of the problem to solve. Therefore, the collective control system is in charge of organizing the MAVs in the scenario in order to increase the accuracy of the fleet location, and consequently, the speed at which a new point of interest is reached.