On-Line Coordination Tasks for Multi-robot Systems Using Adaptive Informative Sampling

Robotic platforms have continued to advance in autonomous capabilities and as such, have been deployed in increasingly more complex scenarios. In this work, we look to address the utilization of multiple robots to take Received Signal Strength Indication (RSSI) measurements within an environment that has been mapped a priori and contains many radio transmitters. The sensor measurements are used to build a Gaussian Process Model of the maximum RSSI across all transmitters in the environment. The algorithm evaluates points of interest within the environment and directs robots based on the highest expected information gain per meter of travel. To coordinate multiple robots, the environment is divided into partitions based on the expected travel cost to visit each evaluation point within the partition. By assigning each robot to a different partition, path disruption is reduced, and exploration is more efficient than exploring random areas. The algorithm was tested and evaluated in both in simulation and on real-world platforms.