Combining the multi-agent flood algorithm with frontier-based exploration in search & rescue applications

Abstract

The Multi-Agent Flood algorithm is an algorithm to control multiple autonomous agents to explore unknown terrain and to find p oints of interest in t he newly uncovered territory. It does so by using a mix of indirect and direct communication. The direct communication via radio signals allows for a faster exchange of data between the agents. This layered communication model allows for a robust data transfer that is needed in SAR scenarios, as the algorithm can still use the indirect part of the model if the direct part is not available anymore. Currently the exploration done by the agents is decided randomly without any intelligent input based on the gathered data. The agents will simply try to move along a line until they will hit an obstacle or a point of interest is found. This work introduces ways to steer the autonomous agents based on the collected data, additionally this data will be used to decide which part of the map should be explored next. Another optimization of the performance is also the increased cooperation between the agents, as the agents will also exchange these destination points, thus reducing the number of agents which will explore the same part of the area, leading to a greater spread of the agents through the terrain.