Negative Updating Combined with Opinion Pooling in the Best-of-n Problem in Swarm Robotics

There is a need for effective collective decision making in decentralised multi-agent and robotic systems. This paper introduces a novel approach to the best-of-n decision problem with large n. It utilises negative feedback obtained from direct pairwise comparison of options and evidence preserving opinion pooling. We present agent-based simulation experiments that explore the effects of pool size and the number of options on the speed of consensus. Robotic simulation experiments are then used to investigate the potential of the approach as a method for solving the best-of-n decision problem in swarm robotic applications. Overall, the results suggest that the proposed approach is highly scalable with regards to n.