A Statistical Framework for Estimating the Success Rate of Liquid-Mimetic Swarms

Abstract

This chapter presents a novel statistical framework for risk assessment of robotic swarm operations. This research has arisen from the crucial problem of predicting (before actual deployment) what portion of the swarm will succeed in accomplishing its mission. Our scouts framework uses a small group of expendable “scout” agents to predict the success probability for the entire swarm, thereby preventing a loss of a large portion of the swarm. The framework is based on two probability formulas—the standard Bernoulli trials formula and a novel Bayesian formula that was derived from fundamental probabilistic principles. For experimental testing, the framework is applied to a navigation task, where simulated agents move to a goal location through a field of obstacles.