Mohsen Raoufi (Science of Intelligence), From State Estimation to Collective Estimation, and from Individuality to Complexity in Swarm Robotic Systems

Using swarm optimization algorithms as heuristic solutions in various engineering problems, including the state estimation of nonlinear systems, has been an inspiration to me for my SCIoI project. We started our project by studying the “Wisdom of Crowds” effect, i.e. the notion that the average of many imperfect estimations, under the right conditions, can potentially yield a perfect estimation. However, achieving an accurate estimation in a distributed manner requires individuals to finely balance many tradeoffs, e.g. the exploration-exploitation trade-off. In addition to discussing a few of these tradeoffs in my talk, I will demonstrate how the interaction of agents leads to more complex behavior, particularly when the connectivity of the communication network is limited. For instance, this limitation results in the emergence of “echo-chambers” in the collectives.

In the second part, I will talk about the inter-individual variations we observed during real robot experiments with Kilobots. This side project opened up a new dimension to the complexity of the individual and collective behaviors in swarm robotic systems.

This talk will take place in person at SCIoI.

Photo by Alina Grubnyak on Unsplash