Anyone who has ever had a child knows that it is nearly impossible to control multiple children at once. Researchers have had difficulty getting swarms of robots to work together unless they carefully choreograph their interactions, like airplanes in formation, with increasingly sophisticated components and algorithms. With the robots on hand, however, basic, unreliable, and unaware of sophisticated programming mechanisms for organized behavior, what can be done with certainty?
Daniel Goldman, Dunn Family Professor of Physics and ADVANCE Professor of Computing, led a team of researchers to demonstrate that even the most basic robots can perform tasks well beyond the capabilities of a single or a small team. The team achieved their goal of completing tasks by using “dumb robots” (basically mobile granular particles) so efficiently that they eliminated all sensors, communication, memory, and computation — achieved by exploiting the robot’s physical characteristics.
According to Randall, the team’s simple BOBbots were named for the granular physics pioneer Bob Behringer. In order to study aspects of the system inaccessible in the lab, Georgia Tech physics student Shengkai Li studied the system with precise computer simulations. The team’s experimentation was supplemented with computer simulations.
According to Goldman, BOBbots, though simple machines, can cooperate to clear debris that is too heavy for one to shift alone. Most people build increasingly complex and expensive robots to ensure coordination, but we wanted to find out how simple robots could accomplish complex tasks.”
Scientists at the University of Chicago developed a model that describes particles moving on a chessboard, and published their results in the journal Science Advances on April 23, 2021. An abstract mathematical model of the BOBbots was developed based on ideas from probability theory, statistical physics, and stochastic algorithms, they found that as magnetic interactions increased, the model underwent a phase change, from being dispersed to aggregating into large, compact clusters, similar to phase changes seen in physical systems such as water and ice.
The rigorous analysis revealed not only how the BOBbots are built, but also how our algorithm has revealed an inherent strength, which has, in turn, allowed some of them to be defective or unpredictable,” says Randall, who is also an IT professor and Georgia Tech’s Associate Professor of Mathematics.
BOBbots and their collective motion:
The movement and interactions of BOBbots were designed to incorporate the salient features of the stochastic algorithm while replacing all sensory inputs, communication, and computation with physical morphology. A base of elastic brushes connects to a motor with an eccentric rotating mass vibration (ERM). The brushes convert vibrations caused by the rotation of the ERM into locomotion. As a result of the asymmetry of our propulsion mechanism design, the BOBbots (43) traverse largely circular trajectories that are randomized through their initial conditions, although unlike SOPS particles they occur with a constant velocity. Material and Methods provide more details.
The same as in the aggregation algorithm that discourages participants from moving away from locations with a lot of neighbours, each BOBbot has loose magnets contained inside shells that always orient themselves to the most attractive BOBbots nearby. The probability of a bot being removed from its neighbours is negatively related to the attractive strength generated by the number of magnets engaged, with the probability of motion corresponding geometrically and inversely to the number of neighbours.
Designing an arena within defined boundaries presented a challenge both in experiments and simulations. These BOBbots may stay at the boundary or in a corner, affecting system dynamics by encouraging aggregate agglomerations to form where they would not have otherwise. The Team of researchers used uniform airflow to push BOBbots away from the boundary to solve these problems and they also used sim simulations to show how the flow affects BOBbot behavior.