Finding consensus without computation
Anıl Özdemir, Melvin Gauci, Salomé Bonnet and Roderich Groß
Abstract
We propose a computation-free solution to the problem where a group of embodied agents need to collectively choose one of multiple options in their environment.
The agents are equipped with a line-of-sight sensor that extracts one ternary digit (trit) of information from the environment. At every time step, they directly map this information onto constant-value motor commands. They do not communicate with each other nor do they store run-time information.
We evaluate the proposed control strategy with both simulated and physical e-puck robots. The robots are expected to collectively choose, and move to, one of two options of equal value.
The simulation studies show that the strategy is robust against sensory noise, scalable to large swarm sizes, and generalises to the problem of choosing between unequal options.
The physical experiments (50 trials conducted with a group of 20 e-puck robots) show that the group achieves consensus in 96% of the trials.
Given the extremely low hardware requirements of the strategy, it opens up new possibilities for the design of swarms of robots that are small in size (≪ 10−3 m) and large in numbers (≫ 103).
Highlight video
This video is 10 times faster than real time.
Physical experiments
View the YouTube playlist showing all 50 physical experiment trials.