Mapping With Uniformly Controlled Stochastic Swarms
Multi-robot mapping of static worlds has presented a challenge of combining sensor data from each robot to generate accurate maps. Although much progress has been made towards solving this problem, this method is inapplicable to robots without onboard computation. For medical purposes such as mapping vasculature, this limitation is necessary to make the robots sufficiently small. We propose an alternative mapping technique that generates a grid-based Bayesian probability map using physical observation instead of sensor inputs. Our method utilizes a swarm of particles that is controlled by a uniform input but is also affected by random noise. To answer the question of whether random movement significantly affects the efficiency of existing algorithms, we ran simulations of swarm exploration and recorded their speed and accuracy. These results could help us refine current algorithms to better accommodate medical imaging environments.