Swarm control :

Abstract

There were two problems which were worked upon: one was steering and splitting of shepherding and sheep swarm using predator agents of autonomous ground vehicles; and the other was to find the location of the source of a hazardous substance which emits odor like nuclear waste, harmful chemical, gas leaks and etc. The latter problem is motivated by the fact that using humans for this task is dangerous because of the harmful nature of the gases emitted. A swarm of robots can detect the source efficiently. Swarms can be inspired from: biology(a swarm of fishes), from physics (Physicomimetics swarm) and follow a particular formation (circular). The former problem was a build up to the work done in the earlier semesters, steering of robots having a heading angle using bio inspired swarms gives strong proof that these types of swarms can be used in real world as well. Gazebo, a real world simulator was used for simulating this. Source localization using a single robot has a dependency on sensors (plume/gas), the movement of the robot is in the gradient direction. Challenges are due to turbulent environments, a lot of time to find the source and single point failure Therefore, multiple robots to localise source efficiently with fault tolerance.