Swarm Robotics: Harvesting of Moving Swarms Represented by a Markov Process
This project presents methods of harvesting moving swarm agents, with two different experimental applications. We investigated motion planning for one or more robot(s). These methods differ from traditional motion planning problems because the agents move. This allows areas that were previously cleared to become recontaminated. The movement of agents is represented by a Markov process that encodes the agents' preferred regions and their speed of motion. There are two different categories of controllers presented in this project, one for single robot applications and a second for multi-robot applications. We conducted experiments using the single robot controllers. For the first application we studied a destructive survey of mosquitoes. We researched behavior and other characteristic of mosquitoes to design an electrified screen that can be carried on a UAV. For the second application we studied the harvesting of moving swarms using an autonomous robotic boat with an acoustic larvicide unit in a body of water.