A Framework for Distributed Mobile Robotics (Paperback)

This thesis presents a means of coordinating a many-to-many distributed marsupial robotic team in order to maximize the amount of time that the robots deployed by the marsupials are able to effectively carry out their mission. Solutions to this problem are important when distributed robotic teams may be used to augment or replace humans who would otherwise be exposing themselves to hazardous or potentially fatal situations such as search and rescue in collapsed structures or identification, monitoring, and removal of harmful environmental contamination. The goal of this coordination is to minimize the power expended obtaining more energy by the deployed robots. This is accomplished by continuously relocating energy sources to meet the needs of the team. We will show that the optimal solutions to this problem are NP-hard. To solve this problem when scaled to large numbers of mobile docking stations and deployable robots, a heuristic-based coordination algorithm is required. Performance of the coordination algorithm is measured based upon the mean % time the deployable robots are able to actively carry-out their mission, are recharging, or are unable to carry out their mission because they are attempting to reach a source of power. Additional metrics include the number of robots which are unable to successfully obtain power and the number of times the robots are able to successfully recharge themselves. The computational complexity of this coordination algorithm is such that in many (but not all cases), the docking stations are able to choose their position using convex optimization techniques ensuring scalability to large numbers of systems. In the cases that convexity is not possible either directly or through relaxations of the problem, docking stations are able to choose their location through gradient descent or other methodologies at higher computational costs. For comparison purposes simulations were conducted that illustrate that when moving the docking stations in accordance to this strategy, there are significant improvements in nearly all metrics compared to having docking stations at fixed locations within the environment.