Doyle, Matthew. “The Propulsion of Reconfigurable Modular Robots in Fluidic Environments.” PhD diss., University of Sheffield, 2019.
URL1 URL2
Reconfigurable modular robots promise to transform the way robotic systems are designed and operated. Fluidic or microgravity environments, which can be difficult or dangerous for humans to work in, are ideal domains for the use of modular systems. This thesis proposes that combining effective propulsion, large reconfiguration space and high scalability will increase the utility of modular robots. A novel concept for the propulsion of reconfigurable modular robots is developed. Termed Modular Fluidic Propulsion (MFP), this concept describes a system that propels by routing fluid though itself. This allows MFP robots to self-propel quickly and effectively in any configuration, while featuring a cubic lattice structure. A decentralized occlusion-based motion controller for the system is developed. The simplicity of the controller, which requires neither run-time memory nor computation via logic units, combined with the simple binary sensors and actuators of the robot, gives the system a high level of scalabilty. It is proven formally that 2-D MFP robots are able to complete a directed locomotion task under certain assumptions. Simulations in 3-D show that robots composed of 125 modules in a variety of configurations can complete the task. A hardware prototype that floats on the surface of water is developed. Experiments show that robots composed of four modules can complete the task in any configuration. This thesis also investigates the evo-bots, a self-reconfigurable modular system that floats in 2-D on an air table. The evo-bot system uses a stop-start propulsion mechanism to choose between moving randomly or not moving at all. This is demonstrated experimentally for the first time. In addition, the ability of the modules to detect, harvest and share energy, as well as self-assemble into simple structures, is demonstrated.