Experiments in design synthesis when behavior is determined by shape
Personal and Ubiquitous Computing
| Hi-index | 0.00 |
This thesis is about heterogeneous modular reconfigurable robotics: a class of physical systems made up of a number of tangible, computational components that operate concurrently. Although this is a new type of system and not in widespread use for any purpose, the current research interest and potential benefits of future effective systems suggest that it will be soon. Two contributions related to the design of these systems are described.In an attempt to inform robot designers, this thesis analyzes the properties and parameters of roBlocks, a modular reconfigurable robotic construction toy. Design issues related to mechanics, data flow, and power are addressed, and the system, its use, and its affordances are described in detail. Design of the roBlocks system provides a valuable case study; design decisions and their alternatives are discussed and the system properties that emerge from them are analyzed.The second contribution involves the design of robotic constructions using the modules of these systems as media. Instances of this class of systems, with distributed, concurrent intelligence, exhibit emergent behavior that can be difficult or impossible to predict. This makes the designer's task of creating a functional robot out of a kit of parts very challenging. This problem is addressed by showing how evolutionary algorithms, in combination with accurate physical simulation, can create viable robots out of a modular kit. An evolutionary, automated design program is presented that evolves robots in a manner based on genetic programming techniques and tests them in 3D physical simulation against user-specified objectives.The complex, emergent behavior that is produced by combinations of modular robots (and there are lots of combinations) is hard to design without the aid of tools. This thesis offers tools in the form of (a) an analysis and case study for describing and designing this class of systems; (b) a physical kit that can be used to explore modular robots composed of smaller, heterogeneous, concurrent robots; and (c) a series of algorithms that evolve, or design, candidate configurations in order to satisfy design requirements.