Mobile robots that are able to move about and effectively negotiate their environment are attractive for a wide variety of applications. Such applications include surveillance, inspection, and mobile sensing where robots often present cost-effective alternatives to human labor. Other applications include those that are potentially hazardous to humans; examples of these include search and rescue, monitoring and maintenance of toxic environments, and planetary exploration.
A vast majority of research into mobile robots has been limited to structured environments such as research labs, indoor office environments, industrial settings, maintained roads, etc. As the number of mobile robot applications grows, so does the need for such systems to be able to operate in unstructured (general) environments. Such environments often exhibit a wide variety of terrain including uneven surfaces and significant terrain irregularities. In some cases hazardous areas can be tactically avoided with careful path planning, but in general this is not always possible and obstacles must be negotiated directly. For these applications it is imperative that the robot exhibits a sufficient level of mobility to be able to perform required tasks.
In addition to mobility requirements, many mobile robot applications are further constrained by limitations on physical size and/or cost. It is often the case that small (inexpensive) robots are preferable if not required. In general, however, it is the case that miniature mobile robots sacrifice mobility in exchange for their small size. Additionally, the increased design complexity of miniature systems often increases both design and manufacturing cost. In this thesis we present a relatively new and unexplored form of robotic locomotion called tumbling which addresses many of the aforementioned existing limitations of miniature mobile robots; the thesis is comprised of three main parts.
In the first, tumbling and tumbling robots are defined and discussed in detail as well as other useful notation. Additionally, we present a classification of tumbling robots along with a catalog of existing designs to establish the state of the art. This treatment marks the first of its kind and establishes the first formal definitions with respect to tumbling locomotion for mobile robots. In the second, we examine terrainability of the class of serial multiply actuated tumbling robots by looking at the underlying principles of tumbling interactions with several idealized obstacles. Specifically, we derive configuration equations that relate terrainability to the parameters of an idealized tumbling robot. The results are supported through experimentation using the Adelopod, a physical tumbling robot developed as part of this thesis, over a variety of repeatable terrains. Finally, we conclude by examining the maneuverability for the class of serial multiply actuated tumbling robots and begin to address motion planning for such devices. We present results of several planning algorithms as well as a method for deriving useful distance metrics for significant planning speedup and increased path quality. Results of applying such metrics are presented.