Current abstractions, design methodologies, and design tools are useful but inadequate for modern mobile robot design. By viewing robotics systems as an interactive and reactive agent and environment combination, and focusing on the interactions between the two, particularly those interactions that result in task accomplishment, one arrives at the interaction space abstraction.

The role of abstractions, formalisms and models are discussed, with emphasis on several specific abstractions used for robotics as well as the strengths and shortcomings of each. The role of design methodologies is also discussed, again with emphasis on several currently used in robotics. Finally, design tools and the use thereof are briefly discussed.

The concept of interaction spaces as an abstraction and a formalism is developed specifically for use in robot design. Types of elements within this formalism are developed, defined, and described. A formal nomenclature is introduced for these elements based on Simulink blocks. This nomenclature is used for descriptive models and the Simulink blocks are used for predictive models.

The interaction space abstraction is combined with the concept of exploration-based design to create a design methodology specifically adapted for use in descriptive modeling of autonomous robots. This process is initially developed around a simple wall-following robot, then is expanded around a multi-agent foraging system and an urban search and rescue robot model, each of which demonstrates different aspects and capabilities of interaction space modeling as a design methodology.

A design tool based on iterative simulation is developed. The three specific examples above are used to perform quantitative simulation and the results are discussed with emphasis on determination and quantification of factors necessary for task accomplishment. These simulations are used to illustrate how to explore the design space and evaluate trade offs between design parameters in a system.