Towards Robot Systems Architecture

Just as special purpose computers and mainframes grew into the general purpose personal computers we use everyday, special purpose industrial robots are evolving into more general purpose personal robots. As robots become more capable and universal, their applications are less well-defined or even unknown at design time. We will have to design robots for classes of tasks rather than specific applications. Having guidelines for how to best organize, interface, and implement robot systems and reason about trade-offs, as we do in computer architecture, will become crucial for success. In this paper we introduce and adapt some useful notions and principles from computer architecture to robot systems architecture. We argue that notions such as locality of reference, balanced architectures, and boundedness (in terms of IO, memory, and CPU) can be leveraged in robot systems design, and in particular, in the design of distributed robot systems.