Information theoretic unified modelling of autonomous systems

Unified models for an autonomous mobile robot and its environment using information theoretic formalism are proposed. These are obtained by regarding both the robot and the environment as an information processing system. For such a system, a canonical decomposition is elaborated according to its effectability and perceptibility properties. Effectability and perceptibility are defined based on the transmission rate between system variables. It is further shown that the processing power of the whole system is additively shared by input-filtering, internal interconnections, internal system noise, output disturbance and the net input-output transmission rates. Based on these results, suggestive rules and guide-lines for mobile robot design are obtained. Although these results are on a high level of abstraction, they can be very useful in the efficient assignment of the information processing resources on-board the robot.