ABSTRACT
Real-time intelligent robot controllers are required to achieve the level of autonomy necessary in unstructured or “nonengineered” operating domains. For mobile robots, the operating domain, whether an indoor industrial workplace, outdoor paved road, or planetary surface, dictates acceptable design and control approaches. One common requirement, independent of the operating domain, is a capability for intelligent navigation. Traditional approaches based on sequential task decomposition have met with difficulty in achieving real-time response. Controllers are now being developed that are modeled after natural processes and that advocate a behavioral decomposition of tasks with quasi-parallel execution. These behavior-based systems facilitate real-time intelligence by decomposing motion control capabilities into a set of special-purpose routines that operate concurrently.
