SPOTT: A Mobile Robot Control Architecture for Unknown or Partially Known Environments

Planning with Incomplete Information for Robot Problems

Volume

Planning with Incomplete Information for Robot Problems

Contents

A robot control architecture called SPOTT 1 is proposed and implemented as a real-time and parallel system of concurrently executing and cooperating modules. SPOTT provides a bridge for linking behavioral (i.e., reactive) and symbolic control. The control system is a realtime AI system which is responsible for dynamically adapting to changing environmental circumstances in order to successfully execute and complete a set of navigational tasks for an autonomous mobile robot. SPOTT consists of a behavioral controller, a local dynamic path planner, and a global path planner, as well as a map and a graphical user interface. The behavioral control formalism is called TR+ and is based on an adaptation and extension of the Teleo-Reactive (TR) formalism introduced by Ntis Nilsson. TR+ rules make decisions which affect actuator control and map database maintenance. A dynamic local path planner continually polls the map database in order to navigate around newly encountered obstacles. The local dynamic path planner is based on a potential field method using harmonic functions, which are guaranteed to have no spurious local minima. The global planning module advises the local planning module of the effects of a global goal. We have developed a real-time parallel implementation of SPOTT using a message passing software package called PVM. Navigational experiments have consisted of moving the robot in an office and laboratory environment to known and unknown spatial locations with no or a partial a priori map.