No abstract

No abstract

A considerable amount of work the problem of robot navigation in has been reported on known static terrains. Algorithms have been propoged and implemented to search for an optimum path to the goal, taking into account the finite size and shape of the robot. Not as much work has been reported on robot navigation in unknown, unstructured, or dynamic environments.A robot navigating in an unknown environment must explore with its sensors, construct an abstract representation of its global environment to plan a path to the goal, and update or revise its plan based on accumulated data obtained and processed in real-time.The core of the navigation program for the CESAR robots is a production system developed on the expert-systemshell CLIPS which runs on an NCUBE hypercube on board the robot. The production system can call on C-compiled navigation procedures. The production rules can read the sensor data and address the robot's effectors. This architecture was found efficient and flexible for the development and testing of the navigation algorithms; however, in order to process intelligently unexpected emergencies, it was found necessary to be able to control the production system through externally generated asynchronous data. This led to the design of a new asynchronous production system, APS, which is now being developed on the robot. This paper will review some of the navigation algorithms developed and tested at CESAR and will discuss the need for the new APS and how it is being integrated into the robot architecture.

No abstract

No abstract