Architecture of the control system of a mobile robot based on the use of a real-time multitasking executive with an expert system

Abstract

A hierarchical control system of an autonomous mobile robot which has to travel in a quasi-static environment is studied and implemented. The architecture of the control system involves three levels: planner, navigator, and pilot. From a cartography of the environment, the planner elaborates all the possible paths to reach a goal from a starting point. The algorithm used is based on the traversability concept. The planner level does not operate in real time. With the information given by the planner, the navigator defines a constrained optimal trajectory. The navigator is able to avoid an unexpected obstacle on the defined trajectory and the navigator works under real-time constraints and a multitasking mode. The pilot manages the sensors (odometry, telemetry) and the feedback control of the robot actuators. These tasks are decoupled from those executed by the main processor. The three hierarchical levels are designed with a real-time multitasking operating system associated with an expert system. >