Fuzzy Control for Autonomous Ground Vehicles

摘要

The Center for Intelligent Machines and Robotics (CIMAR) at the University of Florida has worked in the area of autonomous ground vehicles (AGVs) for several years under the sponsorship of the Air Force Research Laboratory at Tyndall Air Force Base, Florida. The objective of the work is to develop technological capabilities that can be applied to a variety of Air Force needs and application areas. Recently, one of these capabilities required the design of a modular architecture for autonomous vehicle navigation. This new architecture, which is currently under development, is called Modular Architecture eXperimental (MAX). One of the unique features of this architecture is a generic message for controlling the motion of any autonomous vehicle. This paper describes a control technique, which uses this generic message, for navigating various autonomous ground vehicles. The resulting technique uses two fuzzy model reference learning controllers (FMRLCs). One FMRLC controls the vehicle linear velocity, and the other controls the vehicle angular velocity. Both controllers are designed from parameters that are defined in the MAX interface document. They have been implemented and tested successfully on three different vehicles, a Kawasaki Mule with Ackerman steering, a K2A robot with three wheel synchronous drive, and a tracked vehicle called All-purpose Remote Transport System (ARTS).