Navigation of an Autonomous Robot Vehicle

by Arturo L. Rankin, Univ of Florida, Gainesville, United States,
David G. Armstrong, II, Univ of Florida, Gainesville, United States,
Carl D. Crane, III, Univ of Florida, Gainesville, United States,



Document Type: Proceeding Paper

Part of: Robotics for Challenging Environments

Abstract:

The U.S. Air Force has developed an autonomous excavator capable of performing a variety of tasks at various Department of Defense facilities. Common to all the tasks is the need for autonomous navigation. Navigation subtasks include path planning, vehicle positioning, path following, and obstacle avoidance. These subtasks are the focus of this paper. An efficient and reliable planar off-line path planner has been developed that is based on the A* search method. For the most complicated field of obstacles, a near-optimal piecewise-linear path is found within a few seconds. An inertial navigation system (INS) determines the vehicle position and orientation in real time (12 Hz). To reduce the drift error, a global positioning system (GPS) periodically supplies the INS with position updates accurate to within 0.5 meters. Path following is accomplished through a novel technique called the `follow-the-carrot' method. Proportional control is used to sequentially steer toward a set of subgoals. Sixteen Polaroid ultrasonic sensors are used to direct the NTV around unexpected obstacles and back on the preplanned path. This software was developed for use on a VME computer operating under the VxWorks operating system. It is currently implemented on a Kawasaki Mule 500 all-terrain vehicle, modified to serve as a navigation test vehicle (NTV). The NTV is currently capable of accurate path tracking at speeds up to 7.0 mph.



Subject Headings: Global positioning systems | Robotics | Probe instruments | Ultrasonic methods | Inertia | Computer software | Errors (statistics) | Excavation | Indiana | United States

Services: Buy this book/Buy this article

 

Return to search