Behavior-Based Control for Autonomous Mobile Robots

Abstract

The harsh nature of planetary surfaces introduces many new constraints into the types of control systems suitable for use in such environments. These include low power requirements, operation within wide temperature extremes, relatively low computing capabilities and onboard memory, and multiple terrain types ranging from featureless flat plains to sheer cliffs. This paper presents the results of ongoing work at JPL and USC in autonomous rover control, which has concentrated on the development of a behavior-based control system called BISMARC. BISMARC includes full sensor and mobility models, geometric constraints, and environmental feedback based on the SRR/FIDO rovers at JPL. It is based on a biologically motivated navigation system derived from a study of human path planning in complicated exterior environments. The system has had over 800 simulation runs of a multiple cooperating rover, multiple sample return mission scenario with an overall success rate of 98.9%, and is currently being tested on the SRR rover in the Planetary Robotics Lab at JPL.