Methods for the Reduction of Odometry Errors in Over-Constrained Mobile Robots

This paper presents an analysis of odometry errors in over-constrained mobile robots, that is, vehicles that have more independent motors than degrees of freedom. Examples of overconstrained vehicles are the various 6-wheeled Mars Rovers like Rocky-7, Rocky-8, or Fido. Based on our analysis we developed two novel measures aimed at reducing odometry errors. We also developed a novel method that serves as a framework for the implementation of the two new measures, as well as for other, conventional error reducing measures. One of the two new measures, called “Fewest Pulses Measure,” makes use of the observation that most terrain irregularities, as well as wheel slip, result in an erroneous overcount of encoder pulses. The second new measure, called “Cross-coupled Control Measure,” optimizes the motor control algorithm of the robot to reduce synchronization errors that would otherwise result in wheel slip with conventional controllers. The novel method that serves as a framework for other measures is based on so-called “Expert Rules.” In this paper we formulate three expert rules aimed at reducing dead-reckoning errors. Two of these expert rules are related to the foregoing discussion on error reducing measures. The third expert rule adds a gyroscope to the system and we re-examine the effectiveness of the odometry error-reducing measures in the context of this addition. In the work described in this paper we modified a Pioneer AT skid-steer platform by providing it with four independent drive motors and encoders. We implemented our errorreducing measures and the expert rule method on this over-constrained platform and present experimental results. ) Affiliation of Authors: 1) Lauro Ojeda University of Michigan Department of Mechanical Engineering Advanced Technologies Lab 1101 Beal Avenue Ann Arbor, MI 48109 Ph: (734) 647-1803 email: [email protected] 2) Johann Borenstein University of Michigan Department of Mechanical Engineering Advanced Technologies Lab 1101 Beal Avenue Ann Arbor, MI 48109 Ph: 734-763-1560 email: [email protected] ) A poster version of this paper was presented in a poster session at the 2003 SPIE AeroSense Symposium, Orlando, FL, April 21-25, 2003, and a shorter version of this paper is included in the proceedings of that conference.