Odometry error correction by sensor fusion for autonomous mobile robot navigation

Abstract

A low cost navigation system is developed which fuses inertial sensor information provided by gyroscopes and odometry information. Both Kalman filters and a rule set based fusion strategy are used to correct the odometry errors in orientation for mobile robots. The fusion system even improves orientation estimation using a gyroscope with an extremely high drift rate. This navigation system is implemented on the autonomous mobile robot B21. The performance of this fusion method is tested on different surfaces and orientation errors do not increase as the robot travels along. Experimental results demonstrate that even in harsh environments with obstacles the performance of the designed navigation system reduces odometry tracking errors. The effects of nonsystematic odometry errors caused by unpredictable large bumps or objects encountered on the floor are reduced.