Precise Navigation for an Agricultural Robot

Abstract: This report is exploring the advantages and disadvantages with some common sensors needed when positioning a mobile robot with Ackerman steering. Both absolute and relative methods are explored. Odometry based on measuring traveled distance and steering angle is tested and adjusted to obtain the best performance. A Kalman filter is constructed to improve steering angle measurements. The tests are based on an agricultural robot tended to follow a row of sugar beets while weeding it. Best position estimate was obtained with the Kalman filter filtering the steering angle information from the steering motor and the information from the difference in velocity on the rear wheels. The average error in this measurement was 6.84%/m.