INS-aided odometry and laser scanning data integration for real time positioning and map-building of skid-steered vehicles

An enhanced odometry method coupled with an inertial navigation sensor is introduced for skid-steered vehicle positioning in outdoor environments. In the proposed scheme, robot positioning and attitude is estimated based on an experimentally derived kinematic model. Besides, a low-cost INS unit is used to improve the quality of the rotational velocity of the robot as it is obtained by differential odometry. It is concluded that a reliable positioning can be derived with the use of a comprehensive DIA procedure for outlier removal and the fusion of odometry with inertial data through a discrete Kalman filter. The map-building component of the system is based on the integration of a pair of high sensitivity 2D laser scanners suitably mounted on the vehicle. Finally, an optimum estimate of the robot position and attitude is derived through data fusion of the dead-reckoning and map-matching information. The technique has been implemented on-board an experimental skid-steered vehicle in cases of extreme motion, including runs with steep turns and variable velocity.