High resolution digital elevation modeling with on-vehicle sensors

High resolution elevation mapping is useful for precise field operations. To achieve a good operation performance, the requirement of the field shape accuracy can be centimeter-level in some tasks. This paper presents a method of conducting high resolution elevation mapping with vehicle-borne sensors. To achieve this purpose, a laser range sensor obtains geometric information of the ground surface while an Inertial Navigation System (INS) and vision sensor provide high resolution measurements of the moving vehicle's current location and attitude. The high frequency information from the INS captures the vehicle dynamics, and the low frequency vision sensor data enables vehicle self-localization by eliminating drift error in the INS measurements. A test run of this system was conducted experimentally. And the inclusion of vision-based measurements was shown to improve the accuracy of the ground elevation comparing with an INS-only method.