Nonlinear regression analysis of rut profile data for optimal data storage and efficient terrain condition analysis
IEEE Geoscience and Remote Sensing Letters
School of Engineering
During various operations of military and civil importance, the vehicles need to move in different off-road conditions. Several soil parameters that differ in spatial and temporal domains impact the possibility of vehicular movement. Conventional evaluation of the prevalent state of soil being cumbersome and time-consuming, the information extracted from the rut profile of the vehicle is used. Several techniques ranging from manual to advanced laser scanners are there to monitor this profile. The laser scan contains a vast point cloud dataset. It brings in the data storage and retrieval issue for an efficient computational framework for real-time mobility guidance. Little research exists for addressing the storage issue of this sort of rut profile data. Here, rut profile data captured manually in different field running conditions are used for studying the optimization aspects. Various mathematical formulations are analyzed, and nonlinear regression analysis is attempted on the commonly observed rut shapes. Using the proposed modified bell-shaped function on these common shapes, the R2 value moved above 0.98. It could fetch an additional compression of over 80% on straight patches and 71% on turnings. Although complex scenarios could be dealt with separately, an optimal representation of the most common rut shapes can aid the development of an efficient decision support system. © 2004-2012 IEEE.