RAPID APPROXIMATION OF OFF-ROAD VEHICLE MOBILITY WITH LIMITED TERRAIN KNOWLEDGE USING DIMENSIONLESS PARAMETERS

In the field of autonomous off-road land locomotion, the ability to predict vehicle mobility is crucial to successful navigation. Unfortunately, it is often difficult to quantify the mobility of a given ground vehicle without effectively mapping the terrain in great detail. Detailed mapping is not always feasible, yet it is still desirable and often necessary to have a high-level estimate of vehicle performance for a given terrain. For example, estimates on how much energy/fuel a vehicle will need and how long it will take to complete a mission are basic pre-requisites to starting almost any task with an autonomous or human-operated vehicle. This paper proposes a method for finding these high-level estimates with limited terrain knowledge and basic vehicle modeling. The method utilizes dimensionless representations of obstacle fields that allow one to simulate behavior within a specific environment, and then abstracts those results to estimate mobility in a large number of other environments. The inputs to the algorithm are dimensionless parameters representative of the obstacle field and vehicle size and the output is a dimensionless parameter representing the estimated traversal distance. The distance parameter can then be used to determine high-level estimates of vehicle energy/fuel usage and mission completion time. The results also demonstrate how uncertainty in terrain knowledge is propagated through to the distance estimates. Ultimately, the process allows the results to be generalized to maps of any scale by utilizing dimensionless parameters and provides a framework for using dimensionless parameters of obstacle field geometries to estimate mobility metrics.