Abstract: Military ground vehicles operate in off-road environments traversing different terrains under various environmental conditions. There has been an increasing interest towards autonomous off-road vehicle navigation, leading to the needs of terrain traversability assessment through sensing. These methods utilized data-driven approaches on classical robotic perception sensing modalities (RGB cameras, Lidar, and depth cameras) positioned in front of ground vehicles in order to observe approaching terrain. Classical robotic sensing modalities, though effective for describing environment geometry and object detection and tracking, aren’t able to directly observe features related to compaction and moisture content which have significant effects on the moduli properties governing terrain mechanics. These methods then become very specialized to specific regions and environmental conditions which are inevitably subject to change. Radio wave-based sensing modes have been shown in studies to have success in observing different terrain surface and subsurface conditions such as compaction and moisture presence. We study the usability of emerging, portable and front mountable radar imaging sensors to provide real-time radio spectra information of the in-coming terrain area. In this study, we use a radar transceiver array operating in the 6.2-6.9 GHz spectral range to develop a radar image/soil moisture dataset, where beamforming is used to recover radar images of lab soil samples of various moisture content levels. The radar images are constructed at various distances from the soil surface and various spatial resolutions to support a local path planning scenario. Support vector machine (SVM) classifier and support vector regression (SVR) models are trained on the dataset and tested on lab data and in-field data. Classifier and regression model results indicate that normalized local radar image statistics are able to distinguish moisture levels at various distances and spatial resolutions.