Modelling of Terrain Surfaces Using Aerial Radar Mapping
Aerial scene mapping is often done via visual methods, where many 2D images are combined together to create 3D maps. There are several disadvantages to this approach, however, including weather interference, inconsistent or nonexistent lighting, or fast-moving objects, which often appear as either noise or a blur on the created map. Radar technology is mostly used in automotive applications, like lane keeping and adaptive cruise control. It can also be used, however, for aerial mapping of scenes, by mounting the radar to a drone. The radar can then be used to generate a point cloud, that can either replace or complement point clouds and maps from other sensors. We present an approach for generating a point cloud map using aerial radar. We then present an algorithm for aggregation of points based on pose, and a method to create accurate meshes over mapped terrain, to facilitate path planning by ground-based robots.