US Patent 10424079 Unsupervised approach to environment mapping at night using monocular vision

A trained feature network receives an image captured under low illumination conditions and pose data corresponding to the image. The trained feature network identifies a feature within the image and analyzes the image to extract feature information corresponding to the feature from the image. Based on the image and the pose data, geo-location information corresponding to the feature is determined. The geo-location information is appended to the extracted feature information. The feature information is stored as part of a feature map layer of a digital map. At least a portion of the digital map is provided to a routing and navigation system, for example, for performing vehicle localization under the particular condition.