Using GANs to Enable Semantic Segmentation of Ranging Sensor Data

Ranging sensors, such as radar and lidar, onboard the vehicle are considered to be very robust under changing environmental conditions. Largely owing to this reputation, they have found broad applicability in driver assistance, and consequently in autonomous driving systems. On the other hand, they lack precision. This makes classification tasks of the measurement data rather difficult. In this paper, we propose a method for semantic segmentation of the ranging sensors data using generative adversarial networks. Utilizing the fully unsupervised learning algorithm, we convert the sensor data to artificial, camera-like, environmental images that are further used as input for semantic image segmentation algorithms.