Fusing LiDAR and radar data to perform SLAM in harsh environments

Abstract

LiDAR sensors are very popular for mapping and localisation with mobile robots, yet they cannot handle harsh environments, containing smoke, fog, dust, etc. On the other hand, radar sensors can overcome these situations, but they are not able to represent an environment in the same quality as a LiDAR due to their limited range and angular resolution. In the following article, we present further results regarding SLAM involving the mechanical pivoting radar (MPR), which is a 2D high bandwidth radar scanner that was introduced in Fritsche et al. (Radar and LiDAR sensor fusion in low visibility environments, 2016, [8]). We present two strategies for fusing MPR and LiDAR data to achieve SLAM in an environment with low visibility. The first approach is based on features and requires the presence of landmarks, which can be extracted with LiDAR and MPR. The second SLAM approach is based on scan registration and requires a scan fusion between the two sensors. In the end, we show our experiments, involving real fog, in order to demonstrate, how our approaches make SLAM possible in harsh environments.