LIDAR/Inertial-based Navigation Methods for Autonomous Drone Operations in GNSS-denied Urban Environments
Over the last years, the number of applications for drones has grown significantly, driving interest in techniques to ensure safe operations, particularly in challenging environments like urban areas and with an increasing level of autonomy. Currently, many navigation methods require GNSS, however vulnerabilities associated with GNSS performance at very low urban flight levels due to effects like signal degradation and multipath exist. Thus, alternative navigation solutions, based on sensors like camera, LIDAR and RADAR, are focus of many research projects. This presentation introduces a LiDAR-IMU-based navigation system designed to address these issues by enabling reliable positioning without relying on GNSS. Three LiDAR-based approaches are covered: (1) Digital Scene Matching using octomaps and a global database, (2) local pose estimation from consecutive LiDAR scans without a global map, and (3) image-based pose estimation using 2D heatmaps and optical flow. A short theoretical introduction into the three methods is given as well as results from flight tests in an urban area close to Berlin are demonstrated.