Hong Kong UrbanNav: An Open-Source Multisensory Dataset for Benchmarking Urban Navigation Algorithms

Abstract Accurate positioning in urban canyons remains a challenging problem. To facilitate the research and development of reliable precise methods using multiple sensors canyons, we built multisensory dataset, UrbanNav, collected diverse, scenarios Hong Kong. The dataset provides multi-sensor data, including data from multi-frequency global navigation satellite system (GNSS) receivers, an inertial measurement unit (IMU), light detection ranging (lidar) units, cameras. Meanwhile, ground truth (with centimeter-level accuracy) is postprocessed by commercial software NovAtel integrated GNSS real-time kinematic fiber optics gyroscope system. In this paper, sensor systems, spatial temporal calibration, formats, scenario descriptions are presented detail. benchmark performance several existing provided as baseline. Based on evaluations, conclude that can provide satisfactory results middle-class canyon if appropriate receiver algorithms applied. Both visual lidar odometry deep whereas tunnels still major challenge. Multisensory integration with aid IMU promising solution for achieving seamless cities. its entirety be found GitHub at https://github.com/IPNL-POLYU/UrbanNavDataset.