UrbanLoco

UrbanLoco is a multi-sensor dataset for benchmarking mapping and localization algorithms in highly urbanized environments. It contains 13 trajectories totaling over 40 km across San Francisco and Hong Kong — urban canyons, bridges, tunnels, and sharp turns — with synchronized LiDAR, cameras (360° in San Francisco; fish-eye sky-camera in Hong Kong), IMU, and GNSS measurements, with ground truth from a NovAtel SPAN CPT 7 + RTK system.

Why it matters. Existing public AV datasets either offered limited sensor coverage or avoided the most challenging urban scenarios. UrbanLoco captures real-world conditions where GNSS signals are severely degraded by multipath and where LiDAR/camera-based methods struggle with dynamic objects — providing a realistic stress test for autonomous-vehicle navigation algorithms.

Outputs: open dataset under CC BY-NC-SA 4.0; GitHub repository; accompanying ICRA 2020 paper by Wen, Zhou, Zhang, Fahandezh-Saadi, Bai, Zhan, Tomizuka, and Hsu. Released in collaboration with the Mechanical Systems Control Lab at UC Berkeley.