A long-term autonomous driving dataset from Europe with multi-sensor data (GPS-RTK, LiDAR, cameras, IMU) for localization and mapping research.
The UTBM Robocar Dataset is a long-term, multi-sensor dataset for autonomous driving research, collected from a vehicle equipped with GPS-RTK, LiDAR, cameras, and an IMU. It addresses the need for real-world, varied environmental data to develop and test algorithms for vehicle localization, mapping, and long-term autonomy.
Researchers and engineers in robotics and autonomous systems focusing on localization, SLAM, and perception, particularly those using ROS and working with sensor fusion.
It offers a unique long-term perspective with seasonal changes, provides high-precision GPS-RTK ground truth, and includes ready-to-use ROS baselines, enabling direct benchmarking and reducing setup time for experiments.
EU Long-term Dataset with Multiple Sensors for Autonomous Driving
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Includes synchronized GPS-RTK, LiDAR, cameras, and IMU data, enabling comprehensive sensor fusion for localization and mapping research, as highlighted in the dataset features.
Captures data over several months with variations like sunny, cloudy, rainy, and night conditions, providing a realistic testbed for algorithm robustness in changing environments.
Offers ROS launch files for state-of-the-art methods like Hector pose estimation, LOAM, and LeGO-LOAM, reducing setup time and facilitating direct algorithm comparison.
Uses GPS-RTK for accurate vehicle positioning ground truth, essential for rigorous evaluation of localization algorithms, though issues are noted in the README.
The README acknowledges issues with radar data (issue #13), limiting its utility for full perception system development compared to other datasets.
Licensed under CC BY-NC-SA 4.0, which prohibits commercial use and may hinder industrial research or product development efforts.
Data is only in ROS bag files, requiring familiarity with ROS tools and workflows, making it less accessible for researchers using other frameworks like PyTorch or TensorFlow directly.
While baselines are provided, the README assumes prior ROS knowledge, and issues like GPS-RTK problems (issues #1, #7) are noted but not fully resolved, potentially complicating evaluation.