A ROS-based dataset and tools for autonomous vehicle development with seasonal multi-sensor data from Ford vehicles.
Ford AV Dataset is an open-source autonomous vehicle dataset containing multi-sensor data collected across different seasons. It provides LiDAR, GPS, and IMU data along with tools for visualization and processing within the ROS ecosystem. The dataset helps researchers and developers test and improve autonomous driving algorithms using real-world seasonal variations.
Autonomous vehicle researchers, robotics engineers, and computer vision developers working on perception, localization, and mapping systems for self-driving cars.
Developers choose this dataset because it offers real-world multi-season data from actual Ford vehicles with complete ROS integration. Unlike synthetic datasets, it provides authentic seasonal variations and comes with ready-to-use tools for visualization and data conversion.
Autonomous Vehicle Seasonal Dataset
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Captures LiDAR, GPS, and IMU data across different seasons and weather conditions, providing authentic variations for testing perception algorithms in changing environments.
Includes ROS packages for visualization, map loading, and data conversion, with launch files and Rviz plugins that streamline setup within existing ROS workflows.
Publishes ground plane reflectivity and 3D pointcloud maps based on vehicle pose, with tunable parameters like publish_rate and neighbor_dist for optimized performance.
Provides Python scripts (e.g., bag_to_csv.py) to convert ROS bag files to CSV format, enabling analysis outside the ROS ecosystem without proprietary tools.
Relies on Python 2.x and specific older ROS versions (Kinetic, Melodic, Noetic), which may conflict with modern development environments and lack long-term support.
Pointcloud processing is computation-intensive, potentially causing lag on systems with 16GB RAM, requiring manual tuning of parameters that can be complex for beginners.
Heavily tied to ROS for core functionalities like visualization and map loading, making it less accessible for teams using other frameworks or preferring raw data handling.