fsd-test-bench

This dataset by Mainfranken Racing e.V. aims to lower the entry barrier for new Formula Student Driverless teams by serving as a baseline benchmark for state estimation algorithms. The dataset comprises recordings from multiple Formula Student Driverless disciplines across various track layouts. The recordings primarily contain raw sensor measurements, enabling the development and evaluation of custom state estimation pipelines. Ground-truth data for vehicle position and velocity are provided through RTK-GPS measurements and a ground-speed sensor. Furthermore, ground-truth maps are available for most tracks, created through precise RTK-GPS surveying. These maps provide a reliable reference for validation, benchmarking, and performance evaluation. ℹ️ Info: The Dataset was recorded via ROS-Bags on Ros Noetic on Ubuntu 20.04. A software stack for working with the dataset, including the required ROS Message Types, is provided at: https://github.com/mfr-driverless/minimal-fsd-stack 🏷️ Contents 🏷️ Name Map Lidar_raw Lidar_detections Cam_raw Cam_compressed Cam_depth Cam_detections RTK_pos Groundspeed Acceleration/acceleration1-_track1.bag Acceleration/maps/track1.csv ✓ ✘ ✓ ✓ ✓ ✘ ✓ ✓ Acceleration/acceleration2-_track1.bag Acceleration/maps/track1.csv ✓ ✘ ✘ ✓ ✘ ✘ ✓ ✓ Acceleration/acceleration3-_track2.bag Acceleration/maps/track2.csv ✓ ✘ ✓ ✓ ✓ ✘ ✓ ✓ Acceleration/acceleration4-_track2.bag Acceleration/maps/track2.csv ✓ ✘ ✓ ✓ ✓ ✘ ✓ ✓ Acceleration/acceleration5-_track2.bag Acceleration/maps/track2.csv ✘ ✓ ✘ ✓ ✘ ✓ ✓ ✓ Acceleration/acceleration6-_fscz.bag - ✘ ✓ ✘ ✓ ✘ ✓ ✓ ✓ Autocross/autocross1-_track1.bag Autocross/maps/track1.csv ✓ ✘ ✓ ✓ ✓ ✘ ✓ ✓ Autocross/autocross2-_track2.bag Autocross/maps/track2.csv ✓ ✘ ✓ ✓ ✓ ✘ ✓ ✓ SkidPad/skidpad1.bag SkidPad/maps/track1.csv ✓ ✘ ✓ ✓ ✓ ✘ ✓ ✓ SkidPad/skidpad2.bag SkidPad/maps/track1.csv ✘ ✓ ✘ ✓ ✘ ✓ ✓ ✓ SkidPad/skidpad3.bag SkidPad/maps/track1.csv ✓ ✘ ✓ ✓ ✓ ✘ ✓ ✓ SkidPad/skidpad4.bag SkidPad/maps/track1.csv ✓ ✘ ✘ ✘ ✘ ✘ ✘ ✓ Trackdrive/trackdrive1-_track1.bag Trackdrive/maps/track1.csv ✓ ✓ ✘ ✓ ✘ ✓ ✓ ✓ Trackdrive/trackdrive2-_track2.bag Trackdrive/maps/track2.csv ✓ ✓ ✘ ✓ ✘ ✓ ✘ ✓ Trackdrive/trackdrive3-_track2.bag Trackdrive/maps/track2.csv ✘ ✓ ✘ ✓ ✘ ✓ ✓ ✓ Trackdrive/trackdrive4-_fscz.bag - ✘ ✓ ✘ ✘ ✘ ✓ ✘ ✘ Data Explanation Lidar_raw: Raw LiDAR pointcloud (OUSTER OS1-64) – /ouster/points Lidar_detections: Detected cones by the LiDAR Perception – /lidar/cone_position_cloud Cam_raw: Raw uncompressed camera image (Stereolabs ZED2) – /zed2/zed_node/left/image_rect_color and /zed2/zed_node/right/image_rect_color Cam_compressed: Compressed camera image – /zed2/zed_node/left/image_rect_color_compressed Cam_depth: Depth image from stereocamera (Stereolabs ZED2) – /zed2/zed_node/depth/depth_registered Cam_detections: Detected cones by the Stereocamera Perception – /stereo_cone_perception/cones RTK_pos: Ground truth position from RTK-GPS (SBG Ekinox Micro) on the car – /sbg/ekf_nav and /sbg/ekf_euler Groundspeed: Ground truth velocity from a groundspeed-sensor (OMS Race) on the car – /oms_race/vel_and_ang Ground Truth Map The corresponding map CSV contains the GPS coordinates of all cones, measured using an RTK GPS, along with their respective colors.