Datasets#

Autoware partners provide datasets for testing and development. These datasets are available for download here.

Istanbul Open Dataset#

The dataset is collected in the following route. Tunnels and bridges are annotated on the image. The included specific areas into the dataset are:

Galata Bridge (Small Bridge)
Eurasia Tunnel (Long Tunnel with High Elevation Changes)
2nd Bosphorus Bridge (Long Bridge)
Kagithane-Bomonti Tunnel (Small Tunnel)
Viaducts, road junctions, highways, dense urban areas...

ist_dataset_route2

Leo Drive - Mapping Kit Sensor Data#

This dataset contains data from the portable mapping kit used for general mapping purposes.

The data contains data from the following sensors:

1 x Applanix POS LVX GNSS/INS System
1 x Hesai Pandar XT32 LiDAR

For sensor calibrations, /tf_static topic is added.

Data Links#

You can find the produced full point cloud map, corner feature point cloud map and surface feature point cloud map here:
- https://drive.google.com/drive/folders/1_jiQod4lO6-V2NDEr3d-M3XF_Nqmc0Xf?usp=drive_link
- Exported point clouds are exported via downsampling with 0.2 meters and 0.5 meters voxel grids.

You can find the ROS 2 bag which is collected simultaneously with the mapping data:
- https://drive.google.com/drive/folders/17zXiBeYlM90gQ5hV6EAWaoBTnNFoVPML?usp=drive_link
- Due to the simultaneous data collection, we can assume that the point cloud maps and GNSS/INS data are the ground truth data for this rosbag.

Additionally, you can find the raw data used for mapping at the below link:
- https://drive.google.com/drive/folders/1HmWYkxF5XvVCR27R8W7ZqO7An4HlJ6lD?usp=drive_link
- Point clouds are collected as PCAP and feature-matched GNSS/INS data exported to a txt file.

Localization Performance Evaluation with Autoware#

The report of the performance evaluation of the current Autoware with the collected data can be found in the link below.

The report documented at 2024-08-28.

https://github.com/orgs/autowarefoundation/discussions/5135

Topic list#

For collecting the GNSS/INS data, this repository is used.

For collecting the LiDAR data, nebula repository is used.

Topic Name	Message Type
`/applanix/lvx_client/autoware_orientation`	`autoware_sensing_msgs/msg/GnssInsOrientationStamped`
`/applanix/lvx_client/imu_raw`	`sensor_msgs/msg/Imu`
`/localization/twist_estimator/twist_with_covariance`	`geometry_msgs/msg/TwistWithCovarianceStamped`
`/applanix/lvx_client/odom`	`nav_msgs/msg/Odometry`
`/applanix/lvx_client/gnss/fix`	`sensor_msgs/msg/NavSatFix`
`/clock`	`rosgraph_msgs/msg/Clock`
`/pandar_points`	`sensor_msgs/msg/PointCloud2`
`/tf_static`	`tf2_msgs/msg/TFMessage`

Message Explanations#

Used drivers for sensors give output in default ROS 2 message types and their own ROS 2 message types for additional information. Following topics are the default ROS 2 message types:

/applanix/lvx_client/imu_raw
- Gives the output of INS system in ENU. Due to the 9-axis IMU, yaw value demonstrates the heading value of the sensor.

/applanix/lvx_client/twist_with_covariance
- Gives the twist output of the sensor.

/applanix/lvx_client/odom
- Gives the position and orientation of the sensor from the starting point of the ROS 2 driver. Implemented with GeographicLib::LocalCartesian.
  
  This topic is not related to the wheel odometry.

/applanix/lvx_client/gnss/fix
- Gives the latitude, longitude and height values of the sensors.
  
  Ellipsoidal height of WGS84 ellipsoid is given as height value.

/pandar_points
- Gives the point cloud from the LiDAR sensor.

Bus-ODD (Operational Design Domain) datasets#

Leo Drive - ISUZU sensor data#

This dataset contains data from the Isuzu bus used in the Bus ODD project.

The data contains data from following sensors:

1 x VLP16
2 x VLP32C
1 x Applanix POS LV 120 GNSS/INS
3 x Lucid Vision Triton 5.4MP cameras (left, right, front)
Vehicle status report

It also contains /tf topic for static transformations between sensors.

Required message types#

The GNSS data is available in sensor_msgs/msg/NavSatFix message type.

But also the Applanix raw messages are also included in applanix_msgs/msg/NavigationPerformanceGsof50 and applanix_msgs/msg/NavigationSolutionGsof49 message types. In order to be able to play back these messages, you need to build and source the applanix_msgs package.

# Create a workspace and clone the repository
mkdir -p ~/applanix_ws/src && cd "$_"
git clone https://github.com/autowarefoundation/applanix.git
cd ..

# Build the workspace
colcon build --symlink-install --packages-select applanix_msgs

# Source the workspace
source ~/applanix_ws/install/setup.bash

# Now you can play back the messages

Also make sure to source Autoware Universe workspace too.

Download instructions#

# Install awscli
$ sudo apt update && sudo apt install awscli -y

# This will download the entire dataset to the current directory.
# (About 10.9GB of data)
$ aws s3 sync s3://autoware-files/collected_data/2022-08-22_leo_drive_isuzu_bags/ ./2022-08-22_leo_drive_isuzu_bags  --no-sign-request

# Optionally,
# If you instead want to download a single bag file, you can get a list of the available files with following:
$ aws s3 ls s3://autoware-files/collected_data/2022-08-22_leo_drive_isuzu_bags/ --no-sign-request
   PRE all-sensors-bag1_compressed/
   PRE all-sensors-bag2_compressed/
   PRE all-sensors-bag3_compressed/
   PRE all-sensors-bag4_compressed/
   PRE all-sensors-bag5_compressed/
   PRE all-sensors-bag6_compressed/
   PRE driving_20_kmh_2022_06_10-16_01_55_compressed/
   PRE driving_30_kmh_2022_06_10-15_47_42_compressed/

# Then you can download a single bag file with the following:
aws s3 sync s3://autoware-files/collected_data/2022-08-22_leo_drive_isuzu_bags/all-sensors-bag1_compressed/ ./all-sensors-bag1_compressed  --no-sign-request

AutoCore.ai - lidar ROS 2 bag file and pcap#

This dataset contains pcap files and ros2 bag files from Ouster OS1-64 Lidar. The pcap file and ros2 bag file is recorded in the same time with slight difference in duration.

Click here to download (~553MB)

Reference Issue