The dataset is uploaded here. The rosbags are all compressed. To use the dataset, you can first download and decompress the rosbags.
(1)Data information
/camera/color/image_raw # D435i image, 1280 * 720, 30Hz
/camera/depth/image_rect_raw # D435i depth image, 1280 * 720, 30Hz
/camera/imu # D435i IMU data
/imu/data # IMU data, 100Hz
/ouster/points # Ouster Lidar point cloud, 10Hz
/ouster/imu # Ouster IMU, 100Hz
/radar/points # Radar point cloud, 10Hz(2)Extrinsic and intrinsic parameters
# Camera intrinsic parameters
FocalLength=[921.376281738281, 921.494750976562]
principalPoint = [624.489196777344, 368.375915527344];
imageSize = [1280, 720];
RadialDistortion = [0,0];
TangentialDistortion = [0,0];
# IMU-Camera (IMU inside the camera) extrinsic parameters
# Transformation from IMU to camera
T = [0.999976,0.000693444,0.00683458, 0.0201056953519583,
-0.000695267,1.0 , 0.000264293, -0.00504042906686664,
-0.0068344,-0.000269039, 0.999977, -0.0114226331934333,
0.0,0.0,0.0,1.0]
# Lidar-Camera extrinsic parameters
t = [0.250620404253477, -0.036564016917019, -0.112904341600731]
R = [0.004811388191894, -1.003550418032114, -0.003116251098663;
0.009193652813973, -0.003796995446461, -0.991628647673876;
1.004854476258477, 0.009184319190004, 0.016948508611267]
PointsInCamera = R * (PointsInLidar - t');
# Radar-Lidar extrinsic parameters
t = [0.073, -0.0103, 0.137];
R = [-0.013746383632443, 0.999905514004713, 0;
-0.999905514004713, -0.013746383632443, 0;
0 , 0 , 1.00];
PointsInLidar = R * PointsInRadar + t';
# Radar-Camera extrinsic parameters
t = [-0.027990630551015, -0.249545008991387, -0.174005935288756];
R = [1.003389457384155, 0.018606122623860, -0.003116251098663;
0.003670257205003, 0.009244979098594, -0.991628647673876;
-0.022996566525891, 1.004633280408180, 0.016948508611267];
PointsInCamera = R * PointsInRadar + t';
# Radar-IMU(independent IMU) extrinsic parameters
R = [0, -1, 0;
1, 0, 0;
0, 0, 1]
t = [-0.032, 0.038, 0.084]
Under preparation....