KAIST数据集及使用

KAIST 复杂城市数据集

KAIST-Urban-数据集-论文阅读

数据集下载：Complex Urban Dataset 复杂城市数据集

KAIST数据集转换为rosbag

1.将 .gz.tar 文件解压到其文件夹中

find . -name 'urban28*.tar.gz' -execdir tar -xzvf '{}' \;

使用这行命令速度较慢，可以一个一个解压

tar -tzvf xx名.tar.gz

2.克隆并构建此存储库

mkdir -p kaist2bag_ws/src
cd kaist2bag_ws/src
git clone git@github.com:RPM-Robotics-Lab/irp_sen_msg.git
git clone git@github.com:tsyxyz/kaist2bag.git
cd ..
catkin build

3.使用路径和所需主题编辑配置文件

在 kaist2bag/config/config.yaml 下

只需修改 dataset 和 save_to 即可。

# dataset: "/home/tao/sda/dataset/kaist/urban30/urban30-gangnam"
dataset: "/mnt/hgfs/Shared/datasets/kaist_Urban/urban27/urban27-dongtan"
# save_to: "/home/tao/sda/dataset/kaist/urban30/urban30-gangnam/bag"
save_to: "/mnt/hgfs/Shared/datasets/kaist_Urban/urban27/urban27-dongtan/bag"

# ================ 传感器开关 ================
# altimeter: true      # 对应 sensor_data/altimeter.csv
# encoder: true        # 对应 sensor_data/encoder.csv
# fog: true            # 对应 sensor_data/fog.csv
# gps: true            # 对应 sensor_data/gps.csv
# vrs: true            # 对应 sensor_data/vrs_gps.csv
# imu: true            # 对应 sensor_data/xsens_imu.csv
# velodyne: true       # 对应 sensor_data/VLP_left 和 VLP_right
# sick: true           # 对应 sensor_data/SICK_back 和 SICK_middle
# stereo: true         # 对应 image/stereo_left 和 stereo_right

# ================ Topic  ================
# 时间戳文件（自动关联）
# - data_stamp.csv
# - *_stamp.csv

# 高度计
altimeter: true
altimeter_topic: "/altimeter_data"

# 编码器
encoder: true
encoder_irp_topic: "/encoder_count" # 原始脉冲计数
encoder_raw_topic: "/joint_states"  # 关节状态

# 光纤陀螺
fog: true
fog_topic: "/dsp1760_data"

# GPS
gps: true               
gps_topic: "/gps/fix"       # 标准GPS数据      
vrs: true
vrs_topic: "/vrs_gps_data"  # 虚拟参考站GPS

# IMU
imu: true
imu_irp_topic: "/xsens_imu_data"  # 原始二进制数据（需确认格式）
imu_raw_topic: "/imu/data_raw"    # ROS标准IMU数据
imu_mag_topic: "/imu/mag"         # 磁力计数据

# Velodyne激光雷达
velodyne: true
velodyne_left_topic: "/ns2/velodyne_points"   # 左雷达点云
velodyne_right_topic: "/ns1/velodyne_points"  # 右雷达点云

# SICK激光雷达
sick: true
sick_back_topic: "/lms511_back/scan"      # 后置SICK扫描数据
sick_middle_topic: "/lms511_middle/scan"  # 中置SICK扫描数据

# 双目相机
stereo: true
stereo_left_topic: "/stereo/left/image_raw"   # 左目图像
stereo_right_topic: "/stereo/right/image_raw" # 右目图像

4.为每种传感器类型创建一个 rosbag 文件

source devel/setup.bash
roslaunch kaist2bag kaist2bag.launch

5.将所有 bag 合并为一个

rosrun kaist2bag mergebag.py merged.bag /mnt/hgfs/Shared/datasets/kaist_Urban/urban27/urban27-dongtan/bag/*.bag

或

rosrun kaist2bag mergebag.py merged.bag <bag_file_1> ... <bag_file_8>

其中 merged.bag 是保存的路径。

合并后，结果如下：

damon@damon-virtual-machine:~/kaist2bag_ws$ rosbag info merged.bag 
path:        merged.bag
version:     2.0
duration:    19:20s (1160s)
start:       Dec 12 2018 10:44:08.84 (1544582648.84)
end:         Dec 12 2018 11:03:29.23 (1544583809.23)
size:        28.6 GB
messages:    2038500
compression: none [24116/24116 chunks]
types:       irp_sen_msgs/LaserScanArray [590075c3653e9b2cc1c75018fcbeedde]
             irp_sen_msgs/altimeter      [e6c99c37e6f9fe98e071d524cc164e65]
             irp_sen_msgs/encoder        [360abb290fcf41163977621351ecd430]
             irp_sen_msgs/fog_3axis      [2f13bb2da647a41df38407a7a8adc90a]
             irp_sen_msgs/imu            [97cd0887761aab329eff97e8b6aab17e]
             irp_sen_msgs/vrs            [1a83e0593b51691190a14fffd0fc6b2c]
             sensor_msgs/Image           [060021388200f6f0f447d0fcd9c64743]
             sensor_msgs/Imu             [6a62c6daae103f4ff57a132d6f95cec2]
             sensor_msgs/JointState      [3066dcd76a6cfaef579bd0f34173e9fd]
             sensor_msgs/MagneticField   [2f3b0b43eed0c9501de0fa3ff89a45aa]
             sensor_msgs/NavSatFix       [2d3a8cd499b9b4a0249fb98fd05cfa48]
             sensor_msgs/PointCloud2     [1158d486dd51d683ce2f1be655c3c181]
topics:      /altimeter_data             11605 msgs    : irp_sen_msgs/altimeter     
             /dsp1760_data             1160508 msgs    : irp_sen_msgs/fog_3axis     
             /encoder_count             116048 msgs    : irp_sen_msgs/encoder       
             /gps/fix                     7017 msgs    : sensor_msgs/NavSatFix      
             /imu/data_raw              116054 msgs    : sensor_msgs/Imu            
             /imu/mag                   116054 msgs    : sensor_msgs/MagneticField  
             /joint_states              116047 msgs    : sensor_msgs/JointState     
             /lms511_back/scan          115870 msgs    : irp_sen_msgs/LaserScanArray
             /lms511_middle/scan        115865 msgs    : irp_sen_msgs/LaserScanArray
             /ns1/velodyne_points        11506 msgs    : sensor_msgs/PointCloud2    
             /ns2/velodyne_points        11505 msgs    : sensor_msgs/PointCloud2    
             /stereo/left/image_raw      11605 msgs    : sensor_msgs/Image          
             /stereo/right/image_raw     11601 msgs    : sensor_msgs/Image          
             /vrs_gps_data                1161 msgs    : irp_sen_msgs/vrs           
             /xsens_imu_data            116054 msgs    : irp_sen_msgs/imu

参考

KAIST数据集转换为rosbag