鱼香ROS社区
    • 版块
    • 最新
    • 未解决
    • 已解决
    • 群组
    • 注册
    • 登录
    紧急通知：禁止一切关于政治&VPN翻墙等话题，发现相关帖子会立马删除封号
     小鱼 ROS 2 新书上线！点击链接查看， 新书配套视频点击链接查看。
    提问前必看的发帖注意事项—— 提问前必看！不符合要求的问题拒绝回答！！
    社区使用指南—如何添加标签修改密码

    cartographer进行大范围建图出现卡顿

    已定时 已固定 已锁定 已移动 未解决
    仿真
    cartographer 仿真
    2
    10
    527
    正在加载更多帖子
    • 从旧到新
    • 从新到旧
    • 最多赞同
    回复
    • 在新帖中回复
    登录后回复
    此主题已被删除。只有拥有主题管理权限的用户可以查看。
    • 24758873092
      量子之海里的卡夫卡
      最后由 编辑

      标题：使用cartographer在仿真环境中进行大范围建图出现卡顿

      背景：

      需要实现大范围建图，实现扫描整个校园，之后进行导航。暂时使用仿真模拟大范围建图，但仿真范围相对于校园是比较小的。

      问题描述：

      硬件平台 nuc x86平台
      系统版本 ubuntu20.04
      ros版本 foxy

      使用cartographer在仿真环境中模拟大范围建图一段时间后，rviz显示走过的路与map更新不统一，map更新较为迟缓甚至出现卡顿，需要等一段时间（可能10分钟，或者半小时，甚至更长）后rviz的map才会更新到最新的点。曾经尝试过等待，卡顿的时间并不是线性的，感觉是指数级增长，且内存占用也比较高，建图一段时间后，内存使用量从原来的2gb左右（正常使用情况，刚运行的时候）上升到7gb左右 如下图所示：
      2cb686e3-5926-4ed7-8bc9-ce01c99b6f78-Screenshot from 2024-04-10 15-24-50.png

      如下图所示，是卡顿的情况：
      23e2b0ac-34eb-49c8-bde2-278c63364f60-Screenshot from 2024-04-10 15-08-52.png
      仿真模型如下图所示：
      6d599aba-4159-4926-884e-2e970be0ea8b-Screenshot from 2024-04-10 15-09-25.png

      刚出现卡顿时一部分rviz 日志：

      [cartographer_node-1] [INFO] [1712731552.511735516] [cartographer_ros]: I0410 14:45:52.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.61 Hz 5.68e-02 s +/- 1.73e-02 s (pulsed at 99.71% real time)
[cartographer_node-1] [INFO] [1712731552.511765476] [cartographer_ros]: I0410 14:45:52.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 6.62e-04 s (pulsed at 99.44% real time)
[cartographer_node-1] [INFO] [1712731567.527431246] [cartographer_ros]: I0410 14:46:07.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.36 Hz 5.76e-02 s +/- 1.88e-02 s (pulsed at 99.32% real time)
[cartographer_node-1] [INFO] [1712731567.527496576] [cartographer_ros]: I0410 14:46:07.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 6.08e-04 s (pulsed at 99.69% real time)
[cartographer_node-1] [INFO] [1712731582.551846932] [cartographer_ros]: I0410 14:46:22.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.84 Hz 5.61e-02 s +/- 1.73e-02 s (pulsed at 99.15% real time)
[cartographer_node-1] [INFO] [1712731582.551876766] [cartographer_ros]: I0410 14:46:22.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 4.38e-04 s (pulsed at 99.82% real time)
[cartographer_node-1] [INFO] [1712731597.566277359] [cartographer_ros]: I0410 14:46:37.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.79 Hz 5.62e-02 s +/- 1.78e-02 s (pulsed at 99.54% real time)
[cartographer_node-1] [INFO] [1712731597.566311384] [cartographer_ros]: I0410 14:46:37.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 5.10e-04 s (pulsed at 99.64% real time)
[cartographer_node-1] [INFO] [1712731612.617004872] [cartographer_ros]: I0410 14:46:52.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.62 Hz 5.68e-02 s +/- 1.68e-02 s (pulsed at 99.52% real time)
[cartographer_node-1] [INFO] [1712731612.617035914] [cartographer_ros]: I0410 14:46:52.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 6.82e-04 s (pulsed at 99.43% real time)
[cartographer_node-1] [INFO] [1712731627.630846753] [cartographer_ros]: I0410 14:47:07.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.59 Hz 5.68e-02 s +/- 1.65e-02 s (pulsed at 99.46% real time)
[cartographer_node-1] [INFO] [1712731627.630900505] [cartographer_ros]: I0410 14:47:07.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 6.30e-04 s (pulsed at 99.96% real time)
[cartographer_node-1] [INFO] [1712731628.793949106] [cartographer_ros]: I0410 14:47:08.000000 24051 motion_filter.cc:42] Motion filter reduced the number of nodes to 47.9781%.
[cartographer_node-1] [INFO] [1712731642.663164354] [cartographer_ros]: I0410 14:47:22.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.73 Hz 5.64e-02 s +/- 1.61e-02 s (pulsed at 99.38% real time)
[cartographer_node-1] [INFO] [1712731642.663195643] [cartographer_ros]: I0410 14:47:22.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 4.97e-04 s (pulsed at 100.04% real time)
[cartographer_node-1] [INFO] [1712731657.712615266] [cartographer_ros]: I0410 14:47:37.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.75 Hz 5.63e-02 s +/- 1.67e-02 s (pulsed at 99.52% real time)
[cartographer_node-1] [INFO] [1712731657.712651190] [cartographer_ros]: I0410 14:47:37.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 5.49e-04 s (pulsed at 99.73% real time)
[cartographer_node-1] [INFO] [1712731672.750544375] [cartographer_ros]: I0410 14:47:52.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.59 Hz 5.68e-02 s +/- 1.68e-02 s (pulsed at 99.40% real time)
[cartographer_node-1] [INFO] [1712731672.750597528] [cartographer_ros]: I0410 14:47:52.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 7.02e-04 s (pulsed at 99.91% real time)
[cartographer_node-1] [INFO] [1712731687.823128666] [cartographer_ros]: I0410 14:48:07.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.65 Hz 5.67e-02 s +/- 1.63e-02 s (pulsed at 99.28% real time)
[cartographer_node-1] [INFO] [1712731687.823228259] [cartographer_ros]: I0410 14:48:07.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 4.68e-04 s (pulsed at 99.89% real time)
[cartographer_node-1] [INFO] [1712731702.862857259] [cartographer_ros]: I0410 14:48:22.000000 24051 collated_trajectory_builder.cc:72] odom rate: 20.99 Hz 4.76e-02 s +/- 1.69e-02 s (pulsed at 100.36% real time)
[cartographer_node-1] [INFO] [1712731702.864018212] [cartographer_ros]: I0410 14:48:22.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 2.87e-04 s (pulsed at 99.80% real time)

      等待30分钟后一部分rviz日志：

      [cartographer_node-1] [INFO] [1712733927.665480146] [cartographer_ros]: I0410 15:25:27.000000 24051 collated_trajectory_builder.cc:72] odom rate: 15.85 Hz 6.31e-02 s +/- 2.21e-02 s (pulsed at 99.96% real time)
[cartographer_node-1] [INFO] [1712733927.668576325] [cartographer_ros]: I0410 15:25:27.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 8.68e-04 s (pulsed at 99.75% real time)
[cartographer_node-1] [INFO] [1712733934.773953652] [cartographer_ros]: I0410 15:25:34.000000 24051 motion_filter.cc:42] Motion filter reduced the number of nodes to 36.3149%.
[cartographer_node-1] [INFO] [1712733942.694428955] [cartographer_ros]: I0410 15:25:42.000000 24051 collated_trajectory_builder.cc:72] odom rate: 16.19 Hz 6.18e-02 s +/- 2.04e-02 s (pulsed at 99.84% real time)
[cartographer_node-1] [INFO] [1712733942.694463213] [cartographer_ros]: I0410 15:25:42.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 7.31e-04 s (pulsed at 99.54% real time)
[cartographer_node-1] [INFO] [1712733957.738089993] [cartographer_ros]: I0410 15:25:57.000000 24051 collated_trajectory_builder.cc:72] odom rate: 15.58 Hz 6.42e-02 s +/- 2.16e-02 s (pulsed at 99.92% real time)
[cartographer_node-1] [INFO] [1712733957.738118990] [cartographer_ros]: I0410 15:25:57.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 8.99e-04 s (pulsed at 99.31% real time)
[cartographer_node-1] [INFO] [1712733972.801014655] [cartographer_ros]: I0410 15:26:12.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.79 Hz 5.62e-02 s +/- 1.92e-02 s (pulsed at 99.33% real time)
[cartographer_node-1] [INFO] [1712733972.809865571] [cartographer_ros]: I0410 15:26:12.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 7.49e-04 s (pulsed at 99.81% real time)
[cartographer_node-1] [INFO] [1712733987.852491175] [cartographer_ros]: I0410 15:26:27.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.27 Hz 5.79e-02 s +/- 2.08e-02 s (pulsed at 99.08% real time)
[cartographer_node-1] [INFO] [1712733987.852517236] [cartographer_ros]: I0410 15:26:27.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 7.49e-04 s (pulsed at 99.87% real time)
[cartographer_node-1] [INFO] [1712734002.953640554] [cartographer_ros]: I0410 15:26:42.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.38 Hz 5.75e-02 s +/- 1.89e-02 s (pulsed at 99.21% real time)
[cartographer_node-1] [INFO] [1712734002.954095649] [cartographer_ros]: I0410 15:26:42.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 8.36e-04 s (pulsed at 99.94% real time)
[cartographer_node-1] [INFO] [1712734018.003455380] [cartographer_ros]: I0410 15:26:58.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.05 Hz 5.87e-02 s +/- 2.23e-02 s (pulsed at 100.39% real time)
[cartographer_node-1] [INFO] [1712734018.005777640] [cartographer_ros]: I0410 15:26:58.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 7.85e-04 s (pulsed at 99.80% real time)
[cartographer_node-1] [INFO] [1712734033.051672753] [cartographer_ros]: I0410 15:27:13.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.08 Hz 5.85e-02 s +/- 2.19e-02 s (pulsed at 100.08% real time)
[cartographer_node-1] [INFO] [1712734033.052880943] [cartographer_ros]: I0410 15:27:13.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 7.76e-04 s (pulsed at 99.45% real time)
[cartographer_node-1] [INFO] [1712734035.161934350] [cartographer_ros]: I0410 15:27:15.000000 24051 motion_filter.cc:42] Motion filter reduced the number of nodes to 35.9477%.
[cartographer_node-1] [INFO] [1712734048.053592713] [cartographer_ros]: I0410 15:27:28.000000 24051 collated_trajectory_builder.cc:72] odom rate: 17.54 Hz 5.70e-02 s +/- 1.95e-02 s (pulsed at 99.14% real time)
[cartographer_node-1] [INFO] [1712734048.053625189] [cartographer_ros]: I0410 15:27:28.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 8.84e-04 s (pulsed at 99.90% real time)
[cartographer_node-1] [INFO] [1712734063.086749043] [cartographer_ros]: I0410 15:27:43.000000 24051 collated_trajectory_builder.cc:72] odom rate: 16.69 Hz 5.99e-02 s +/- 2.02e-02 s (pulsed at 99.19% real time)
[cartographer_node-1] [INFO] [1712734063.086786005] [cartographer_ros]: I0410 15:27:43.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 7.94e-04 s (pulsed at 99.68% real time)
[cartographer_node-1] [INFO] [1712734078.141211460] [cartographer_ros]: I0410 15:27:58.000000 24051 collated_trajectory_builder.cc:72] odom rate: 16.83 Hz 5.94e-02 s +/- 1.99e-02 s (pulsed at 99.08% real time)
[cartographer_node-1] [INFO] [1712734078.141237015] [cartographer_ros]: I0410 15:27:58.000000 24051 collated_trajectory_builder.cc:72] scan rate: 5.00 Hz 2.00e-01 s +/- 7.76e-04 s (pulsed at 99.87% real time)

      尝试过的解决方法：

      一开始修改过lua文件，根据cartographer官方提供的调优文档修改过，但效果并不明显。
      后又修改过launch文件中地图分辨率从原来的0.05修改为0.4，以及修改map更新频率从原来的0.5修改为1，建图范围增大的较为明显。但还是会出现卡顿的情况。

      小鱼小 1 条回复 最后回复 回复 引用 0
      • 24758873092 2475887309 将这个主题转为问答主题，在
      • 小鱼小
        小鱼 技术大佬 @2475887309
        最后由 编辑

        @2475887309 大场景确实会消耗更多的内存，而且还是在仿真情况下，需要对参数进一步的调整，不止是更新频率和分辨率，还要调整子图数量之类的参数，建议附上完整的参数配置，并详细尝试调整其他参数进行测试。

        小鱼 ROS 2 新书上线！书籍购买链接： https://m.tb.cn/h.gFDBx39
        新书配套视频：https://www.bilibili.com/video/BV1GW42197Ck/

        24758873092 1 条回复 最后回复 回复 引用 0
        • 24758873092
          量子之海里的卡夫卡 @小鱼
          最后由 编辑

          @小鱼 大佬您好，参数配置是根据您的教程来的，就launch文件修改了一下分辨率以及更新地图，其他的例如lua文件没有修改。现在需要修改的是哪部分参数啊

          小鱼小 1 条回复 最后回复 回复 引用 0
          • 小鱼小
            小鱼 技术大佬 @2475887309
            最后由 编辑

            @2475887309 你倒是把你现在的参数贴出来呀，大兄弟

            小鱼 ROS 2 新书上线！书籍购买链接： https://m.tb.cn/h.gFDBx39
            新书配套视频：https://www.bilibili.com/video/BV1GW42197Ck/

            24758873092 1 条回复 最后回复 回复 引用 0
            • 24758873092
              量子之海里的卡夫卡 @小鱼
              最后由 编辑

              @小鱼 不好意思，昨天忙着做其他事情去了，就没看论坛

              lua配置文件如下```
              include "map_builder.lua"
              include "trajectory_builder.lua"

              options = {
              map_builder = MAP_BUILDER,
              trajectory_builder = TRAJECTORY_BUILDER,
              map_frame = "map",
              tracking_frame = "base_link", --base_link
              -- base_link改为odom,发布map到odom之间的位姿态
              published_frame = "odom", --odom
              odom_frame = "odom", --odom
              -- true改为false，不用提供里程计数据
              provide_odom_frame = false,
              -- false改为true，仅发布2D位资
              publish_frame_projected_to_2d = true,
              -- false改为true，使用里程计数据
              use_odometry = true,
              use_nav_sat = false,
              use_landmarks = false,
              -- 0改为1,使用一个雷达
              num_laser_scans = 1,
              -- 1改为0，不使用多波雷达
              num_multi_echo_laser_scans = 0,
              -- 10改为1，1/1=1等于不分割
              num_subdivisions_per_laser_scan = 1,
              num_point_clouds = 0,
              lookup_transform_timeout_sec = 0.2,
              submap_publish_period_sec = 0.3,
              pose_publish_period_sec = 5e-3,
              trajectory_publish_period_sec = 30e-3,
              rangefinder_sampling_ratio = 1.,
              odometry_sampling_ratio = 1.,
              fixed_frame_pose_sampling_ratio = 1.,
              imu_sampling_ratio = 1.,
              landmarks_sampling_ratio = 1.,
              }

              -- false改为true，启动2D SLAM
              MAP_BUILDER.use_trajectory_builder_2d = true
              -- 0改成0.10,比机器人半径小的都忽略
              TRAJECTORY_BUILDER_2D.min_range = 0.10 --0.10
              -- 30改成3.5,限制在雷达最大扫描范围内，越小一般越精确些
              TRAJECTORY_BUILDER_2D.max_range = 30 --5
              -- 5改成3,传感器数据超出有效范围最大值
              TRAJECTORY_BUILDER_2D.missing_data_ray_length = 30. --3.0
              -- true改成false,不使用IMU数据，大家可以开启，然后对比下效果
              TRAJECTORY_BUILDER_2D.use_imu_data = false
              -- false改成true,使用实时回环检测来进行前端的扫描匹配
              TRAJECTORY_BUILDER_2D.use_online_correlative_scan_matching = true
              -- 1.0改成0.1,提高对运动的敏感度
              TRAJECTORY_BUILDER_2D.motion_filter.max_angle_radians = math.rad(0.1) --0.1

              -- 0.55改成0.65,Fast csm的最低分数，高于此分数才进行优化。
              POSE_GRAPH.constraint_builder.min_score = 0.65 --0.65
              --0.6改成0.7,全局定位最小分数，低于此分数则认为目前全局定位不准确
              POSE_GRAPH.constraint_builder.global_localization_min_score = 0.7 --0.7

              -- 设置0可关闭全局SLAM
              POSE_GRAPH.optimize_every_n_nodes = 0

              return options

              launch文件如下

              import os
              from launch import LaunchDescription
              from launch.substitutions import LaunchConfiguration
              from launch_ros.actions import Node
              from launch_ros.substitutions import FindPackageShare

              def generate_launch_description():
              # 定位到功能包的地址
              pkg_share = FindPackageShare(package='fishbot_cartographer').find('fishbot_cartographer')

              #=====================运行节点需要的配置=======================================================================
# 是否使用仿真时间，我们用gazebo，这里设置成true
use_sim_time = LaunchConfiguration('use_sim_time', default='true')
# 地图的分辨率--0.05
resolution = LaunchConfiguration('resolution', default='0.4')
# 地图的发布周期--0.5
publish_period_sec = LaunchConfiguration('publish_period_sec', default='1')
# 配置文件夹路径
configuration_directory = LaunchConfiguration('configuration_directory',default= os.path.join(pkg_share, 'config') )
# 配置文件
configuration_basename = LaunchConfiguration('configuration_basename', default='fishbot_2d.lua')


#=====================声明三个节点，cartographer/occupancy_grid_node/rviz_node=================================
cartographer_node = Node(
    package='cartographer_ros',
    executable='cartographer_node',
    name='cartographer_node',
    output='screen',
    parameters=[{'use_sim_time': use_sim_time}],
    arguments=['-configuration_directory', configuration_directory,
               '-configuration_basename', configuration_basename])

occupancy_grid_node = Node(
    package='cartographer_ros',
    executable='occupancy_grid_node',
    name='occupancy_grid_node',
    output='screen',
    parameters=[{'use_sim_time': use_sim_time}],
    arguments=['-resolution', resolution, '-publish_period_sec', publish_period_sec])

rviz_node = Node(
    package='rviz2',
    executable='rviz2',
    name='rviz2',
    # arguments=['-d', rviz_config_dir],
    parameters=[{'use_sim_time': use_sim_time}],
    output='screen')

#===============================================定义启动文件========================================================
ld = LaunchDescription()
ld.add_action(cartographer_node)
ld.add_action(occupancy_grid_node)
ld.add_action(rviz_node)

return ld
              1 条回复 最后回复 回复 引用 0
              • 24758873092
                量子之海里的卡夫卡
                最后由 编辑

                格式有点不对，我重新发以下配置
                lua文件配置

                include "map_builder.lua"
include "trajectory_builder.lua"

options = {
  map_builder = MAP_BUILDER,
  trajectory_builder = TRAJECTORY_BUILDER,
  map_frame = "map",
  tracking_frame = "base_link",  --base_link
  -- base_link改为odom,发布map到odom之间的位姿态
  published_frame = "odom",   --odom
  odom_frame = "odom",  --odom 
  -- true改为false，不用提供里程计数据
  provide_odom_frame = false,
  -- false改为true，仅发布2D位资
  publish_frame_projected_to_2d = true,
  -- false改为true，使用里程计数据
  use_odometry = true,
  use_nav_sat = false,
  use_landmarks = false,
  -- 0改为1,使用一个雷达
  num_laser_scans = 1,
  -- 1改为0，不使用多波雷达
  num_multi_echo_laser_scans = 0,
  -- 10改为1，1/1=1等于不分割
  num_subdivisions_per_laser_scan = 1,
  num_point_clouds = 0,
  lookup_transform_timeout_sec = 0.2,
  submap_publish_period_sec = 0.3,
  pose_publish_period_sec = 5e-3,
  trajectory_publish_period_sec = 30e-3,
  rangefinder_sampling_ratio = 1.,
  odometry_sampling_ratio = 1.,
  fixed_frame_pose_sampling_ratio = 1.,
  imu_sampling_ratio = 1.,
  landmarks_sampling_ratio = 1.,
}


-- false改为true，启动2D SLAM
MAP_BUILDER.use_trajectory_builder_2d = true
-- 0改成0.10,比机器人半径小的都忽略
TRAJECTORY_BUILDER_2D.min_range = 0.10  --0.10
-- 30改成3.5,限制在雷达最大扫描范围内，越小一般越精确些
TRAJECTORY_BUILDER_2D.max_range = 30   --5
-- 5改成3,传感器数据超出有效范围最大值
TRAJECTORY_BUILDER_2D.missing_data_ray_length = 30. --3.0
-- true改成false,不使用IMU数据，大家可以开启，然后对比下效果
TRAJECTORY_BUILDER_2D.use_imu_data = false
-- false改成true,使用实时回环检测来进行前端的扫描匹配
TRAJECTORY_BUILDER_2D.use_online_correlative_scan_matching = true 
-- 1.0改成0.1,提高对运动的敏感度
TRAJECTORY_BUILDER_2D.motion_filter.max_angle_radians = math.rad(0.1) --0.1

-- 0.55改成0.65,Fast csm的最低分数，高于此分数才进行优化。
POSE_GRAPH.constraint_builder.min_score = 0.65       --0.65
--0.6改成0.7,全局定位最小分数，低于此分数则认为目前全局定位不准确
POSE_GRAPH.constraint_builder.global_localization_min_score = 0.7     --0.7

-- 设置0可关闭全局SLAM
POSE_GRAPH.optimize_every_n_nodes = 0

return options

                launch文件配置

                import os
from launch import LaunchDescription
from launch.substitutions import LaunchConfiguration
from launch_ros.actions import Node
from launch_ros.substitutions import FindPackageShare


def generate_launch_description():
    # 定位到功能包的地址
    pkg_share = FindPackageShare(package='fishbot_cartographer').find('fishbot_cartographer')
    
    #=====================运行节点需要的配置=======================================================================
    # 是否使用仿真时间，我们用gazebo，这里设置成true
    use_sim_time = LaunchConfiguration('use_sim_time', default='true')
    # 地图的分辨率--0.05
    resolution = LaunchConfiguration('resolution', default='0.4')
    # 地图的发布周期--0.5
    publish_period_sec = LaunchConfiguration('publish_period_sec', default='1')
    # 配置文件夹路径
    configuration_directory = LaunchConfiguration('configuration_directory',default= os.path.join(pkg_share, 'config') )
    # 配置文件
    configuration_basename = LaunchConfiguration('configuration_basename', default='fishbot_2d.lua')

    
    #=====================声明三个节点，cartographer/occupancy_grid_node/rviz_node=================================
    cartographer_node = Node(
        package='cartographer_ros',
        executable='cartographer_node',
        name='cartographer_node',
        output='screen',
        parameters=[{'use_sim_time': use_sim_time}],
        arguments=['-configuration_directory', configuration_directory,
                   '-configuration_basename', configuration_basename])

    occupancy_grid_node = Node(
        package='cartographer_ros',
        executable='occupancy_grid_node',
        name='occupancy_grid_node',
        output='screen',
        parameters=[{'use_sim_time': use_sim_time}],
        arguments=['-resolution', resolution, '-publish_period_sec', publish_period_sec])

    rviz_node = Node(
        package='rviz2',
        executable='rviz2',
        name='rviz2',
        # arguments=['-d', rviz_config_dir],
        parameters=[{'use_sim_time': use_sim_time}],
        output='screen')

    #===============================================定义启动文件========================================================
    ld = LaunchDescription()
    ld.add_action(cartographer_node)
    ld.add_action(occupancy_grid_node)
    ld.add_action(rviz_node)

    return ld
                小鱼小 1 条回复 最后回复 回复 引用 0
                • 小鱼小
                  小鱼 技术大佬 @2475887309
                  最后由 编辑

                  @2475887309 在 cartographer进行大范围建图出现卡顿 中说：

                  TRAJECTORY_BUILDER_2D.use_online_correlative_scan_matching = true

                  比如这个参数就可以关掉，应该能剩不少算力

                  以下内容由GPT给出：

                  减小扫描匹配参数：

                  TRAJECTORY_BUILDER_2D.min_range: 减小此参数可以忽略机器人附近太近的数据，减少计算量。
                  TRAJECTORY_BUILDER_2D.max_range: 限制传感器数据的最大范围可以减少计算负载，避免处理超出此范围的数据。
                  TRAJECTORY_BUILDER_2D.missing_data_ray_length: 调整此参数可以有效处理缺失或无效的传感器数据，从而减少不必要的计算。
                  禁用未使用的功能：

                  TRAJECTORY_BUILDER_2D.use_imu_data: 如果不使用IMU数据，将此设置为false可以防止与IMU数据相关的不必要处理。
                  use_nav_sat、use_landmarks: 如果不使用卫星导航或地标，则确保这些参数为false。
                  优化扫描匹配参数：

                  TRAJECTORY_BUILDER_2D.use_online_correlative_scan_matching: 如果系统允许，您可以禁用在线扫描匹配以减少实时计算。但是，这可能会影响SLAM的质量。
                  调整运动滤波器灵敏度：

                  TRAJECTORY_BUILDER_2D.motion_filter.max_angle_radians: 减小此参数会增加对运动的敏感性，这可能会改善对运动的响应，但也可能略微增加计算量。您可以根据机器人的动态特性进行微调。
                  约束构建参数：

                  POSE_GRAPH.constraint_builder.min_score: 降低此值可能会减少在优化过程中考虑的约束数量，从而减少计算负载。但这可能会影响生成地图的准确性。
                  POSE_GRAPH.constraint_builder.global_localization_min_score: 同样，降低此值可能会减少全局定位尝试的次数，节省计算资源。
                  全局SLAM优化：

                  POSE_GRAPH.optimize_every_n_nodes: 将此设置为0可以有效地禁用全局SLAM优化，这可以节省大量计算资源。但是，这会以随时间降低地图质量为代价。

                  小鱼 ROS 2 新书上线！书籍购买链接： https://m.tb.cn/h.gFDBx39
                  新书配套视频：https://www.bilibili.com/video/BV1GW42197Ck/

                  24758873092 2 条回复 最后回复 回复 引用 0
                  • 24758873092
                    量子之海里的卡夫卡 @小鱼
                    最后由 编辑

                    @小鱼 好的大佬，我尝试一下

                    1 条回复 最后回复 回复 引用 0
                    • 24758873092
                      量子之海里的卡夫卡 @小鱼
                      最后由 编辑

                      @小鱼 大佬，我又回来了，在修改以上提到的参数过后，确实有所改善，但是建图效果并没有之前的好了，而且还是会出现卡顿的情况，只不过就是比调整之前的建图范围大了一些。
                      然后我去cartographer的github上面寻找到一个issue，问题与我类似，但似乎并没有解决，这个问题似乎是因为迁移到ros2中形成的一个bug，我把链接放下面，如果大佬有时间的话，可否看一下，如果是因为这个原因，我是否可以更换算法，比如您发的Lego-LOAM这个算法

                      https://github.com/ros2/cartographer_ros/issues/36
                      小鱼小 1 条回复 最后回复 回复 引用 0
                      • 小鱼小
                        小鱼 技术大佬 @2475887309
                        最后由 编辑

                        @2475887309 可以试试一些其他版本的建图方法，比如slam_toolbox 或者 gmapping等

                        小鱼 ROS 2 新书上线！书籍购买链接： https://m.tb.cn/h.gFDBx39
                        新书配套视频：https://www.bilibili.com/video/BV1GW42197Ck/

                        1 条回复 最后回复 回复 引用 0
                        • 第一个帖子
                          最后一个帖子
                        皖ICP备16016415号-7
                        Powered by NodeBB | 鱼香ROS