trimof

Appearance

ROS2 Navigation Sentry Deployment Guide ​

1 Dependencies ​

There're no rosdep commands because building from source is recommended.

1.1 livox_sdk2 ​

Follow the official guide.

livox sdk github

1.2 Third Packages ​

Third packages are included in our gitlab project due to stability issues. Don't use the official packages.

We have made some changes from the original packages.

sh
git clone https://mirrors.sustech.edu.cn/git/artinx/third_package.git --branch develop
git clone https://mirrors.sustech.edu.cn/git/artinx/third_package.git --branch develop

1.2.1 livox_ros_driver2 ​

Go to package directory and build it using the build.sh script.

sh
./build.sh ROS2
./build.sh ROS2

After which, go to workspace directory and use colcon build --symlink-install to build the whole packages.

Next, you have to set up static ip for mid-360.Follow the instructions on the manual. When you are connected to mid-360, its route would be located at 192.168.1.xx. Set up your ethernets' ip to 192.168.1.xx as well(Follow the instructions on this guide)

If you want to connect to lidar on-boot, use netplan to do the job. However, using netplan would unable the remote-desktop function. Adding a router to distribute the ip would be a solution, but was not tried yet.

To learn more about mid-360, go to inofficial mid-360 doc site.

1.2.2 Fast lio ​

Fast-lio Github

Fast-lio is an odometry predicting algorithm combining IMU and point cloud data. Basically, if you can get the odometry, you can do SLAM(Simultaneously Localization and Mapping) easily.

To completely understand what SLAM is, I recommend you to read through this book written by Bo Gao.

Fast-lio is broadcasting several msgs. /odometry, /real_odometry and /cloud_registered. /odometry is a stamped msg, showing the position relative to the initial position of lidar.

TIP

Mid-360 take the pose of the lidar base when initialized as the the horizon plane of its axis. This will cause some problems but somehow solved by little tricks.

We're using fast-lio for ROS2, which is not officially supported. Compared to ROS1, the map accumulating function(save as a map) is not accomplished yet.

1.3 navigation2 (ros2 navigation) ​

navigation2 github

Navigation is a huge project related to mountains of researches.

Building from source is recommended. Follow the official guide.

1.4 sentry_serial_node ​

sentry_serial_node is based on chenjun's RV framework, responsible for communicating with the lower computer.

INFO

We're controlling gimbal to navigate in RMUC2023 due to differential control method. We would change that to controlling chassis in next season.

1.5 Pointcloud to Laserscan ​

Pointcloud to laserscan is for 2d AMCL. The input is pointcloud msg /cloud_registered and the output is laser scan msg/scan.

2 Tuning and Configuration Guide ​

st=>start: Begin io_livox=>inputoutput: IMU+Lidar Msgs Input io_serial_in=>inputoutput: Serial Input target=>operation: Goal target_cond=>condition: Check if there's a Goal fast_lio=>operation: Fast_lio point_cloud_msg=>operation: Filtered Pointcloud and Odometry costmap=>operation: Local Costmap & Global Costmap,Pointcloud to LaserScan & AMCL global_path=>operation: Global Path (by navfn or hybrid A*) controller=>operation: Controller(Differential) io_serial_out=>inputoutput: Serial Output ui_position=>operation: Position In RMUC Map st(bottom)->io_livox->fast_lio->point_cloud_msg->costmap->io_serial_in->target_cond(yes)->global_path->controller->io_serial_out(left)->fast_lio(left) target_cond(no)->fast_lio(top)

2.1 FAST_LIO ​

In config/mid360.yaml,

yaml
preprocess:
            blind: 0.5
            lidar_type: 1                # 1 for Livox serials LiDAR, 2 for Velodyne LiDAR, 3 for ouster LiDAR, 
            scan_line:  4
            timestamp_unit: 3
            scan_rate: 10
preprocess:
            blind: 0.5
            lidar_type: 1                # 1 for Livox serials LiDAR, 2 for Velodyne LiDAR, 3 for ouster LiDAR, 
            scan_line:  4
            timestamp_unit: 3
            scan_rate: 10

blind is the distant from the lidar origin to the valid data sphere.

2.2 Sentry Description ​

The tf information is concluded in a urdf file. Change the configuration if you want to change the position of lidar on your robot.

2.3 Navigation ​

Official Configuration Guide

2.3.1 AMCL ​

AMCL is for relocalization, adjusting the tf between the initial pose of the odometry and

yaml
amcl:
  ros__parameters:
    alpha1: 0.2
    alpha2: 0.2
    alpha3: 0.2
    alpha4: 0.2
    alpha5: 0.2
		...
    robot_model_type: "nav2_amcl::DifferentialMotionModel"
		...
    tf_broadcast: true
amcl:
  ros__parameters:
    alpha1: 0.2
    alpha2: 0.2
    alpha3: 0.2
    alpha4: 0.2
    alpha5: 0.2
		...
    robot_model_type: "nav2_amcl::DifferentialMotionModel"
		...
    tf_broadcast: true

alphaX are the estimation of noise level of your odomtry system, each representing the translational or angular noise.Further Tuning Guide

Since we're using a really precise odomtry system, we would like to set the all 5 alphas to 0.0005 . If you don't want to use relocalization, set tf_broadcast to false, and use another node to broadcast the tf relationship between the odom and map.

2.3.2 Controller (DWB) ​

yaml
controller_server:
  ros__parameters:
    controller_frequency: 20.0
    min_x_velocity_threshold: 0.001
    min_y_velocity_threshold: 0.5
    min_theta_velocity_threshold: 0.001
    failure_tolerance: 0.3
    ...
    general_goal_checker:
      stateful: True
      plugin: "nav2_controller::SimpleGoalChecker"
      xy_goal_tolerance: 0.25
      yaw_goal_tolerance: 0.25
    # DWB parameters
    FollowPath:
      plugin: "dwb_core::DWBLocalPlanner"
      debug_trajectory_details: True
      min_vel_x: 0.0
      min_vel_y: 0.0
      max_vel_x: 0.26
      max_vel_y: 0.0
      max_vel_theta: 1.0
      min_speed_xy: 0.0
      max_speed_xy: 0.26
      min_speed_theta: 0.0
      acc_lim_x: 2.5
      acc_lim_y: 0.0
      acc_lim_theta: 3.2
      decel_lim_x: -2.5
      decel_lim_y: 0.0
      decel_lim_theta: -3.2
      vx_samples: 20
      vy_samples: 5
      vtheta_samples: 20
      sim_time: 1.7
      linear_granularity: 0.05
      angular_granularity: 0.025
      transform_tolerance: 0.2
      xy_goal_tolerance: 0.25
      ...
controller_server:
  ros__parameters:
    controller_frequency: 20.0
    min_x_velocity_threshold: 0.001
    min_y_velocity_threshold: 0.5
    min_theta_velocity_threshold: 0.001
    failure_tolerance: 0.3
    ...
    general_goal_checker:
      stateful: True
      plugin: "nav2_controller::SimpleGoalChecker"
      xy_goal_tolerance: 0.25
      yaw_goal_tolerance: 0.25
    # DWB parameters
    FollowPath:
      plugin: "dwb_core::DWBLocalPlanner"
      debug_trajectory_details: True
      min_vel_x: 0.0
      min_vel_y: 0.0
      max_vel_x: 0.26
      max_vel_y: 0.0
      max_vel_theta: 1.0
      min_speed_xy: 0.0
      max_speed_xy: 0.26
      min_speed_theta: 0.0
      acc_lim_x: 2.5
      acc_lim_y: 0.0
      acc_lim_theta: 3.2
      decel_lim_x: -2.5
      decel_lim_y: 0.0
      decel_lim_theta: -3.2
      vx_samples: 20
      vy_samples: 5
      vtheta_samples: 20
      sim_time: 1.7
      linear_granularity: 0.05
      angular_granularity: 0.025
      transform_tolerance: 0.2
      xy_goal_tolerance: 0.25
      ...

Either using differential or omnidirectional controller is determined on your parameters. Becareful not to set the yaw value to zero or the controller server would collapse. Rest of the parameters are trivial according to their names.

2.3.3 Costmap ​

yaml
local_costmap:
  local_costmap:
    ros__parameters:
      update_frequency: 5.0
      publish_frequency: 2.0
      global_frame: odom
      robot_base_frame: base_link
      rolling_window: true
      width: 3
      height: 3
      resolution: 0.05
      robot_radius: 0.22
      plugins: ["voxel_layer", "inflation_layer"]
      inflation_layer:
        plugin: "nav2_costmap_2d::InflationLayer"
        cost_scaling_factor: 3.0
        inflation_radius: 0.55
      voxel_layer:
        plugin: "nav2_costmap_2d::VoxelLayer"
        enabled: True
        publish_voxel_map: True
        origin_z: 0.0
        z_resolution: 0.05
        z_voxels: 16
        max_obstacle_height: 2.0
        mark_threshold: 0
        observation_sources: scan
        scan:
          topic: /scan
          max_obstacle_height: 2.0
          clearing: True
          marking: True
          data_type: "LaserScan"
          raytrace_max_range: 3.0
          raytrace_min_range: 0.0
          obstacle_max_range: 2.5
          obstacle_min_range: 0.0
local_costmap:
  local_costmap:
    ros__parameters:
      update_frequency: 5.0
      publish_frequency: 2.0
      global_frame: odom
      robot_base_frame: base_link
      rolling_window: true
      width: 3
      height: 3
      resolution: 0.05
      robot_radius: 0.22
      plugins: ["voxel_layer", "inflation_layer"]
      inflation_layer:
        plugin: "nav2_costmap_2d::InflationLayer"
        cost_scaling_factor: 3.0
        inflation_radius: 0.55
      voxel_layer:
        plugin: "nav2_costmap_2d::VoxelLayer"
        enabled: True
        publish_voxel_map: True
        origin_z: 0.0
        z_resolution: 0.05
        z_voxels: 16
        max_obstacle_height: 2.0
        mark_threshold: 0
        observation_sources: scan
        scan:
          topic: /scan
          max_obstacle_height: 2.0
          clearing: True
          marking: True
          data_type: "LaserScan"
          raytrace_max_range: 3.0
          raytrace_min_range: 0.0
          obstacle_max_range: 2.5
          obstacle_min_range: 0.0

robot_radius can be changed to footprint for square robot chassis. Checkt this for further information.

raytrace_range should conclude the obstacle_range in order to rapidly remove the trace of the dynamic obstacle.

Because we're using pointcloud to get costmap, change the datatype and topic under scan.

global_costmap is same as the local_costmap, presenting information for path calculation.

3 Map ​

3.1 Make and Save a Map ​

Save a 3D map and project it to make a 2D map

In fast-lio, change the save map parameter from false to true, you will get a 3D pcd file. Then, using pcd2pgm link you can get a 2D map you like.

Save a 2D map using /scan and gmapping

Using gmapping and our /scan topic, you can save a map using map_saver.

3.2 The Projection from Actual Map and Referee System ​

The manipulator of the sentry will pass a point on the Referee System which is different from position of the real terrain. I used Least Square to calculate the linear projection.

PDF.