Autonomous Tunnel Driving package

This package is for running mobile robot under tunnel environment.

The code is tested with Leo Rover mobile platform and Rover Robotics Rover Pro

Used Sensors are

• 45-degree tilted 2D Lidar: for aisle-follow driving
• 3D LiDAR on mobile robot: for detecting obstacles, mapping, localization

Consisting Nodes are

• Aisle detect Node

  • Subscribe
    • /rp/scan (sensor_msgs/LaserScan)
  • Publish:
    • /cluster_line (sensor_msgs/PointCloud2): detected aisle
    • /aisle_points (sensor_msgs/PointCloud2): robot's direction on line, line center, line start, line end
  • Parameters
    • line_thresh (defalt 0.5): Ransac threshold of line model
    • aisle_width (defatl 0.6): Horizontal width to be cropped from lidar scan

• Obstacles Node

  • Subscribe
    • /velodyne_points (sensor_msgs/PointCloud2)
  • Publish:
    • /cropped_obs (sensor_msgs/PointCloud2): detected obstacles within the aisle zone
    • /obs_dists (std_msgs/Float32MultiArray): Right and Left obstacles's distances
  • Parameters
    • obs_x_min / obs_x_max (default 0.0 / 1.0): x range of obstacle detect region (m)
    • obs_y_min / obs_y_max (default -0.3 / 0.3): y range of obstacle detect region (m)
    • obs_z_min / obs_z_max (default -0.1 / 0.5): z range of obstacle detect region (m)

• Localization Node

  • Subscribe
    • /amcl_pose (geometry_msgs/PoseWithCovarianceStamped): amcl_pose topic can be published by running amcl package simultaneously.
    • /move_base_simple/goal (geometry_msgs/PoseStamped): The target position can be set with 2D Nav Goal of Rviz.
  • Publish:
    • /localization_data (std_msgs/Float32MultiArray): [0]-global dist error, [1]-global angle error, [2]-arrival flag, [3]-rotating flag
  • Parameters
    • global_dist_boundary (default 0.3): Global distance boundary error to arrival (m)
    • global_ang_boundary (default 0.05): Global angular boundary error to arrival (rad)

• Cmd Puslish Node

  • Subscribe (data from other nodes)
    • /joystick (sensor_msgs::Joy): amcl_pose topic can be published by running amcl package simultaneously.
    • /obs_dists (std_msgs/Float32MultiArray)
    • /aisle_points (sensor_msgs/PointCloud2)
    • /localization_data (std_msgs/Float32MultiArray)
    • /lidar_driving (std_msgs/Bool) : Should be puslished manually to execute the /cmd_vel publishing function.
  • Publish:
    • /cmd_vel (geometry_msgs/Twist): Final cmd_vel combining information from multiple nodes
  • Parameters
    • Kpy_param (default 1.1): Rotation control gain for aisle driving
    • Kpy_param_rot (default 0.01): Rotation control gain for static rotation
    • linear_vel (default 0.0): Linear velocity of mobile robot (m/s)
    • robot_width (default 0.45): Robot width (m)
    • obs_coefficient (default 0.5): Obstacle avoidance control ratio
    • front_obs(default 0.6): Distance of front obstacles to be avoided
    • boundary_percent (default 0.02): Percentage of free width on both sides of the aisle for safe driving
    • spare_length (default 1.5): Additional driving after avoiding obstacles
    • amcl_driving (default false): Whether to check the location of mobile robot
    • check_obstacles (default false): Whether to check for obstacles

Required ROS Package

• slam-gmapping
• amcl

sudo apt-get install ros-melodic-slam-gmapping
sudo apt-get install ros-melodic-amcl

Launch files

• only_driving.launch : driving using scan data with no mapping and localization
• mapping.launch : Generates map with autonomous driving (gmapping ros package is included)
• amcl_driving.launch : Round trip driving with localization (amcl ros package is included)
• simulation.launch : run with bag file

Running

roslaunch leo_driving only_driving.launch
roslaunch leo_driving mapping.launch
roslaunch leo_driving amcl_driving.launch
roslaunch leo_driving simulation.launch