Skip to content

Latest commit

 

History

History
105 lines (80 loc) · 3.39 KB

README.md

File metadata and controls

105 lines (80 loc) · 3.39 KB
Raw

Path Planning Stack

Main scripts

In the folder scripts, you can find:

  • sim_launcher_navigation for launching a path planning simulation with all the required nodes (mapping, traversability analysis, trajectory control)
  • sim_launcher_ugv for launching the nodes of a single robot (mapping, path planner, trajectory control): this script is used by sim_launcher_navigation
  • save_map for saving the robot map and trajectory (please, use ugv1 for building and saving maps or trajectories)
  • kill_vrep_sim or kill_gazebo_sim for killing all the launched nodes and V-REP/gazebo.

How to run a path planning simulation

Open a new terminal, enter in the root folder of the repo, and run:
$ source source_all.bash
$ rosrun path_planner sim_launcher_navigation

or you can also run:
$ source source_all.bash
$ roscd path_planner/scripts
$./sim_launcher_navigation

N.B.: in the script sim_launcher_navigation, you can find some input variables for setting different things, e.g. you can change the V-REP world and the number of robots (see the variables WORLD_NAME and NUM_ROBOTS).

In order to kill all the launched nodes and V-REP, run:
$ rosrun path_planner kill_vrep_sim
or $ ./kill_vrep_sim (from path_planner/scripts)

Information about the nodes

Node: traversability

Description: provides traversability analysis

Subscribed topics: /dynjoinpcl: 3d map

Published topics: /clustered_pcl/wall: subset of map relative to non-traversable parts (walls, obstacles, ...) /clustered_pcl/no_wall: subset of map relative to traversable parts (ground, ramp, stairs) /clustered_pcl/segmented: ? /normals_pcl: point cloud normals /normals_markers: point cloud normals marker /trav/label: points labeled with semantic information (a priori cost based on the type of traversable part) /trav/density: points labeled with local point cloud density /trav/roughness: points labeled with local point cloud roughness /trav/clearance: points labeled with information about distance to closest obstacle /trav/traversability: costmap summarizing the 4 cost components (label, density, roughness, clearance)

Parameters: ClusteringPcl: max_superable_height: 0.2 (maximum height of a surmountable obstacle) normal_clustering_thres: 0.1 (?) normal_radius_search: 0.3 (normal estimation radius search) pcl_cluster_tolerance: 0.3 (?)

Node: pathPlanner

Description: path planning facility

Subscribed topics: /trav/traversability: traversability costmap /clustered_pcl/wall: (don't know why this guy is here) /goal_topic: geometric goal location

Published topics: /robot_path: geometric path towards goal (if any)

Node: mapping

Description: aggregates point clouds from laser scanner and remove outdated dynamic obstacles from map

Subscribed topics: /dynamic_point_cloud: 3d point cloud from laser scanner

Published topics: /dynjoinpcl: aggregated point cloud /normals_marker: debug normals estimation

Parameters: DynamicJoinPcl: global_frame: /map laser_frame: /laser leaf_size: 0.035 (downsampling leaf size) num_subdivisions: 50 (number of subdivisions. higher number -> higher accuracy. beware: too much high number -> out of memory) NormalEstimationPcl: flatness_curvature_threshold: 0.2 kernel_type: 0 laser_frame: /laser num_threads: 4 radius: 0.2 smoothing: 1.0