Robotics Second Project - Mapping and Autonomous Navigation

This repository contains the source code for the generation of an environment map by performing SLAM, and of robot navigation using the DWA local planner.

Code Organization

• CMakeList.txt this file contains a description of the project's source files and targets
• info.txt this file contrains the registration number, name, last name of the authors
• package.xml this file contains a collection of metadata components
• launch/* this folder contains the project launch files
• maps_raw/* this folder contains the .pgm and .yaml files of the generated maps
• src/navigation.cpp this executable instantiates the navigation node, reads the new goal from the waypoints.csv file (defined in the /navigation/csv_path parameter) and publishes it via move base action
• src/tf_publisher.cpp this executable instantiates the tf_publisher node and publishes the tf between odom and t265 frames based on the t265 odometry
• stage/* this folder contains the .pgm, .yaml, .png and .world files of the final chosen map
• config_mapping.rviz this file contains the rviz configuration parameters for the generation of the map
• config_nav.rviz this file contains the rviz configuration parameters for the robot navigation
• waypoints.csv this file contains the waypoints read by the navigation node to set the new goal
• waypoints.png this file contains an image of the map chosen for navigation, with the pose and sequence number of the selected waypoints

Installation

Prerequisites:

• Linux environment
• ROS noetic

Generating the Map:

Launchfile for generating the map with the provided single plane scanner data (from /scan topic):

roslaunch second_project slam_toolbox_scan.launch

Launchfile for generating the map with the 2D data converted from the 3D laser (includes a pointcloud_to_laserscan node)

roslaunch second_project slam_toolbox_2d_scan.launch

After launching either one of these, a rosbag should be played in order to generate the map.

Both launchfiles were run with two separate bags, and the resulting maps were stored in the maps_raw folder. These commands also launch the tf_publisher node together with other static transforms, and RViz with the config_mapping.rviz configuration.

After the map was generated, it was saved by running:

rosrun map_server map_saver -f <map_name>

Running the Navigation:

The map chosen for navigation was the map generated from the converted 2D scanner by playing the first bag; noise was removed in post-processing.

The algorithm chosen for local path planning is DWA.

roslaunch second_project amcl.launch

This also launches the navigation node and RViz with the config_navigation.rviz configuration.

The navigation node reads a sequence of waypoints, one at a time, from the waypoints.csv file; each waypoint represents the new goal that the robot attempts to reach.

This is the map with the chosen waypoints:

Nodes and Other Packages Used:

• pointcloud_to_laserscan to convert velodyne pointcloud to laserscan
• slam_toolbox with gmapping to generate the map (detailed config in cfg/st_config_scan.yaml)
• dwa_local_planner for local planning
• tf and tf2_ros for static transforms
• rviz for visualization
• stage_ros for stage simulation
• move_base for the navigation stack
• amcl for localization
• map_server for saving/serving the map
• amcl for autonomous navigation