Simulation

Repository containing different projects involving any kind of robotic simulation

Table of contents

• Project 1
• Project 2 (Chase White pixel)
• Project 3 (MCL localization)
• Project 4 (RTAB Map)
• Project 5 (Home Service)

Project 1

Spawning a simple three wheeled robot in a simulation environment.

Build and compile

• Compiling the code in scripts directory is necessary to make the plugin work.
• Make a directory names as build, mkdir build
• Build directory can be located on the same level as CMakeLists.txt
• After making a build directory, open it cd build, then use the following commands to compile the code

$ cmake ../
$ make

Post building and compilation, you need to set a path for gazebo to know where to check your plugin too.

$ export GAZEBO_PLUGIN_PATH=${GAZEBO_PLUGIN_PATH}:/home/path_to_build_directory/build

Run

• Use the command gazebo worldSubmit1 to run the simulation

Project 2

Chasing a white colored ball in gazebo simulation by using camera feed.

Setup and compile

• First the two packages my_robot and ball_chaser are to be made in src folder of your catkin workspace.
• Then use catkin_make in one level before your src directory of catkin workspace to build all the executables according to instructions in CMakeLists.txt files.

Run

• Use the commands roslaunch my_robot world.launch and roslaunch ball_chaser ball_chasr.launch to run the simulation and move around the ball to test.

NOTE:

• Play around with torque and velocity values to get smooth turning for different wheel joint coordinates of robot.

Project 3

Observing behaviour of MCL in RViz while tweaking different parameters in AMCL package.

• Particles after initial pose estimation

• Particles after 1 update

• Particles after 6 updates

• Particles after 30-40 updates

Project 4

Making a map using Real-Time Apearance Based Mapping (RTAB Map).

Link to database files(result + exported 3D Map).

• The yellow dots (in the cube figure) are showing distinct features. (below occupancy grip)

• Figures shows the resultant map in RViz in a slanted side view.

• Figures shows the resultant map in RViz in top view.

Project 5

Writeup:-

• In this project we used all the previous knowledge of all previous courses to make a robot capable of mapping and localizing simultaneously (SLAM) in order to pick up virtual objects and drop them off at a given position.
• Official repos of turtlebot3 were used to obtain this functionality along with "self made" nodes to pick and place virtual objects.
• Changes have been done to official repos of turtlebot3 and turtlebot3_simulations in order to run mapping and localization on our world and later on our map (pick and place)
• For localization, amcl package was used (located in turtlebot3/turtlebot3_navigations)and mapping was done via slam_gmapping package
• For navigation, ROS Navigation stack was used.
• The ROS navigation stack creates a path for our robot based on Dijkstra's algorithm, a variant of the Uniform Cost Search algorithm, while avoiding obstacles on its path.
• Shell scripts have been made to run different nodes via xterm.
• In order to visualize, I did not run any additional command via xterm in scripts because the launch files of turtlebot3 already included those, although the path of RViz configuration file in turtlebot3_navigation was changed in order to run our RViz configuration.
• add_markers package contains the add_markers node which adds virtual objects in rviz.
• pick_objects package contains the pick_objects node which sends two goals (pose) for robot to travel to by path planning and it is synced with add_markers node to add and delete package accordingly to make it look like robot picked up the virtual object.

NOTE:-

• In case you are getting error related to dependencies, try to use rosdep

• Do not use original turtlebot packages, the one in repository are modified and must be used only.

• Figures shows the robot going to pick virtual object.

• Figures shows the robot going to drop after picking up.