race_car

This project provides a simple solution for a racecar navigation. The simulation environment is adapted from eufs_sim

Demo

Demo of racecar navigating in simulation environment by only using image data

Demo of racecar passing a small U-circle

Install Prerequisites

This project is created for ROS-Kinetic / ROS-Melodic, where python 2.7 is used. However, part of this project's codes are only available for python 3. Several steps for using python 3 along with python 2.7 is listed:

• Ubuntu 16.04 + ROS Kinetic

sudo apt-get install ros-kinetic-ackermann-msgs ros-kinetic-twist-mux ros-kinetic-joy ros-kinetic-controller-manager ros-kinetic-robotnik-msgs ros-kinetic-velodyne-simulator ros-kinetic-effort-controllers ros-kinetic-velocity-controllers ros-kinetic-joint-state-controller ros-kinetic-gazebo-ros-control ros-kinetic-robotnik-msgs

• Ubuntu 18.04 + ROS Melodic

sudo apt-get install ros-melodic-ackermann-msgs ros-melodic-twist-mux ros-melodic-joy ros-melodic-controller-manager ros-melodic-robotnik-msgs ros-melodic-velodyne-simulator ros-melodic-effort-controllers ros-melodic-velocity-controllers ros-melodic-joint-state-controller ros-melodic-gazebo-ros-control

Install this project

• Download and Compile

  Create a ROS workspace if you don't have one (here, the workspace is called race_car_ws): mkdir -p ~/race_car_ws/src. Copy the contents in this repository into the src file you just created.

  Open a terminal, navigate to your workspace and build the simulation:

  cd ~/race_car_ws/src
  catkin_make
  

  Navigate inside your workspace and give permission to several .py or .sh files:

  cd ~/race_car_ws/src/auto_pilot_param/scripts
  chmod +x naive_controller.py
  
  cd ~/race_car_ws/src/eufs_gazebo/nodes
  chmod +x ground_truth_republisher.py
  
  cd ~/race_car/src/sim_simple_controller/nodes
  chmod +x twist_to_ackermannDrive.py
  
  cd ~/race_car_ws/src/scripts
  chmod +x install_python2.sh
  chmod +x install_python3.sh
  

• Enable gazebo display

  Put the 3 model packages inside models in this repository into .gazebo/models.

• Enable python3 usage

  This project is created for ROS-Kinetic / ROS-Melodic, where python 2.7 is used. However, part of this project's codes are only available for python 3. Several steps for using python 3 along with python 2.7 is listed:

  • eliminate any source commands in your ~/.bashrc

    source commands, for example: source /opt/ros/melodic/setup.bash will include python 2.7 directory, and will conflict with python 3. Please delete all source commands in "~/.bashrc", and put them in a new file called ~/.bashrc_ros1.

    Remember to add source command of this workspace into ~/.bashrc_ros1

    source /home/username/race_car_ws/devel/setup.bash

  • (optional) create a conda environment for python 3

    It is recommended to create a python 3 environment with conda. To install conda, follow the conda installation guide

    create a new conda environment called racecar

    conda create --name racecar python=3.8

  • Install all python 3 packages

    Open a new terminal, activate your python 3 environment (e.g. conda activate racecar)

    Navigate to /scripts inside your workspace

    cd ~/race_car_ws/src/scripts

    Install those packages by

    ./install_python3.sh

  • Install python 2 packages

    Open a new terminal, activate python 2 environment

    Navigate to /scripts inside your workspace

    cd ~/race_car_ws/src/scripts

    Install those packages by

    ./install_python2.sh

• Download weights for cone-detection by yolov3

  Cone detection using yolov3 is an important part for the racecar to navigate in the simulation environment. The model weights for cone-detection in simulation can be found at link_google_drive or 微云 Please download the weights from any of the links and put it into

  ~/race_car_ws/src/scripts/detect

Run demo files

• step 1

  Open the first terminal, activate ROS workspaces by using

  source ~/.bashrc_ros1

  In the terminal, launch simulation by

  roslaunch eufs_gazebo training_track.launch

  you can also change the launch file in the command above to big_track.launch

• step 2

  Open another terminal, activate the python 3 environment, for example

  conda activate racecar

  navigate to /scripts/detect in your workspace

  cd ~/race_car_ws/src/scripts/detect/

  start the server for naive racecar control

  python cone_detection_server_param.py

  wait until you see "Control server started"

• step 3

  Open the third terminal, activate ROS workspaces by using

  source ~/.bashrc_ros1

  In the terminal, run the ros node for naive racecar control by

  rosrun auto_pilot_param naive_controller.py

  Once you run this command, the racecar inside simulation should start running.

Project Introduction

To learn more about this project, please see racecar_project_introduction.pdf