💻 mcav_autonomy

The urban-driving autonomous stack of the Monash Connected Autonomous Vehicles team. Designed to run on a StreetDrone Twizy and in the CARLA simulation environment.

Installation - Using Docker (recommended)

Requirements

• Docker (install on Ubuntu using sudo apt install docker.io)
• Nvidia-docker2 (For optional GPU support)

Installation

• Clone the repository:

git clone git@github.com:Monash-Connected-Autonomous-Vehicle/mcav_autonomy.git

• Run the container:

cd mcav_autonomy
docker/run.sh

• See docker/README.md for more details on using docker in this project.

Installation - From Source

Requirements

• Ubuntu 22.04
• ROS2 Humble Hawksbill

Installation

• Create a workspace: mkdir -p ~/mcav_ws/src && cd ~/mcav_ws/src
• Install vcstool: sudo apt install python3-vcstool or pip3 install vcstool
• Clone this repo: git clone git@github.com:Monash-Connected-Autonomous-Vehicle/mcav_autonomy.git
• Change directory: cd mcav_autonomy
• Create a space for source dependencies: mkdir external/
• Clone source dependencies: vcs import external/ < ros.repos
• Change directory: cd ~/mcav_ws
• Install ROS dependencies: rosdep install --from-paths src -i -r -y
• Build the packages: colcon build --symlink-install

Project Structure

This should result in a directory structure similar to the following:

mcav_ws/                                                     
├── build
├── install
└── src
    └── mcav_autonomy
        ├── autonomy_launch         # Launch files for entire stack
        ├── data
        │   ├── pointclouds         # Pointcloud map files (.pcd)
        │   ├── rosbags             # Recorded rosbags
        │   └── waypoints           # Global waypoints (.csv) 
        ├── data_recording          # Launch files for creating rosbags
        ├── docker                  # Dockerfiles and run scripts
        ├── external                # External dependencies that come in source code form
        │   ├── SD-VehicleInterface # Sends actuation commands, reads GPS/IMU and speedometer
        │   └── kiss-icp            # Performs lidar-based odometry (see github.com/PRBonn/kiss-icp)
        ├── mcav_interfaces         # Shared ROS Message and Service definitions
        ├── project_tracker         # Object detection and tracking
        ├── pure_pursuit            # Control system
        ├── sensors_launch          # Launch files for the sensors
        ├── simulation              # CARLA & simple kinematics simulation setups
        ├── velocity_planner        # Local planner
        └── visualisation
            └── vehicle_model       # 3D vehicle models for visualisation 

Pointcloud maps, rosbags, waypoints and other files that are data, not code, should be stored in data/pointclouds, data/rosbags, data/waypoints etc.

External Dependencies

This project relies on:

• MCAV's fork of the StreetDrone Vehicle Interface for communication with the vehicle actuation
• KISS-ICP for lidar-based odometry

Running

How to run in CARLA

Terminal 1 (Launch CARLA server):

• /opt/carla-simulator/CarlaUE4.sh

Terminal 2 (Launch CARLA ros-bridge):

• cd ~/mcav_ws && source install/setup.bash
• ros2 launch carla_ros_bridge carla_ros_bridge.launch.py

Terminal 3 (Launch autonomy stack):

• cd ~/mcav_ws && source install/setup.bash
• ros2 launch autonomy_launch carla.launch.py waypoints_file:=/home/mcav/Sheng/control_ws/town01_path1.csv

How to run on Alienware

Setup CAN and enter Docker container:

./main_launch.sh

Launch the stack:

src
ros2 launch autonomy_launch streetdrone.launch.xml