MCTR

<table> <tr> <td align="center" valign="bottom"> <img src="./first_vision.gif" height="300" /> <br /> _{first_vision} </td> <td align="center" valign="bottom"> <img src="./third_vision.gif" height="300" /> <br /> _{third_vision} </td> <td align="center" valign="bottom"> <img src="./digital_vision.gif" height="300" /> <br /> _{digital_vision} </td> </tr> </table>

More details please see ：

bilibili:https://www.bilibili.com/video/BV1LdqWBXEBo/?vd_source=3484564f7d7d8958edef5897d0494ad6

youtube:

<div id="Main"></div>

Main project

This project consists of two main modules:

<nav> <ul> <li><a href="#module1">F1tenth Simulator && Physical Controller</a></li> <li><a href="#module2">Carla Simulator Controller</a></li> <li><a href="#module3">Create Digital Twin Map</a></li> <li><a href="#module4">Custom Vehicle Import</a></li> </ul> </nav>

Project Struct

  project_mctr/
  │
  ├──controller/                      
  │   ├── controller_manager.py        
  │   ├── mctr/                        
  ├──requirements.txt   #For carla                 
  ├──launch_carla.sh                 
  ├──carla-ros-bridge/
     └── carla_ws/
         └── src/
             └── carla_stack/  
  ├──rcarla_example/
  │    ├── digital_twin.ply 
  │    ├── map.obj                
  │    └── tmp.ply                   

Prerequisites

To run the MCTR race stack make sure that you have assembled and setup a car following the official F1TENTH instructions.

Please install strictly in the following order:

• Ubuntu 20.04: Double System or WSL with Windows 11
• ROS Noetic
• Catkin Build Tools (Installing on Ubuntu with apt-get)
• FORZA_ETH RaceStack
• CCMA

Note:

Add the following aliases to your ~/.bashrc for convenience:

alias sros1="source /opt/ros/noetic/setup.bash"
alias sf1="sros1; source ~/catkin_ws/devel/setup.bash" # for convenience, no need>
alias mctr="sf1"

Now, you can simply type mctr in the terminal to activate the environment.

<!-- F1tenth Simulator && Physical Controller --> <h2 id="module1"></h2> <a href="#Main">

F1tenth Simulator && Physical Controller</a>

How to use F1tenth Simulator

MCTR race stack runs on the Forza ETH stack. Merge or replace folders controller with our files.

Ensure it subscribes to the following ROS topics:

| Topic Name | Message Type | Description | | ----------------------------------- | ---------------------------- | ------------------------------------------------------------- | | /scan | sensor_msgs/LaserScan | LiDAR scan data for environment and obstacle perception | | /car_state/odom | nav_msgs/Odometry | Current vehicle velocity | | /car_state/pose | geometry_msgs/PoseStamped | Vehicle pose in map frame (position + orientation) | | /car_state/odom_frenet | nav_msgs/Odometry | Vehicle position in Frenet frame (currently unused) | | /vesc/sensors/imu/raw | sensor_msgs/Imu | Raw IMU acceleration data (for steering adjustment) | | /local_waypoints | f110_msgs/WpntArray | Locally planned waypoints (short-term trajectory) | | /global_waypoints | f110_msgs/WpntArray | Global trajectory for initialization | | /perception/obstacles | f110_msgs/ObstacleArray | Detected obstacles (position, velocity, static/dynamic) | | /state_machine | std_msgs/String | Current vehicle control mode (e.g., FTGONLY, TRAILING) | | /l1_param_tuner/parameter_updates | dynamic_reconfigure/Config | Real-time parameter tuning (e.g., L1 controller PID settings) |

After setup, you can launch the base system and our algorithm in the F1tenth Simulator with:

launch stack_master base_system.launch sim:=true map_name:=test_map

Another terminal:

roslaunch stack_master time_trials.launch ctrl_algo:=MCTR racecar_version:=NUC2

If you have a F1tenth car, choose the sim to false:

roslaunch stack_master base_system.launch sim:=false map_name:=test_map

<!--Carla Simulator Controller--> <h2 id="module2"></h2> <a href="#Main">

Carla Simulator Controller</a>

Environment Setup

# 1. Create a virtual environment (recommended: Python 2.7.18)
python2.7 -m venv carla_env

# 2. Activate the environment
source carla_env/bin/activate

# 3. Upgrade pip
pip install --upgrade pip

# 4. Install dependencies
pip install -r requirements.txt

Add this alias to your ~/.bashrc to quickly source the Carla environment:

alias carla='source ~/carla-ros-bridge/catkin_ws/devel/setup.bash'

Now, you can simply type carla in the terminal to activate the environment.

Pre-Installation

Install CARLA and CARLA-ROS-BRIDGE:

• CARLA
• CARLA-ROS-BRIDGE

For your convenience, you can directly use our integrated Carla-ros-bridge.

carla-ros-bridge/
  └── carla_ws/
      └── src/
          └── carla_stack/
              ├── carla_spawn_objects/   
              ├── carla_controller/ 
              └── spawn_f1tenth/             

If you are using the official carla-ros-bridge, please replace or add our package to the location below.

carla-ros-bridge/
  └── catkin_ws/
      └── src/
          ├── vision_opencv/
          └── ros-bridge/
              ├── carla_spawn_objects/   
              ├── carla_controller/ 
              └── spawn_f1tenth/             

How to use Carla Simulator

All startup commands are designed as scripts to facilitate program startup. You can run the script launch_carla.sh anywhere (recommended in the carla folder). Start carla:

./launch_carla.sh

Click Play in the Carla after launching.

Start CARLA-ROS-Bridge (in a new terminal):

# Source Carla environment
carla

# Launch ROS bridge with ego vehicle
roslaunch carla_ros_bridge carla_ros_bridge_with_example_ego_vehicle.launch

Launch the MCTR 3D Algorithm (in another terminal):

roslaunch carla_controller controller_manager.launch ctrl_algo:=DELAUNAY_FTG

<!-- Digital Twin Map Creation--> <h2 id="module3"></h2><a href="#Main">

Digital Twin Map Creation</a>

Pre-Installation

We used rcarla, developed by ETH, to build the digital twin mesh in Carla simulator:

• Install rcarla

Obtain Point Cloud Data

Use 2D/3D LiDAR to capture point cloud data (.pcd/.ply files). Example track point cloud data digital_twin.ply is provided for mesh generation.

You need to use a point cloud processing tool (such as CloudCompare) to delete point clouds that are irrelevant to the track and obtain a pre-processed point cloud file.

Please place the point cloud files in the rcarla/digital_twin_creator/data folder.

Generate Mesh Files

cd rcarla/digital_twin_creator
./setup.sh
source venv/bin/activate
python3 digital_twin_creator.py data/<YOUR_POINTCLOUD_FILE>

This will generate tmp.ply and map.obj files. You can rename them as needed. Example track point cloud data tmp.ply and map.obj generated by digital_twin.ply are provided.

Create Digital Twin Map

Start Carla:

cd ~/carla
./launch_carla.sh

Import the .obj file in UE4, create a new level to build the map, or open example maps like Test_map or Digital_Twin_map:

cp PATH/TO/Digital_Twin_map/Carla/Maps/* PATH/TO/carla/Unreal/CarlaUE4/Content/Carla/Maps

Launch New Map:

Modify the carla_ros_bridge_with_example_ego_vehicle.launch file in /carla-ros-bridge/catkin_ws/src/ros-bridge/carla_ros_bridge/launch:

<arg name="town" default='<YOUR_MAP_NAME>'/>

Modify vehicle spawn point:

<arg name="spawn_point" default="<X,Y,Z,ROLL,PITCH,YAW>"/>

Click Play in the UE4 window, then in a new terminal, launch the vehicle:

cd ~/carla-ros-bridge/catkin_ws
./launch_ros_bridge_with_ego.sh

<!-- Custom Vehicle Import--> <h2 id="module4"></h2><a href="#Main">

Custom Vehicle Import</a>

Vehicle Specifications:

Wheelbase: 37cm Axle track: 50cm <br> CARLA Vehicle Asset (Length: 77cm, Width: 36cm) Usage

Import Vehicle Assets

1. Open CARLA.
2. Create a new folder in the Content Browser.
3. Right-click the folder and select Open in File Explorer.

Configure Wheel Assets

1. In the Content Browser, navigate to Content/Carla/Blueprints/Vehicles.
2. Place the Wheel_shape_small.uasset in this folder.

Register Vehicle in VehicleFactory

1. In Content/Carla/Blueprints/Vehicles, double-click VehicleFactory.
2. Click on Vehicle.
3. In the right panel, click the plus icon to add a new element.
4. Fill in the configuration as shown in the documentation.

Spawn Custom Vehicle

• When spawning vehicles, use the vehicle name rgv.