CARLA Autonomous Driving Leaderboard

https://leaderboard.carla.org
Admin
The CARLA Autonomous Driving Leaderboard is an open platform designed to evaluate the driving...
Tokens:19,357
Snippets:176
Trust Score:9.9
Update:11 months ago
Show doc for...
Tokens:
Raw
### Install Leaderboard Python Dependencies

Source: https://leaderboard.carla.org/get_started_v1_0

Installs the Python dependencies required for the Leaderboard. Execute this command after navigating to the Leaderboard root directory.

```shell
cd ${LEADERBOARD_ROOT} # Change ${LEADERBOARD_ROOT} for your Leaderboard root folder
pip3 install -r requirements.txt
```

--------------------------------

### Install Scenario Runner Python Dependencies

Source: https://leaderboard.carla.org/get_started_v1_0

Installs the Python dependencies required for the Scenario Runner. Execute this command after navigating to the Scenario Runner root directory.

```shell
cd ${SCENARIO_RUNNER_ROOT} # Change ${SCENARIO_RUNNER_ROOT} for your Scenario_Runner root folder
pip3 install -r requirements.txt
```

--------------------------------

### Clone Leaderboard Repository

Source: https://leaderboard.carla.org/get_started_v1_0

Clones the Leaderboard v1.0 repository from GitHub. This command should be executed in your desired project directory.

```shell
git clone -b leaderboard-1.0 --single-branch https://github.com/carla-simulator/leaderboard.git
```

--------------------------------

### Python: Override Agent Setup Method

Source: https://leaderboard.carla.org/get_started_v1_0

The `setup` method is used for agent initialization and definitions. It's called when a route is initialized and can accept a configuration file path. The minimum requirement is to set the agent's track modality using the `Track` enum.

```Python
from leaderboard.autoagents.autonomous_agent import Track
...
def setup(self, path_to_conf_file):
    self.track = Track.SENSORS # At a minimum, this method sets the Leaderboard modality. In this case, SENSORS

```

--------------------------------

### Run CARLA Server

Source: https://leaderboard.carla.org/get_started_v1_0

Starts the CARLA simulator server with specified quality level, port, and resolution. This command should be executed in a separate terminal.

```shell
cd ${CARLA_ROOT}
./CarlaUE4.sh -quality-level=Epic -world-port=2000 -resx=800 -resy=600
```

--------------------------------

### Run Carla Leaderboard Test Script

Source: https://leaderboard.carla.org/get_started_v1_0

This command runs a test script that launches a pygame window for manual agent control. It cycles through multiple towns, loading a single test route in each, and guides the user to follow the route and traffic rules.

```shell
./test_run.sh
```

--------------------------------

### Install CARLA Python API Requirements

Source: https://leaderboard.carla.org/get_started_v1_0

Installs the necessary Python dependencies for the CARLA Python API. This command should be executed within your activated conda environment after navigating to the CARLA root directory.

```shell
conda create -n py37 python=3.7
conda activate py37
cd ${CARLA_ROOT}  # Change ${CARLA_ROOT} for your CARLA root folder

pip3 install -r PythonAPI/carla/requirements.txt
```

--------------------------------

### Create and Prepare Leaderboard Run Script

Source: https://leaderboard.carla.org/get_started_v1_0

Creates an executable bash script named 'test_run.sh' in the Leaderboard root directory. This script will be used to configure and launch the leaderboard evaluation.

```shell
touch test_run.sh
chmod +x test_run.sh
```

--------------------------------

### Override Agent Setup Method

Source: https://leaderboard.carla.org/get_started_v2_0

Overrides the `setup` method within an agent class to perform initialization tasks, such as setting the participation track. This method is called automatically when a route starts and can accept a configuration file path. It requires importing `Track` enum from `leaderboard.autoagents.autonomous_agent`.

```python
from leaderboard.autoagents.autonomous_agent import Track
...
def setup(self, path_to_conf_file):
    self.track = Track.SENSORS # At a minimum, this method sets the Leaderboard modality. In this case, SENSORS
```

--------------------------------

### Clone Scenario Runner Repository

Source: https://leaderboard.carla.org/get_started_v1_0

Clones the Scenario Runner repository from GitHub. This repository is used for managing and executing scenarios for the Leaderboard.

```shell
git clone -b leaderboard --single-branch https://github.com/carla-simulator/scenario_runner.git
```

--------------------------------

### Create ROS Agent Entry Point

Source: https://leaderboard.carla.org/get_started_v2_1

Defines the entry point for a ROS-based agent. It includes specifying the ROS package, launch file, and parameters required for ROS integration.

```python
from leaderboard.autoagents.ros1_agent import ROS1Agent

def get_entry_point():
    return 'MyROSAgent'


class MyRosAgent(ROS1Agent):
    ....
    def get_ros_entry_point(self):
        return {
            "package": "my_ros_agent",
            "launch_file": "my_ros_agent.launch",
            "parameters": {}
        }

```

--------------------------------

### Source Bash Profile

Source: https://leaderboard.carla.org/get_started_v1_0

Applies the environment variable changes made to the ~/.bashrc file to the current terminal session.

```shell
source ~/.bashrc
```

--------------------------------

### Clone Scenario Runner Repository

Source: https://leaderboard.carla.org/get_started_v2_1

Clones the CARLA Scenario Runner repository from GitHub, targeting the 'leaderboard-2.1' branch. This component is necessary for running simulation scenarios.

```bash
git clone -b leaderboard-2.1 --single-branch https://github.com/carla-simulator/scenario_runner.git
```

--------------------------------

### Clone CARLA Leaderboard Repository

Source: https://leaderboard.carla.org/get_started_v2_1

Clones the CARLA leaderboard repository from GitHub, specifically checking out the 'leaderboard-2.1' branch. This is essential for obtaining the leaderboard code.

```bash
git clone -b leaderboard-2.1 --single-branch https://github.com/carla-simulator/leaderboard.git
```

--------------------------------

### Clone CARLA Leaderboard Repository

Source: https://leaderboard.carla.org/get_started_v2_0

Clones the CARLA Leaderboard repository from GitHub. It specifically checks out the 'leaderboard-2.0' branch to ensure compatibility with the described version. This command should be run in the directory where you want to place the leaderboard code.

```bash
git clone -b leaderboard-2.0 --single-branch https://github.com/carla-simulator/leaderboard.git
```

--------------------------------

### Clone CARLA ROS Bridge

Source: https://leaderboard.carla.org/get_started_v2_1

Clones the CARLA ROS bridge repository, ensuring the 'leaderboard-2.1' branch is selected. This is for setting up ROS-based autonomous agents.

```bash
git clone --recurse-submodules -b leaderboard-2.1 --single-branch https://github.com/carla-simulator/ros-bridge
```

--------------------------------

### Create ROS Agent Entry Point

Source: https://leaderboard.carla.org/get_started_v2_0

Creates a ROS-specific entry point for agents, specifying ROS package, launch file, and parameters. This builds upon the standard entry point by adding ROS integration details. It requires importing `ROS1Agent`.

```python
from leaderboard.autoagents.ros1_agent import ROS1Agent

def get_entry_point():
    return 'MyROSAgent'


class MyRosAgent(ROS1Agent):
    ....
    def get_ros_entry_point(self):
        return {
            "package": "my_ros_agent",
            "launch_file": "my_ros_agent.launch",
            "parameters": {}
        }
```

--------------------------------

### Clone Scenario Runner Repository

Source: https://leaderboard.carla.org/get_started_v2_0

Clones the CARLA Scenario Runner repository from GitHub. Similar to the Leaderboard, it ensures the 'leaderboard-2.0' branch is used for consistency. Run this command in the desired location for the scenario_runner code.

```bash
git clone -b leaderboard-2.0 --single-branch https://github.com/carla-simulator/scenario_runner.git
```

--------------------------------

### Python: Define Agent Sensors

Source: https://leaderboard.carla.org/get_started_v1_0

The `sensors` method defines the sensors required by the agent. It returns a list of dictionaries, where each dictionary specifies the sensor type, ID, and its intrinsic/extrinsic parameters. Ensure sensor types and counts adhere to CARLA's limitations to avoid setup failures.

```Python
def sensors(self):
    sensors = [
        {'type': 'sensor.camera.rgb', 'id': 'Center',
         'x': 0.7, 'y': 0.0, 'z': 1.60, 'roll': 0.0, 'pitch': 0.0, 'yaw': 0.0, 'width': 300, 'height': 200, 'fov': 100},
        {'type': 'sensor.lidar.ray_cast', 'id': 'LIDAR',
         'x': 0.7, 'y': -0.4, 'z': 1.60, 'roll': 0.0, 'pitch': 0.0, 'yaw': -45.0},
        {'type': 'sensor.other.radar', 'id': 'RADAR',
         'x': 0.7, 'y': -0.4, 'z': 1.60, 'roll': 0.0, 'pitch': 0.0, 'yaw': -45.0, 'fov': 30},
        {'type': 'sensor.other.gnss', 'id': 'GPS',
         'x': 0.7, 'y': -0.4, 'z': 1.60},
        {'type': 'sensor.other.imu', 'id': 'IMU',
         'x': 0.7, 'y': -0.4, 'z': 1.60, 'roll': 0.0, 'pitch': 0.0, 'yaw': -45.0},
        {'type': 'sensor.opendrive_map', 'id': 'OpenDRIVE', 'reading_frequency': 1},
        {'type': 'sensor.speedometer', 'id': 'Speed'},
    ]
    return sensors

```

--------------------------------

### Define CARLA Environment Variables

Source: https://leaderboard.carla.org/get_started_v1_0

Sets environment variables for CARLA, Scenario Runner, and Leaderboard, and updates the PYTHONPATH. These variables are crucial for the modules to locate each other. Add these to your ~/.bashrc file.

```shell
export CARLA_ROOT=PATH_TO_CARLA_ROOT
export SCENARIO_RUNNER_ROOT=PATH_TO_SCENARIO_RUNNER
export LEADERBOARD_ROOT=PATH_TO_LEADERBOARD
export PYTHONPATH="${CARLA_ROOT}/PythonAPI/carla/':'${SCENARIO_RUNNER_ROOT}':'${LEADERBOARD_ROOT}':'${CARLA_ROOT}/PythonAPI/carla/dist/carla-0.9.10-py3.7-linux-x86_64.egg':${PYTHONPATH}
```

--------------------------------

### Run Carla Leaderboard Evaluation Script

Source: https://leaderboard.carla.org/get_started_v1_0

This command executes the main evaluation script for the Carla Leaderboard. It takes the previously set environment variables to run the simulation and evaluate the agent's performance on various scenarios and routes across different towns.

```shell
./scripts/run_evaluation.sh
```

--------------------------------

### Run Leaderboard Evaluation

Source: https://leaderboard.carla.org/get_started_v2_1

Executes the leaderboard evaluation script using a provided bash script. This script simplifies the process of running the `leaderboard_evaluator.py` with necessary arguments, launching a pygame window for agent interaction.

```bash
./run_leaderboard.sh
```

--------------------------------

### Define CARLA Environment Variables

Source: https://leaderboard.carla.org/get_started_v2_1

Sets essential environment variables for CARLA, Scenario Runner, and Leaderboard, and updates the PYTHONPATH to include necessary directories. These variables allow different modules to locate each other.

```bash
export CARLA_ROOT=PATH_TO_CARLA_ROOT
export SCENARIO_RUNNER_ROOT=PATH_TO_SCENARIO_RUNNER
export LEADERBOARD_ROOT=PATH_TO_LEADERBOARD
export PYTHONPATH="${CARLA_ROOT}/PythonAPI/carla/":"${SCENARIO_RUNNER_ROOT}":"${LEADERBOARD_ROOT}":"${CARLA_ROOT}/PythonAPI/carla/dist/carla-0.9.14-py3.7-linux-x86_64.egg":${PYTHONPATH}
```

--------------------------------

### Define Autonomous Agent Entry Point in Python

Source: https://leaderboard.carla.org/get_started_v1_0

This Python code defines the `get_entry_point` function and a basic `MyAgent` class that inherits from `AutonomousAgent`. The `get_entry_point` function is used to automatically instantiate the agent, and `MyAgent` serves as a template for custom autonomous agent implementations.

```python
from leaderboard.autoagents.autonomous_agent import AutonomousAgent

def get_entry_point():
    return 'MyAgent'

class MyAgent(AutonomousAgent):
...
```

--------------------------------

### Python: Override Agent Destroy Method

Source: https://leaderboard.carla.org/get_started_v1_0

The `destroy` method is called at the end of each route and can be overridden for any necessary cleanup operations, such as releasing resources or memory.

```Python
def destroy(self):
    pass

```

--------------------------------

### Python: Override Agent Run Step Method

Source: https://leaderboard.carla.org/get_started_v1_0

The `run_step` method is executed at each time step to produce a control command for the vehicle. It receives sensor data and the current timestamp as input and must return a `carla.VehicleControl` object.

```Python
def run_step(self, input_data, timestamp):
    control = self._do_something_smart(input_data, timestamp)
    return control

```

--------------------------------

### Clone CARLA ROS Bridge Repository

Source: https://leaderboard.carla.org/get_started_v2_0

Clones the CARLA ROS bridge repository, which is necessary for ROS-based agents. This command includes the '--recurse-submodules' flag to fetch all necessary submodules and specifically checks out the 'leaderboard-2.0' branch.

```bash
git clone --recurse-submodules -b leaderboard-2.0 --single-branch https://github.com/carla-simulator/ros-bridge
```

--------------------------------

### Install EvalAI CLI

Source: https://leaderboard.carla.org/submit_v2_0

Installs the EvalAI command-line interface using pip. This is a prerequisite for interacting with the EvalAI platform for submissions.

```shell
pip install evalai
```

--------------------------------

### Install EvalAI CLI

Source: https://leaderboard.carla.org/submit_v2_1

Installs the EvalAI command-line interface tool using pip. This tool is necessary for interacting with the EvalAI platform for submissions.

```bash
pip install evalai

```

--------------------------------

### Set Environment Variables for Carla Leaderboard Evaluation

Source: https://leaderboard.carla.org/get_started_v1_0

These shell commands set environment variables required for running the Carla Leaderboard evaluation. They define paths to scenario and route files, simulation parameters, and the team agent script. These are crucial for configuring the evaluation process.

```shell
export SCENARIOS=${LEADERBOARD_ROOT}/data/all_towns_traffic_scenarios_public.json
export ROUTES=${LEADERBOARD_ROOT}/data/routes_devtest.xml
export REPETITIONS=1
export DEBUG_CHALLENGE=1
export TEAM_AGENT=${LEADERBOARD_ROOT}/leaderboard/autoagents/human_agent.py
export CHECKPOINT_ENDPOINT=${LEADERBOARD_ROOT}/results.json
export CHALLENGE_TRACK_CODENAME=SENSORS
```

--------------------------------

### Capture Sensor Data Script (Shell)

Source: https://leaderboard.carla.org/get_started_v2_1

This command executes the `capture_sensor_data.py` script, which is used in conjunction with CARLA logs to capture sensor data. The script allows users to configure sensors, weather, recorders, and output folders before running the log replay.

```bash
python capture_sensor_data.py

```

--------------------------------

### Define Agent Sensors Configuration

Source: https://leaderboard.carla.org/get_started_v2_0

Overrides the `sensors` method to define the sensors required by the agent. This method returns a list of dictionaries, where each dictionary describes a sensor with its type, ID, and other attributes like extrinsics and field of view. The sensor definitions must adhere to CARLA's coordinate system and predefined limits.

```python
def sensors(self):
    sensors = [
        {'type': 'sensor.camera.rgb', 'id': 'Center',
         'x': 0.7, 'y': 0.0, 'z': 1.60, 'roll': 0.0, 'pitch': 0.0, 'yaw': 0.0, 'width': 300, 'height': 200, 'fov': 100},
        {'type': 'sensor.lidar.ray_cast', 'id': 'LIDAR',
         'x': 0.7, 'y': -0.4, 'z': 1.60, 'roll': 0.0, 'pitch': 0.0, 'yaw': -45.0},
        {'type': 'sensor.other.radar', 'id': 'RADAR',
         'x': 0.7, 'y': -0.4, 'z': 1.60, 'roll': 0.0, 'pitch': 0.0, 'yaw': -45.0, 'fov': 30},
        {'type': 'sensor.other.gnss', 'id': 'GPS',
         'x': 0.7, 'y': -0.4, 'z': 1.60},
        {'type': 'sensor.other.imu', 'id': 'IMU',
         'x': 0.7, 'y': -0.4, 'z': 1.60, 'roll': 0.0, 'pitch': 0.0, 'yaw': -45.0},
        {'type': 'sensor.opendrive_map', 'id': 'OpenDRIVE', 'reading_frequency': 1},
        {'type': 'sensor.speedometer', 'id': 'Speed'},
    ]
    return sensors
```

--------------------------------

### Example Leaderboard Carla Route Results JSON

Source: https://leaderboard.carla.org/evaluation_v2_0

This JSON structure represents the results of a single route simulation in the Leaderboard Carla project. It includes general route information, a summary of infractions committed by the agent, scores achieved, and timing metadata. This data is crucial for evaluating agent performance.

```json
{
    "index": 0,
    "route_id": "RouteScenario_0_rep0",
    "status": "Failed - Agent deviated from the route",
    "num_infractions": 2,
    "infractions": {
        "collisions_layout": [],
        "collisions_pedestrian": [],
        "collisions_vehicle": [],
        "red_light": [
            "Agent ran a red light 203 at (x=341.25, y=209.1, z=0.104)"
        ],
        "stop_infraction": [],
        "outside_route_lanes": [],
        "min_speed_infractions": [],
        "yield_emergency_vehicle_infractions": [],
        "scenario_timeouts": [],
        "route_dev": [
            "Agent deviated from the route at (x=95.92, y=165.673, z=0.138)"
        ],
        "vehicle_blocked": [],
        "route_timeout": []
    },
    "scores": {
        "score_route": 25.4,
        "score_penalty": 0.8,
        "score_composed": 20.32
    },
    "meta": {
        "route_length": 530.846,
        "duration_game": 85.55,
        "duration_system": 21.034
    }
}

```

--------------------------------

### Example JSON Results for a Completed Route

Source: https://leaderboard.carla.org/evaluation_v2_1

This JSON object represents the results of a single simulated route. It includes route identification, status, a summary of infractions committed by the agent, detailed scores, and metadata about the route such as its length and duration. This structure is generated by the leaderboard upon completion or failure of a route.

```json
{
    "index": 0,
    "route_id": "RouteScenario_0_rep0",
    "status": "Failed - Agent deviated from the route",
    "num_infractions": 2,
    "infractions": {
        "collisions_layout": [],
        "collisions_pedestrian": [],
        "collisions_vehicle": [],
        "red_light": [
            "Agent ran a red light 203 at (x=341.25, y=209.1, z=0.104)"
        ],
        "stop_infraction": [],
        "outside_route_lanes": [],
        "min_speed_infractions": [],
        "yield_emergency_vehicle_infractions": [],
        "scenario_timeouts": [],
        "route_dev": [
            "Agent deviated from the route at (x=95.92, y=165.673, z=0.138)"
        ],
        "vehicle_blocked": [],
        "route_timeout": []
    },
    "scores": {
        "score_route": 25.4,
        "score_penalty": 0.8,
        "score_composed": 20.32
    },
    "meta": {
        "route_length": 530.846,
        "duration_game": 85.55,
        "duration_system": 21.034
    }
}
```

--------------------------------

### Leaderboard Route Results JSON Example

Source: https://leaderboard.carla.org/evaluation_v1_0

This JSON object represents the performance results for a single route in the Carla leaderboard. It includes route status, a list of infractions, calculated scores, and metadata such as route length and duration. The 'infractions' field details specific events like running red lights or deviating from the route.

```json
{
    "index": 0,
    "route_id": "RouteScenario_0_rep0",
    "status": "Failed - Agent deviated from the route",
    "num_infractions": 2,
    "infractions": {
        "collisions_layout": [],
        "collisions_pedestrian": [],
        "collisions_vehicle": [],
        "red_light": [
            "Agent ran a red light 203 at (x=341.25, y=209.1, z=0.104)"
        ],
        "stop_infraction": [],
        "outside_route_lanes": [],
        "route_dev": [
            "Agent deviated from the route at (x=95.92, y=165.673, z=0.138)"
        ],
        "route_timeout": []
    },
    "scores": {
        "score_route": 25.4,
        "score_penalty": 0.8,
        "score_composed": 20.32
    },
    "meta": {
        "route_length": 530.846,
        "duration_game": 85.55,
        "duration_system": 21.034
    }
}
```

--------------------------------

### Build CARLA Leaderboard Docker Image

Source: https://leaderboard.carla.org/submit_v1_0

This command invokes the `make_docker.sh` utility script from the leaderboard repository to build the Docker image for your agent. The script uses the previously defined environment variables and Dockerfile to create an image named `leaderboard-user`. This image encapsulates your agent and its dependencies for submission.

```bash
${LEADERBOARD_ROOT}/scripts/make_docker.sh

```

--------------------------------

### Source ROS and Agent Environment in Dockerfile

Source: https://leaderboard.carla.org/submit_v1_0

This command appends sourcing commands for ROS and the agent's environment to the `agent_sources.sh` file within the Dockerfile. It's essential for setting up the necessary paths and configurations for the agent to run within the leaderboard's environment. Ensure the ROS_DISTRO variable is correctly set.

```dockerfile
RUN /bin/echo \
  'source /opt/ros/$ROS_DISTRO/setup.bash; \
  source $TEAM_CODE_ROOT/install/setup.bash;' >> ${HOME}/agent_sources.sh

```

--------------------------------

### Push Docker Image for Submission

Source: https://leaderboard.carla.org/submit_v2_0

Pushes a Docker image to the EvalAI platform for submission to a specific track (phase). Replace '<image>:<tag>' with your Docker image name and tag, and '<phase_name>' with the target track's name.

```shell
evalai push <image>:<tag> --phase <phase_name>
```

--------------------------------

### Push Docker Image to CARLA Leaderboard

Source: https://leaderboard.carla.org/submit_v2_1

Pushes a Docker image to the CARLA Leaderboard for a specified phase. Replace '<image>:<tag>' with your Docker image name and tag, and '<phase_name>' with the target track (e.g., SENSORS, MAP).

```bash
evalai push <image>:<tag> --phase <phase_name>

```

--------------------------------

### Set Environment Variables for CARLA Leaderboard Build

Source: https://leaderboard.carla.org/submit_v1_0

These `export` commands set essential environment variables required for building the CARLA Leaderboard Docker image. Ensure that the paths for CARLA_ROOT, SCENARIO_RUNNER_ROOT, LEADERBOARD_ROOT, and TEAM_CODE_ROOT accurately point to their respective locations on your system. These variables are typically added to the `.bashrc` file.

```bash
export CARLA_ROOT=...
export SCENARIO_RUNNER_ROOT=...
export LEADERBOARD_ROOT=...
export TEAM_CODE_ROOT=...

```

--------------------------------

### Build Agent Docker Image Script

Source: https://leaderboard.carla.org/submit_v2_1

This script builds the Docker image for the agent. It takes an optional ROS distribution argument. Ensure environment variables like CARLA_ROOT, SCENARIO_RUNNER_ROOT, LEADERBOARD_ROOT, and TEAM_CODE_ROOT are set.

```shell
${LEADERBOARD_ROOT}/scripts/make_docker.sh [--ros-distro|-r ROS_DISTRO]

```

--------------------------------

### Build Docker Image for CARLA Agent

Source: https://leaderboard.carla.org/submit_v2_0

This script builds a Docker image for your CARLA agent. It ensures necessary environment variables are set and copies contents into the Dockerfile. This is crucial for encapsulating your agent for evaluation on the CARLA Leaderboard.

```shell
${LEADERBOARD_ROOT}/scripts/make_docker.sh [--ros-distro|-r ROS_DISTRO]
```

--------------------------------

### Set EvalAI User Token

Source: https://leaderboard.carla.org/submit_v2_1

Authenticates the EvalAI CLI with a user-specific token. The token can be retrieved from the 'Submit' section of the EvalAI website and is crucial for associating submissions with the correct user.

```bash
evalai set_token <token>

```

--------------------------------

### Set EvalAI User Token

Source: https://leaderboard.carla.org/submit_v2_0

Authenticates the EvalAI CLI with a user-specific token. This token is required to associate submissions with your user account and can be found in the 'Submit' section of the EvalAI website.

```shell
evalai set_token <token>
```

--------------------------------

### Route Representation (World Coordinates)

Source: https://leaderboard.carla.org/evaluation_v1_0

This snippet demonstrates route representation using world coordinates. Waypoints are dictionaries with 'x', 'y', and 'z' components, associated with a RoadOption.

```python
[({'x': 153.7, 'y': 15.6, 'z': 0.0}, RoadOption.LEFT),
 ({'x': 148.9, 'y': 67.8, 'z': 0.0}, RoadOption.RIGHT),
 ...
 ({'x': 180.7, 'y': 45.1, 'z': 1.2}, RoadOption.STRAIGHT)]
```

--------------------------------

### Route Representation (GPS Coordinates)

Source: https://leaderboard.carla.org/evaluation_v1_0

This snippet shows the representation of a route using GPS coordinates. Each waypoint is a dictionary containing 'z', 'lat', and 'lon', paired with a RoadOption enum.

```python
[({'z': 0.0, 'lat': 48.99822669411668, 'lon': 8.002271601998707}, RoadOption.LEFT),
 ({'z': 0.0, 'lat': 48.99822669411668, 'lon': 8.002709765148996}, RoadOption.RIGHT),
 ...
 ({'z': 0.0, 'lat': 48.99822679980298, 'lon': 8.002735250105061}, RoadOption.STRAIGHT)]
```
Tokens:
Raw