CARLA server needs to be running in the background for the data collection to work.

Running the script on lab computer (E5-5044):

• Navigate to CARLA directory in Rishi's directory.
• Launch CARLA using make launch
• After watiing for Unity Engine to load, press 'Play' button in Unity Engine.
• Run main.py (in PyCharm or terminal)

The main.py script performs the following key functions:

1. CARLA Connection: It connects to a running instance of the CARLA simulator, allowing interaction with the virtual environment.

2. Vehicle Setup: The script spawns a Tesla Model 3 (or any other specified vehicle) in the CARLA environment at a predefined spawn point.

3. Waypoint Generation: A series of waypoints are generated relative to the vehicle's spawn point. These waypoints dictate the path the vehicle will attempt to follow.

4. Model Predictive Control (MPC): The core of the script is an MPC setup. This includes defining the optimization problem where the vehicle's trajectory is optimized to follow the generated waypoints. The optimization problem uses a bicycle model for vehicle dynamics, constraints on control inputs (steering and acceleration), and a cost function that penalizes deviation from the desired trajectory and control effort.

5. Solver Configuration: IPOPT, an interior point optimizer, is configured with specific tolerances and parameters to solve the optimization problem at each time step.

6. Data Collection: During the simulation, the script collects data on the planned trajectory (open-loop data), the actual trajectory followed by the vehicle (closed-loop data), and the residuals (differences between the predicted and actual states).

7. Visualization and Debugging: The script generates plots for each iteration, showing the planned and actual trajectories, which are helpful for debugging and analysis. These plots are saved in a specified directory for later review.

8. Data Storage: At the end of the simulation, the collected data (closed-loop, open-loop, and residuals) are saved to CSV files for further analysis.

Upon successful completion of the simulation, the script will generate and save the following:

• A series of plots visualizing the vehicle's trajectory over time.
• CSV files containing detailed data on the vehicle's motion and control inputs:
  • closed_loop_data.csv: Data on the actual trajectory followed by the vehicle.
  • open_loop_data.csv: Data on the planned trajectory as computed by the MPC.
  • residuals_data.csv: Data on the residuals, highlighting the differences between the planned and actual trajectories.