# Matlab Physics for Drones

Project AirSim allows users to import their existing Simulink physics models as an alternative to the built-in Fast Physics. The AirSim server transmits information about the applied vehicular forces and the environmental state which the Simulink model processes to compute the vehicle's kinematic state, all through TCP.

## Prerequisites

Download and install Matlab + Simulink 2022a with the Aerospace Blockset add-on.

## Install Matlab Engine API for Python

In order to connect to Matlab from the Python client, the Matlab Engine API for Python must be installed. More information can be found on the MathWorks [documentation](https://www.mathworks.com/help/matlab/matlab_external/install-the-matlab-engine-for-python.html) but the general installation steps are prescribed below:
1. Run a command prompt as Administrator
1. Activate a Python virtual environment and navigate to the directory where the Matlab API Python library is installed: `cd "C:\Program Files\MATLAB\R2022a\extern\engines\python"`
1. Install the Matlab API Python library: `python setup.py install`

## Example Scripts with Simulink Physics

The example Python client scripts and configuration files can be found in `client/python/example_user_scripts/` and the corresponding Matlab model files can be found in `client/python/example_user_scripts/simulink/`. There are currently two user scripts - `simulink_quadrotor_standardized_model.py` and `simulink_quadtiltrotor_standardized_model.py` - both of which utilize the same Simulink model but with different parameter values for each airframe. The AirSim Simulink interface currently supports up to eight rotors and eight control surfaces.

### Directory Information
- Scripts beginning with `simulink_` in their name utilize Simulink physics. Each script references a scene configuration file.
- The `sim_config/` folder contains both scene configuration files and robot configuration files. Note the former contains a `physics-connection` parameter to connect to Matlab and references the latter in the `robot-config` parameter. The robot configuration file has its `physics-type` parameter set to `matlab-physics`.
- The `simulink/` folder contains the Matlab model loader m-file scripts, the Simulink models, and the S-function mex file. Each Matlab script loads the vehicle parameters, opens the Simulink model, and configures the Matlab Engine for the client script to use Python APIs to control. The Simulink model contains the drone and fixed-wing physics for the vehicle, and references the vehicle parameters loaded via the Matlab script.

### Running the Scripts
1. Open Matlab 2022a.
1. Navigate the Matlab file explorer to the `client/python/example_user_scripts/simulink/` folder.
1. Run the model loader m-file from the **Matlab console**:
   1. ex. `>> load_quadrotor_simulink_physics_model`
1. Launch and run the simulation server.
1. In an activated Python environment, navigate to `client/python/example_user_scripts/` and run the corresponding client script:
    1. ex. `python simulink_quadrotor_standardized_model.py`
1. After the simulation scene has started, it will automatically start the Simulink model through the Matlab Python API.
1. Once the script completes, it will automatically stop the Simulink model through the Matlab Python API. To restart the script, just run it again. To stop the simulation while the it is in progress, press Ctrl-C from the Python script console output to stop the in-progress flight command, stop Simulink, and disconnect the client. The simulation can be started again by simply running the script again.

---

Copyright (C) Microsoft Corporation.  All rights reserved.