usage_instructions.md

Running the Flightmare simulator on the SSD and running a simple RL example on a drone

1. Activate UEFI Mode:

   • Ensure your system is set to UEFI mode boot. This can typically be done by accessing your BIOS settings:

     • Restart your computer and press the key (e.g., F2, Del) to enter BIOS setup.
     • Navigate to the Boot or Startup tab.
     • Select UEFI mode as the boot option rather than legacy BIOS mode.
     • Save and exit BIOS settings.

     ⚠️ WARNING :- BE CAREFULL WHILE CHANGING BOOT MODE FROM LEGACY TO UEFI IN CASE IF YOU HAVE TO DO IT. YOU MAY LOOSE ALL FILES IN YOUR SYSTEM

2. Connect External SSD:

   • Ensure the external SSD (installed in ICY BOX IB-200T-C3 enclosure via USB 3.2) is connected to your system.

3. Boot into Ubuntu 20.04 from External SSD:

   • Restart your computer and access BIOS setup again (if necessary).
   • Select the external SSD where Ubuntu 20.04 is installed as the boot device.
   • Save changes and exit BIOS settings to boot into Ubuntu 20.04 from the external SSD.

4. Unity Rendering Setup:

   • Once Ubuntu 20.04 is booted, Unity rendering for Flightmare is enabled by running the executable RPG_Flightmare.x86_64 from the downloaded Flightmare binary. Double-click this executable in the /home/sdp/Desktop/RPG_Flightmare folder.

four - Running the mentioned file, the RPG_Flightmare GUI will appear as below, on which four different 3D scenes Industrial, Warehouse, Garage and Forest can be chosen.

Running the RL Example Test

In this section the step to run the python RL simulation using the default unity+flightmare executable file and the default 3D environments.

5. Activate Conda Environment:

   • Open a terminal in Ubuntu 20.04 and activate your Conda environment named ENVNAME where Flightmare is set up. Use:

     conda activate ENVNAME

6. Navigate to Flightmare Directory:

   • Change directory to where Flightmare is installed on the external SSD:

     cd /home/sdp/Desktop/flightmare/flightrl

7. Run Sample Reinforcement Learning Script:

   • Navigate to the examples directory:

     cd examples

   • Execute the sample script run_drone_controltest.py to run the simulator with a pre-trained controller. Use the following command:

     python3 run_drone_controltest.py --train 0 --render 1

   • Running the python code, a sample reinforcement learning (RL) training / testing algorithm will be run in the chosen environment (see the figure below):

     

     • And the the results of the simulation will be displayed as below:

     

Additional Information

• For detailed customization and troubleshooting:
  • Check the troubleshoot.md file on the docs/troubleshootbranch.

Running the ROS Example Test

Once everting is installed sucuessfully and Catkin build is succesful.

• Build examples in Flightros

  catkin build flightros -DBUILD_SAMPLES:=ON
  

• Launch example

roslaunch flightros rotors_gazebo.launch

Running the RL Example Test on the Forest Floor 3D model

The steps to run the simulation are the same as the ones mentioned under the Running the Flightmare simulator on the SSD and running a simple RL example on a drone above, except for the part regarding the 4th step and the below instruction should be followed instead:

4. Unity Rendering Setup:
   • Once Ubuntu 20.04 is booted, Unity rendering for Flightmare is enabled by running the executable forest26_06.x86_64 under the directory: /Desktop/flightmare/flightrl/examples

• The UI will look like below:

.

• And the drone during the simulation will look like below: