- Autopilot v1 (alpha): Download
- pyvjoy: Download
- Python 3.5 wih preinstalled packages*: Download
- vJoy: Download
ETS2Autopilot-master.zipand open the directory
- Extract the content of the folder in
- Run vJoySetup.exe
- Copy the file
To start the autopilot gui just execute
If you have used an older version of the autopilot before and recorded some data, just move the
captured-folder to the new version and an automatic migration will start.
You can also copy
settings.py to the new autopilot folder to migrate the settings as well.
After the first start, please open the settings and configure the autopilot. Use the sliders to set up the area for the autopilot. Just select the windscreen.
Next select your controller and set all buttons and axes. Simply click ‘Set’ next to the input field and press the button or move the joystick/steering wheel for the axes.
For ‘vJoy Device’ enter 1.
- Go to Options > Controls
- Select ‘Keyboard + vJoy Device’ as first input
- Select your other controller as second device (maybe there are more devices to set)
- In the autopilot GUI start the recording mode.
- Back in-game set the steering axis by moving your joystick/steering wheel. It should select something like ‘Joy Y-Axis’ (IMPORTANT! It has to be ‘Joy’! Not ‘Joy 2’ or something else.)
- Set the rest of the axes and buttons.
- Now go to Options > Gameplay and disable ‘Automatic Blinker Turn off’. This is necessary to get the right data while recording. The indicator state will get saved too. In the training mode, data with indicator turned on will not get used.
To enter the recording mode, select ‘Recording’ and press Start. Now you should see an image and the state in the status bar.
Press the autopilot button on your controller to start or stop the recording.
To start training mode, just select ‘Training’ and press Start. This process will take a while. Make sure you have collected much data. (~20.000 images)
The trained model is saved in the directory
save. The old model will get overwritten.
Finally! To use the autopilot and relax a bit while playing, select ‘Autopilot’ and press Start. In-game you can press the autopilot button and it will steer for you. If it’s doing weird stuff and you have to take over control, it will save this correction. You will see it in the sequence list as an entry with a note.
If you don’t want to collect your own data and train a model you can try to use my model.
- Model: Download
Download it and extract the content into
ETS2Autopilot-master\save. If the
save directory does not exist, just create one.
This model was trained for roads like in Germany. Just try it out and look how it works.
The downloadable version of python contains the following packages:
- bleach 1.5.0 (via pip)
- enum34 1.1.6 (via pip)
- html5lib 0.9999999 (via pip)
- Markdown 2.6.9 (via pip)
- numpy 1.13.3+mkl (via http://www.lfd.uci.edu/~gohlke/pythonlibs/#numpy (12.11.2017 10:05))
- olefile 0.44 (via pip)
- opencv-python 188.8.131.52 (via pip)
- Pillow 4.3.0 (via pip)
- pip 9.0.1 (via https://bootstrap.pypa.io/get-pip.py (12.11.2017 10:04))
- protobuf 3.4.0 (via pip)
- pygame 1.9.3 (via pip)
- PyQt5 5.9.1 (via pip)
- scipy 0.19.1 (via http://www.silx.org/pub/wheelhouse/scipy-0.19.1-cp35-cp35m-win_amd64.whl (12.11.2017 10:04))
- setuptools 36.7.1 (via pip)
- sip 4.19.5 (via pip)
- six 1.11.0 (via pip)
- tensorflow 1.4.0 (via pip)
- tensorflow-tensorboard 0.4.0rc2 (via pip)
- Werkzeug 0.12.2 (via pip)
- wheel 0.30.0 (via pip)