These instructions are shortcuts for more general knowledge found in donkeycar docs
Install the SW stack using those instructions.
If using OSX then remember to use the same version of Tensorflow (atm 1.8) AND the same Python version (3.5.3) as in the pi. Using the install/envs/mac.yml will install the correct version to conda env 'donkey535' Note!! Keras models are NOT compatible between different tensorflow versions!
We have our own template, create it to your home folder like this:
´donkey createcar --template donkey2futucar ~/futucar´
TODO using the real car
Or using the simulator, just follow donkey instructions
´cd ~/futucar´
´python manage.py train --model=<output-path-to-model> --tub=<path-to-data> --type=<linear|categorical>´
TODO Either using the car
Or the simulator, first run the server and then using the simulator app.
´donkey sim --type <linear|categorical> --model <path-to-model>´
You need the wlan router, turn it on. Turn the car on by plugging in the usb cable to the rasberry pi. The password for the wlan and raspberry pi are written on the router.
Connect to the raspberry pi
´ssh [email protected]´
cd to the mycar dir
´cd ~/mycar´
to run car and record training data run
´python manage.py drive --js´
to see stuff from the camera and debug stuff run
´python manage.py drive´