In this post I wish to outline the build of a compact indoor autonomous car using neural network for navigation. The setup can be used to train and validate autonomous driving neural network models quickly and easily.
The robot is a two-wheel differential steering type. The robot chassis was laser cut out of 5mm clear acrylic, in order to minimize self-shadowing in front of the camera. As you can see there is no caster wheel, rather balance is via a zip-tie tied in a loop. The usual hardware is present, Raspberry Pi 2 and the Adafruit Motor HAT. A 2200mAh 2S battery piowers the setup. A 3A UBEC is used for stepping down the voltage to drive the Pi and PWM generator.
The robot is loaded with a Raspbian image, and uses the Burro autonomous driving platform to steer and control throttle. Burro uses a pre-trained Convolutional Neural Network model trained using Keras and Tensorflow.
The track is available as a PDF file and can be printed on A0 size paper for best results. The default model included in Burro is suitable for this track.
Sorry guys, I’ve been trying to wrestle the editing system into submission but I failed. Anyway, I have a couple of posts detailing the build at my blog https://backyardrobotics.eu, in particular:
