Using the NVIDIA Jetson Nano for banknote classification and to control LEGO power functions’ motors and servomotors feeding single notes to realize a money counting machine.
A detailed project description and source code can be found in this GitHub repository.
Model training and inference are done with the NVIDIA Jetson Nano. Transfer learning (VGG16 base + 3 dense layers (25,10,6)) is used to classify 5, 10, 20 and 50 EUR notes, the background and counterfeit money. To control LEGO Power Functions’ motors and servomotors with the Jetson a PCA9685 and an H-bridge (L293) is used. For more information on the motor control part check out my previous post “Controlling LEGO Technic motor/servo with the Nvidia Jetson” or this video “Jetson Nano - PCA9685 motor/servo control”.
Detailed Project Description & Implementation: