Computational Efficiency: Can Something as Small as a Raspberry Pi Complete the Computations Required to Follow the Path?

Abstract

This chapter explains the development processes of a prototype autonomous toy car. It focuses on the design and implementation of transforming a normal remote control toy car into a self-contained vehicle with the capability to drive autonomously. This would be proven by making it follow a track of any layout. It uses a neural network (NN) in the form of a multilayer perceptron (MLP) to process images in real time to generate a movement instruction. Upon completion, the vehicle demonstrated the ability to be able to follow a track of any layout, while staying between both sides of the track. The collision avoidance system proved to be effective up to a distance of 50 cm in front of the vehicle in order to let it stop prior to hitting an object. The neural network processing of the image in order to classify it in a real time proved to be above the expectation of around 5 FPS and has an accuracy score of over 90%.