Implementasi Algoritma Convolutional Neural Network Pada Kendaraan Tanpa Awak Skala Kecil

Abstract

Autonomous Vehicle is a vehicle capable of navigating the car independently without requiring input from the driver. This research aims to design and manufacture a prototype of an unmanned vehicle that can maneuver across a simple artificial road. This study also aims to analyze the performance of the NVIDIA Jetson Nano in processing deep learning models and driving actuators according to the predictions given by the model. The research stages include designing a prototype, creating an artificial path, taking image data, conducting training, and then implementing the training model on the car prototype. After testing the prototype, the training model made the correct steering angle prediction using epoch 50 with RMSE train and validation, 0.1792 and 0.1896, respectively. NVIDIA Jetson Nano also performs well in computing steering angle predictions with live input from the camera.