Comparison of Machine Learning Algorithm’s on Self-Driving Car Navigation using Nvidia Jetson Nano

Abstract

The vehicle is the most facility for business and living in the present. The vehicle is used in abundant activities, which lead to an increasing number of accidents on the roads. Many statistical reports pointed that more than 80% of accident causes came from direct human causes such as violating the speed limit, illegal overtaking, and suddenly cutting in. Therefore, the self-driving car was rapidly developed by starting a scaled RC-Car platform. This paper presents a self-driving car model to study behavior by using the three machine learning algorithms: SVM, ANN-MLP, and CNN-LSTM in 3-speed levels and 3 scenarios, both with obstacles and without obstacle. The results show that CNN-LSTM is the best accuracy in every scenario and speed levels.