Research of Water Detection in Autonomous Vehicles

Abstract

An autonomous car is a ground vehicle that navigates without human input. These vehicles are expected to reach $60 billion in sales by 2025. But these autonomous vehicles have many drawbacks: one among them is lack of water detection. This problem has created havoc in the normal operation of autonomous vehicles which has interested researchers to develop algorithms to overcome this problem. This research aims to address the fundamental challenges pertaining to this issue using image classification. For this task we first design a general image classification system for water detection and propose a heuristic solution to classify the images. Secondly, we adopt a machine learning technique and develop an algorithm to classify the images. We consider the same image data set for both the models. The results show that the detection of water in three different climatic conditions is feasible and convenient for the proposed model. The results from the proposed image classification system for gray scale and color and the machine learning technique are different, and the image classification model has more accuracy than the machine learning technique.