Data Wrangling

13Obstacle Avoidance Simulation and Real-Time Lane Detection for AI-Based Self-Driving Car

B. Eshwar*, Harshaditya Sheoran, Shivansh Pathak and Meena Rao

Department of ECE, Maharaja Surajmal Institute of Technology, Janakpuri, New Delhi, India

Abstract

This chapter aims at developing an efficient car module that makes the car drive autonomously from one point to another avoiding objects in its pathway through use of Artificial Intelligence. Further, the authors make use of visual cues to detect lanes and prevents vehicle from driving off road/moving into other lanes. The paper is a combination of two simulations; first, the self-driving car simulation and second, real-time lane detection. In this work, Kivy package present in Anaconda navigator is used for simulations. Hough transformation method is used for lane detection in “restricted search area.”

Keywords: Self-driving car, artificial intelligence, real-time lane detection, obstacle avoidance

13.1 Introduction

A self-driving car is designed to move on its own with no or minimal human intervention. It is also called autonomous or driverless car many times in literature [1]. The automotive industry is rapidly evolving and with it the concept of self-driving cars is also evolving very fast. Several companies are focused on developing their own self. Even the tech giants, which are not into “mainstream automobile,” like Google and Uber, seem greatly interested in it. This is due to the ease of driving opportunity that self-driving ...