i-Car: An Intelligent and Interactive Interface for Driver Assistance System

Abstract

The aim of the present research was to reduce accidents by assisting the driver in various aspects of driving such as lane detection, pedestrian and car detection, driver drowsiness detection and rear view parking assistance. The methodology combines the computer vision techniques with pattern recognition, feature extraction, machine learning, object recognition, human computer interaction and parallel processing in a nutshell. The proposed system provides robust extraction of lane markings in various types and alerts the driver attempting to drift from the lane. It also detects the pedestrians and cars which are at a vulnerable distance to be hit by the vehicle and alarms the driver well ahead of time. The system uses eye closure based decision algorithm to detect driver drowsiness in all conditions and also warns by interactive voice early enough to avoid the accidents. It also assists the driver while reversing the vehicle, by providing a clear view of his blind spot areas. Computer vision algorithms like Hough’s Transform, Canny Edge detection and HAAR classifiers were applied to meet the objectives. The integrated module was analyzed and tested in different terrains and various lighting condition to produce an accurate and robust real-time assistance system (Sivaraman et al., 2014). iCar is an innovative prototype in the Information Technology with minimum hardware like low cost webcams. It emerged as an Interactive Technology with an interactive audio, visual, touch and touch-less interfaces. These can assist to avoid accidents in the world by intelligently ignoring certain hardware sensors like IR, UV, Acoustic, Proximity and mechanical devices like costlier LIDAR (Light Detection and Ranging) fitted in Google Car. Present research findings outperform the state of the art research like CalTech (Aly et al., 1997). Attempts of depth sensing even using Microsoft Kinect could be ignored by the present technology, the iCar.

Keywords: iCar; Canny Edge detection; HAAR Classifier; Probabilistic Hough’s; Transform