Toward Smart Cars with Computer Vision for Integrated Driver and Road Scene Monitoring



Download 0.69 Mb.
Page1/3
Date conversion28.05.2018
Size0.69 Mb.
  1   2   3
Workshop on the Challenges and Promises of an Ecological Approach to Robotics

Toward Smart Cars with Computer Vision for
Integrated Driver and Road Scene Monitoring


Alexander Zelinsky

Seeing Machines Pty Ltd

Innovations Building

Canberra, ACT 0200 Australia
Email alex@seeingmachines.com

Introduction

This paper presents the results from an Intelligent Transport System (ITS) project that integrates automated driver monitoring and lane tracking systems. The experimental results from the integration of the lane tracker and the driver monitoring system are presented with an analysis of the driver's visual behaviour in several different driving scenarios.


Studies for road traffic authorities worldwide estimate that about 30% of all fatal car crashes can be attributed to driver inattention and fatigue.

Numerous studies have been performed to analyse signs of driver fatigue through the measurement of the visual demand on the driver. This is often through frame-by-frame analysis or infrared corneal-reflection technologies. While these studies produce valuable results, they are often time consuming and too unreliable for many research purposes. An ITS project has recently been initiated at The Australian National University (ANU) and Seeing Machines which has focused on autonomous driver monitoring and autonomous vehicle control to aid the driver The major aim of this project has been the development of a system of cooperating internal and external vehicle sensors for understanding the behaviour of a driver.


Understanding what the driver looks at is only the first step in investigating the psychology of the driver. Research into the context in which distractions occur and the effect it has on the driver's ability to control the vehicle is an important part of this investigation.
This paper presents the results from the initial phase of this study where a lane tracker was developed using a computer vision based fusing multiple visual cues. The lane tracker was integrated with a driver monitoring system developed by Seeing Machines called faceLAB, to investigate the visual behaviour of the driver in a number of common driving scenarios.
The lane tracker was tested in autonomous driving experiments in a number of

situations including along a freeway and along road with no lane markings.



Experimental Setup

The experimental vehicle is a 1999 Toyota Landcruiser 4WD that has been fitted with several actuation devices to control steering, acceleration and braking.



Vision is the main form of sensing used on the vehicle, which has two different vision platforms installed. A passive set of stereo cameras is mounted on the dashboard facing the driver and is used as part of the faceLAB system for driver monitoring.

Figure 1
An active stereo vision platform, called Cedar designed at ANU, carries 4 cameras - one pair used for stereovision in the near-field, and one pair for far-field stereo experiments and for mid-field to far-field scene coverage. Various other sensors have been fitted to the vehicle including a Global Positioning System (GPS), Inertial Navigation Sensor (INS), and a laser range finder. Figure 1 shows the vehicle, faceLAB camera set and a picture of the Cedar sensor platform that is mounted in place of the rear view mirror.
  1   2   3


The database is protected by copyright ©ininet.org 2016
send message

    Main page