SAE PT-158-2013 Autonomous Vehicles for Safer Driving (To Purchase Call 1-800-854-7179 USA Canada or 303-397-7956 Worldwide).pdf

1、PT-158 Jurgen Autonomous Vehicles for Safer Driving Autonomous Vehicles for Safer Driving Edited by Ronald K. Jurgen Autonomous Vehicles for Safer Driving Edited by Ronald K. Jurgen Progress In Technology s er Ies Progress In Technology s er Ies Self-driving cars are no longer in the realm of scienc

2、e fiction, thanks to the integration of numerous automotive technologies that have matured over many years. Technologies such as adaptive cruise control, forward collision warning, lane departure warning, and V2V/V2I communications are being merged into one complex system. The papers in this compend

3、ium were carefully selected to bring the reader up to date on successful demonstrations of autonomous vehicles, ongoing projects, and what the future may hold for this technology. It is divided into three sections: overview, major design and test collaborations, and a sampling of autonomous vehicle

4、research projects. This book will be of interest to a wide range of readers: engineers at automakers and electronic component suppliers; software engineers; computer systems analysts and architects; academics and researchers within the electronics, computing, and automotive industries; legislators,

5、managers, and other decision-makers in the government highway sector; traffic safety professionals; and insurance and legal practitioners. About the editor After graduating from Rensselaer Polytechnic Institute with a BEE, Ronald K. Jurgen held various technical magazine editorial staff positions, i

6、ncluding 30 years with IEEE Spectrum. Now retired, he is the editor of the Automotive Electronics Handbook and the Digital Consumer Electronics Handbook, and assistant editor of the Electronics Engineers Handbook, Fourth Edition. He is also the editor of more than a dozen SAE International books on

7、automotive electronics.Autonomous Vehicles for Safer DrivingOther SAE books of interest: V2V/V2I Communications for Improved Road Safety and Efficiency By Ronald K. Jurgen (Product Code: PT-154) Automotive E/E Reliability By John Day (Product Code: T-126) Automotive Software Engineering By Joerg Sch

8、aeuffele and Thomas Zurawka (Product Code: R-361) For more information or to order a book, contact SAE International at 400 Commonwealth Drive, Warrendale, PA 15096-0001, USA; phone 877-606-7323 (U.S. and Canada only) or 724-776-4970 (outside U.S. and Canada); fax 724-776-0790; email CustomerService

9、sae.org; website http:/books.sae.org.Autonomous Vehicles for Safer Driving By Ronald K. Jurgen Warrendale, Pennsylvania, USA Copyright 2013 SAE International. eISBN: 978-0-7680-8039-1400 Commonwealth Drive Warrendale, PA 15096-0001 USA E-mail: CustomerServicesae.org Phone: 877-606-7323 (inside USA a

10、nd Canada)724-776-4970 (outside USA) Fax: 724-776-0790 Copyright 2013 SAE International. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, distributed, or transmitted, in any form or by any means without the prior written permission of SAE Internationa

11、l. For permission and licensing requests, contact SAE Permissions, 400 Commonwealth Drive, Warrendale, PA 15096-0001 USA; e-mail: copyrightsae.org; phone: 724-772-4028; fax: 724-772-9765. ISBN 978-0-7680-7993-7 Library of Congress Catalog Number 2013932495 SAE Order Number PT-158 DOI 10.4271/PT-158

12、Information contained in this work has been obtained by SAE International from sources believed to be reliable. However, neither SAE International nor its authors guarantee the accuracy or completeness of any information published herein and neither SAE International nor its authors shall be respons

13、ible for any errors, omissions, or damages arising out of use of this information. This work is published with the understanding that SAE International and its authors are supplying information, but are not attempting to render engineering or other professional services. If such services are require

14、d, the assistance of an appropriate professional should be sought. To purchase bulk quantities, please contact SAE Customer Service e-mail: CustomerServicesae.org phone: 877-606-7323 (inside USA and Canada) 724-776-4970 (outside USA) fax: 724-776-0790 Visit the SAE Bookstore at books.sae.orgDedicati

15、on This book is dedicated to my friend Richard Keaton.vii Table of Contents Introduction 1 Overview: 3 Autonomous Driving A Practical Roadmap (2010-01-2335) Jeffrey D. Rupp and Anthony G. King 5 Major Design and Test Collaborations: 27 Sartre - Safe Road Trains for the Environment Reducing Fuel Cons

16、umption through Lower Aerodynamic Drag Coefficient (2011-36-0060) Arturo Dvila and Mario Nombela 29 Ohio State University Experiences at the DARPA Challenges (2008-01-2718) Keith A. Redmill, Umit Ozguner, Scott Biddlestone, Alex Hsieh, and John Martin 35 Low-Cost Autonomous Vehicles for Urban Enviro

17、nments (2008-01-2717) Mahesh K. Chengalva, Richard Bletsis, and Bernard P . Moss 43 Vehicle Safety Communications Applications: System Design yet there are also potential safety benefits from the pursuit of autonomous vehicles. This paper describes some of the practical obstacles in achieving those

18、goals, and explores the use of near term applications of technologies that will be by-products of pursuing them. This includes a partial history of autonomous vehicle development (Section 2), potential consumer acceptability issues (Section 3), followed by a development roadmap and discussion of som

19、e variables to be addressed before autonomous vehicles become viable (Sections 4 and 5), and ends with a consideration of collaborative relationships that could assist in acceleration of development and issue resolution (Section 6). 2.0. THE CURRENT STATE - PUTTING THE HYPE INTO PERSPECTIVE There ha

20、s been escalating excitement about fully autonomous vehicles in the robotics community for some time and the excitement has now spilled over to the automotive industry. The idea of a self-driving, road-ready vehicle sparks the imagination, and is a familiar concept due to repeated exposures in popul

21、ar culture; be it movies, cartoons, television, magazines, books or games. An exhibit at the 1939 Worlds Fair in New York 1presented a vision where cars would use “automatic radio control” to maintain safe distances, a depiction of transportation as it would be in 1960, then only 21 years into the f

22、uture. One of the earliest attempts at developing an actual vehicle was led by Dr. Robert E. Fenton who joined the faculty at Ohio State University in 1960 and was elected to the National Academy of Engineering in 2003 2 . It is believed that his pioneering research and experimentation in automatic

23、steering, lane changing, and car following resulted in the first demonstration of a vehicle that could drive itself. Since then, Autonomous Driving - A Practical Roadmap 2010-01-2335 Published 10/19/2010 Jeffrey D. Rupp and Anthony G. King Ford Motor Company Copyright 2010 SAE International6 OEMs, u

24、niversities, and governmental agencies worldwide have engineered or sponsored autonomous vehicle projects with different operating concepts and varying degrees of success. Most recently, the Defense Advanced Research Projects Agency (DARPA), an agency of the United States Department of Defense, spon

25、sored three autonomous vehicle challenges. While a number of media friendly successes resulted in good photo ops, those in technical fields and many others readily appreciate the magnitude of work required to mature these vehicles into a viable, real world, design. 2.1. Contemporary Error Rates - We

26、re Way Off In the months preceding the inaugural DARPA Grand Challenge in 2004, William “Red” Whittaker of Carnegie Mellons Robotics Institute, with over 65 robots to his credit, stated “We dont have the Henry Ford, or the Model T, of robotics”, “Robotics is not yet mainstream; its not yet a nationa

27、l conversation.” 3His contributions and those of his students over the next few years would move the needle significantly, but his comments suggest the true nature of the challenge. The error rates of robotically piloted vehicles today are still very high compared to human-piloted vehicles. At the 2

28、005 DARPA Grand Challenge (DGC2) 5 of the 23 finalists successfully finished the 132 mile course, while two years later, at the 2007 DARPA Urban Challenge Event (UCE), 6 of the 11 finalists finished a 60 mile course. The mean mileage between significant errors (failure) at these events was 120 miles

29、 for DGC2 and 100 miles for UCE 4 . The errors cannot be attributed to a single primary cause, rather, multiple simultaneous causes and interactions including sensing, interpretation of the scene and simplification of its full complexity, simplifying assumptions and non- representative tradeoffs bui

30、lt into the algorithms, as well as unintended software bugs and hardware durability. Compare robotically piloted vehicle errors to that of human drivers, who averaged 500,000 miles driven between crashes in 2008 5 . Despite humans being 3-4 orders of magnitude better at driving than robots, crashes

31、of varying severity occur regularly. In 2008 in the United States alone, there were 34,000 fatal crashes and 1.6 million injury crashes. Autonomous vehicles may need to be better drivers than humans, exhibiting fewer errors, to gain acceptance. The error rates inherent in todays autonomous vehicles

32、are unacceptable for real world deployment in the present and will be for some time to come. 2.2. Progress Has Been Slow Recalling the many predictions of a self-driving car over the last four decades, it is obvious that autonomous vehicles have taken and will take far longer than expected, especial

33、ly when it comes to operational safety. Fully autonomous vehicles today are the product of laboratories, test tracks, and prize winning competitions, mainly conducted under favorable conditions with minimal and controlled uncertainties and no penalty for error. With limited success even in ideal sit

34、uations, industry has little choice but to methodically split the problem into attainable steps, learning and developing the necessary enabling technologies along the way. The combination of radio detection and ranging (RADAR) functionalities was patented by Christian Hlsmeyer in 1904 6 , building o

35、n work from the mid-1800s by physicists James Maxwell and Heinrich Hertz. The majority of the development since then has been driven by maritime collision avoidance and military defense applications, including important signal processing extensions such as target velocity estimation based on frequen

36、cy shift as proposed by physicist Christian Doppler. Despite this early start, it wasnt until 1999, with seven years of focused target tracking and controls development as well as electronics miniaturization, that Ford Motor Company launched the worlds first-to-market radar- based ACC system with br

37、aking for an automotive application, on a Jaguar XKR. 7 More than a decade later, advances in sensing technology critical for autonomous vehicle applications are just now accelerating significantly. Functionality of automotive forward-looking radars is increasing, even while prices are decreasing, w

38、ith a drop of 75% over two generations expected in one case. 8The progression to todays state of the art dual mode electronically scanned systems has allowed industry to use the resulting increased accuracy and availability to expand to new customer functions. Digital camera systems have similarly b

39、een in existence for quite some time, with a patent application for “All Solid State Radiation Imagers” filed in 1968 9 , and are now progressing more rapidly too. CMOS imagers have demonstrated increasing sensitivity, dynamic range, and pixel count, while costs have decreased due to the large volum

40、es of consumer electronics applications. More recently, advancements in machine vision algorithms have enabled the evolution from lane tracking to significantly more complex vehicle and pedestrian detection and tracking functions. Fusion sensing systems are also starting to see more automotive applications as well. Combining multiple sensing modalities, fusion leverages the orthogonality that can be established where the strength of one complements the weakness of another. This can create a sensing system with