Further Results on the Use of Binocular Vision for Highway Driving.pdf

下载文档

2
0
约2.78万字
约 12页
2017-04-09 发布于江苏
举报
版权申诉
保障服务

Further Results on the Use of Binocular Vision for Highway Driving.pdf

1、本文档共12页，可阅读全部内容。
2、原创力文档（book118）网站文档一经付费（服务费），不意味着购买了该文档的版权，仅供个人/单位学习、研究之用，不得用于商业用途，未经授权，严禁复制、发行、汇编、翻译或者网络传播等，侵权必究。
3、本站所有内容均由合作方或网友上传，本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺！文档内容仅供研究参考，付费前请自行鉴别。如您付费，意味着您自己接受本站规则且自行承担风险，本站不退款、不进行额外附加服务；查看《如何避免下载的几个坑》。如果您已付费下载过本站文档，您可以点击这里二次下载。
4、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等，请点击“版权申诉”（推荐），也可以打举报电话：400-050-0827(电话支持时间：9:00-18:30)。

Further Results on the Use of Binocular Vision for Highway Driving

Further Results on the Use of Binocular Vision for Highway Driving Joseph Weber and Michael AtkinEngineering Research CenterThe California Institute of TechnologyPasadena CA 91125 ABSTRACT In the 1995 Collision Avoidance and Automated Trac Management Sensors Proceedings we presented astereo vision based approach for detecting the relative location of other vehicles in highway driving. This systemwas able to track other vehicles which were within the stereo pair visual eld of view. In addition, importantcamera position variables such as inclination angle and camera height were dynamically calculated from imagedata without the need for external sensors.In collaboration with industrial partners we have tested the algorithm on additional video information undera number of dierent driving situations. The aect of noise in the camera geometry estimate on the algorithmsperformance was also examined. As a result of these tests and the additional requirement of real time performance,a number of modications were made to the algorithm to increase both its reliability and computational eciency.We present the results of the testing and well as the algorithmic changes here.Keywords: autonomous driving, stereo vision, intelligent vehicles, lane detection, obstacle avoidance. 1 Introduction We have been investigating the use of stereo vision for autonomous driving for a number of years, and havepresented algorithms and preliminary results in the past.3,8,9 Our approach makes extensive use of binocularstereopsis for both longitudinal and lateral vehicle control. Novel aspects of our system include (a) exploitation ofdomain constraints to simplify the search problem in nding binocular correspondences, (b) temporal integrationof the results of the stereo analysis to reliably track obstacles, (c) dealing with crowded trac scenes wheresubstantial segments of the lane boundaries may be occluded and (d) on-line updating of the external cameracalibration with respect to the road.Lon