文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > Probabilistic Models of Driver Behavior
侵权投诉

Probabilistic Models of Driver Behavior

The BATMobile [1] is one approach to the implementation of a system capable of driving autonomously in normal highway traffic. As with human drivers, the BATMobile would be a superior driver if it were able to make predictions about the actions of other dr

Probabilistic Models of Driver Behavior Computer Science Division University of California Berkeley, CA 94720-1776, USA oza@cs.berkeley.edu

Nikunj C. Oza

The BATMobile 1] is one approach to the implementation of a system capable of driving autonomously in normal highway tra c. As with human drivers, the BATMobile would be a superior driver if it were able to make predictions about the actions of other drivers, e ectively giving it the ability to respond before such actions even occur. The ability to predict driver behavior would also be helpful in the creation of accurate models of tra c behavior used in freeway design and analysis. In this report, we discuss the creation of dynamic probabilistic networks (DPNs) that constitute models of driver behavior. Speci cally, we discuss three model structures we have created, the data used to learn the models, and how well the models predict driver behavior. We compare the three models to observe the bene ts of modeling hidden state and using deterministic variables. We also analyze the models to see what aspects of driver behavior the models have learned.

Abstract

1 Introduction The BATMobile 1] is one approach to the implementation of a system capable of driving autonomously in normal highway tra c. It uses dynamic probabilistic networks (DPNs) to model the dynamic and partiallyobservable world of driving. Such a network is to be used by the BATMobile to maintain the current belief state, where the state variables that constitute the complete state may include (for example) the locations and velocities of other cars (actual and observed values), the intentions of other drivers, and the orientation of the road. Of course, many of these variables are unobservable some or all of the time. The main aim of this research is to derive models of the factors a ecting the intentions of other drivers, such as speeding up, slowing down, and changing lanes. Creating models of driver behavior will allow an autonomous driver such as the BATMobile to anticipate potentially dangerous situations, such as another car cutting in front of it. It could also warn human drivers of dangerous situations as part of a driver assistance module. Detailed driver models are also essential in the creation of accurate models of tra c behavior used in freeway design and analysis. These also constitute models of driver preferences and behavior which can be used to assess driver competence 2]. Creating our probabilistic driver models involves identifying the variables essential to predicting human driver behavior, using a learning algorithm 3] to automatically construct a DPN model from data, and assessing how well the model predicts drivers' actions. The data comes from either a driving simulator 1] or from videotapes. From the videotapes the locations of various cars, their speeds, and other information are extracted automatically using computer vision algorithms 4]. In the Transportation community, the type of model we are construct

ing is considered to be a microscopic model because it models the behavior of individual vehicles. This is di erent from macroscopic models which describe general tra c characteristics such as density and ow 8] 16]. In this report, we rst brie y describe microscopic tra c models. In particular, we explain the motivations for developing microscopic models and survey the types of microscopic models that have been postulated and, in some cases, validated1 This previous work motivates our decision to create a DPN model. We 1

Supported by CalTrans (PATH MOU 130) and the NSF Graduate Research Training Program in Spatial Cognition. In this paper, we will limit ourselves to discussing models of highway tra c.

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email