Driver Recognition on Segway

Driver Recognition on Segway

Hiroshi Sato (National Defense Academy of Japan, Japan) and Julien Rossignol (Special Military School of Saint-Cyr, France)
Copyright: © 2012 |Pages: 13
DOI: 10.4018/jalr.2012010107
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

Statistical machine learning approach to understand human behaviors has been attracting considerable amounts of attention in recent years. If the authors understand more about humans, the authors can make more user-friendly machines. In this paper, the authors propose the driver recognition method from their record of manipulations using support vector machine. The authors demonstrate the efficiency of the authors’ method using the Segway. The performance of the recognition is quite good especially when the authors introduce the pre-process with FFT.
Article Preview

1. Introduction

For long years, the paradigm of learning human behaviors has attracted considerable amount of attention for determining if it is possible to build pattern of human behaviors (Pentland, 1999; Atkeson et al., 2000; Qian, Ou, Wu, Meng, & Xu, 2010). These researches were done in different domains: emotion on face recognition, spoken language recognition, recognition of a lane change in vehicle by the driver’s behavior. The control of human dynamic features can enhance security applications such as passwords and fingerprint.

For instance, according to the FBI statistics, auto theft occurs about every 30 seconds in the United States. Of course, some constructors have developed systems that let the owner to stop his vehicle remotely, but if we can invent a process that makes the car intelligent, the car itself would detect that it has been theft. Thus, the stealing could be avoided.

Nuria and Pentland (2000) find a method for detecting driver’s intentions by using a motion based driving simulator. The method relies on the Hidden Markov Models. In this work, Nuria and Pentland attempt to predict the intentions for a given situation (for instance for lane change: emergency lane change, normal lane change and lane keeping), not the driver himself among others.

Our goal is the opposite: We focus on the driver instead of his intention. In this case, Kuge, Yamamura, Shimoyama, and Liu (2000) show that Support Vector Machine is better suited, because SVM can be applied for different kind of trajectory rather than for each trajectory, HMM need to be changed. Huihan et al. use the Support Vector Machine in order to make the distinction between several drivers using a simple simulator. They show that SVM is more powerful than HMM for driver behavior recognition.

In this paper, we present an intelligent driver identification system using the support vector machine (SVM) (Cortes & Vapnik, 1995). SVM has been applied to many areas, such as pattern recognition, regression or equalization. The aim of this research is to improve the safety of the gyro-pod Segway by detecting a non-authorized driver.

The reason we choose Segway as a target vehicle is that human behavior changes drastically from one person to another with this machine and it is totally new way of driving. This topic just establishes the first step of a long study about behavior recognition. If we can identify who drives the Segway, the same technique can be used to estimate more complex things such as the estimation of driver’s intention. It may be more complex, but it can have a huge application.

This paper consists of seven sections. Section 2 describes the experimental apparatus – the vehicle and the sensor – in this research. We propose the method of recognition using SVM in Section 3. Section 4 explains the experimental settings. The experimental result is shown in Section 5 and discussed in Section 6. We conclude this paper in Section 7.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 7: 2 Issues (2017)
Volume 6: 2 Issues (2016)
Volume 5: 1 Issue (2015)
Volume 4: 1 Issue (2014)
Volume 3: 4 Issues (2012)
Volume 2: 4 Issues (2011)
Volume 1: 4 Issues (2010)
View Complete Journal Contents Listing