This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 1 Powered Two-Wheelers Critical Events Detection and Recognition Using Data-Driven Approaches Ferhat Attal, Abderrahmane Boubezoul , Allou Samé, Latifa Oukhellou , and Stéphane Espié Abstract— Driving errors are considered to be the greatest contributory cause in all road accidents and an important contributory cause of most fatal accidents. This is particularly the case for the users of powered two-wheeled vehicles (PTWs), perhaps because PTW riders play a greater role in the control of their vehicles’ stability than four-wheeled vehicle drivers. Thus, observing and analyzing the evolution of riders’ behavior in a real-life context is an important step in the identification of the road environment characteristics that constitute a risk factor for PTW riders. A relevant research issue in naturalistic studies is related to the detection and identification of critical riding events from among the vast amount of data recorded during the experiment. In this paper, two approaches were used to auto- matically detect such critical riding events. First, we formalized this problem in terms of detecting changes in the mean and variance of the signals generated by the acceleration and angular velocity sensors. For this purpose, two steps were performed: 1) a data segmentation and feature extraction step in which the multidimensional time series of accelerometer and angular velocity data were segmented and modeled using a Gaussian mixture model with quadratic logistic proportions; and 2) a clas- sification step in which each detected segment was assigned to the appropriate riding sequence, whether “naturalistic” or “critical” (i.e., fall or near fall), using the k-nearest neighbor algorithm. The second approach was based on online fall detection. This methodology used control charts (a multivariate cumulative sum), an approach that has been traditionally employed for sequential detection. These two algorithms were applied to a database composed of data from a real-life driving experiment. The obtained results show the effectiveness of both methodologies. Index Terms— Powered two-wheelers, critical events detection and recognition, data mining approaches. I. I NTRODUCTION I N RECENT years, the public authorities have been making considerable efforts to reduce the number of road fatalities. This has resulted in the implementation of a number of enforcement measures not only for speed (fixed and mobile radars) but also for alcohol. Although these measures have led Manuscript received January 2, 2017; revised June 27, 2017 and November 23, 2017; accepted January 8, 2018. The Associate Editor for this paper was Y. Gao. (Corresponding author: Abderrahmane Boubezoul.) F. Attal is with the Laboratory of Images, Signals and Intelligent Sys- tems, University of Paris-Est Cr ´ eteil, 94400 Vitry-Sur-Seine, France (e-mail: abderrahmane.boubezoul@ifsttar.fr). A. Boubezoul and S. Espié are with UPE-IFSTTAR/TS2/SIMU&MOTO, F-77447 Marne la Vallée Cedex, France. A. Samé and L. Oukhellou are with UPE-IFSTTAR/COSYS/GRETTIA, F-77447 Marne la Vallée Cedex, France. Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TITS.2018.2797065 to a considerable reduction in road mortality, all categories of road users have not benefited equally from this improvement. In 2013 for example, the statistics for the riders of powered two-wheelers (PTWs) were alarming: although they repre- sented just 2% of road traffic, they accounted for almost a quarter of all fatalities. Road safety statistics also show that PTWs account for 43% of severely injured road users, and what makes matters worse is that the injuries they sustain have serious consequences. Overall, the statistics highlight the high vulnerability of PTW riders compared with other road users. This group accounts for 24% of road mortality in France, but accounts for only 2% of the traffic: in other terms, a PTW rider is at 20 times greater risk of being killed on the road than a car driver [1]. In spite of this excess risk, the PTWs market has grown considerably in recent years. Individuals choose to use a PTW for their home-to-work trips mainly because of shorter journey times and the implementation of new sustainable urban transport policies which aim to reduce private car use in urban areas, especially city centres. Within this context of increasing PTW traffic, the safety of PTW users is assuming considerable importance. It should also be noted that the majority of PTW accidents occur because of interactions with other road users, most noticeably the refusal of car drivers to give the right-of-way. However, single-vehicle accidents, such as falls, are also an important factor. In 2013 in France, one-third of the fatal PTW accidents occurred without an identified third party (37.7%). The loss of PTW control is common largely because of the complexity of PTW dynamics and the intrinsic instability of such vehicles. The analysis of accidents related to this mode of transport led us to classify these accidents into two classes: (1) single-vehicle accidents, such as falls and (2) accidents involving other road users. Thus, for the study of the first type of accident, we were interested in the development of an online fall detection algorithm, for the deployment of airbag jacket or any other alert system. As far as the accident with other road users is a complex and multi-factorial problem, the understanding of the factors that contribute to crashes is a hard and time consuming task for safety researchers. Usually, epidemiological and empirical methods have been used to assess driving safety. Many large-scale research programs have been enacted in Europe and elsewhere to understand the factors that contribute to crashes [2]. During the last few years, thanks to the developments in the embedded electronics and sensors domains, instrumented vehicles are increasingly 1524-9050 © 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.