Analysing Behavioural Data from On-Road Driving Studies: Handling the Challenges of Data Processing
The analysis of real world driving data entails numerous challenges. In this chapter, several strategies are proposed to meet challenges that surface in data storage, data extraction, data correction and data enrichment. The strategies are illustrated wit
- PDF / 630,898 Bytes
- 20 Pages / 476.22 x 680.315 pts Page_size
- 28 Downloads / 240 Views
9.1
Introduction
While driving studies on closed test-tracks or in driving simulators are characterised by a high degree of experimental control (e. g. [1]), the diverse requirements of urban driving cannot be easily simulated within these settings (e. g. concerning scenario programming). Moreover, experimental studies with a high degree of control over potentially confounding variables necessarily fail to reflect the complexity of real world scenarios. Observing driving behaviour in real-traffic environments, however, entails numerous challenges of data handling (similar to naturalistic driving studies), before data analysis, interpretation or modelling can be performed. In behavioural research, this will mostly entail calculations of statistical parameters for specific performance measures [1], e. g. mean speed when entering an intersection. When aiming to predict driving manoeuvres, this also involves the preparation of “clean” datasets, which allow for appropriate data modelling techniques and robust classification results free of interferences. The present chapter aims to outline methodological issues and provides guidelines for analysts to tackle the numerous data (and video) handling challenges that surface when confronted with similar onroad driving studies. Specifically, these data handling challenges refer to:
data storage, extraction of pertinent data, data correction procedures, enrichment of data.
The challenges and possible solutions are illustrated with examples from a study that was conducted as part of the UR:BAN research initiative “Behaviour Prediction and IntenM. Graichen () V. Nitsch B. Färber () Human Factors Institute, Universität der Bundeswehr München Neubiberg, Germany © Springer Fachmedien Wiesbaden GmbH 2018 K. Bengler et al. (eds.), UR:BAN Human Factors in Traffic, ATZ/MTZ-Fachbuch, DOI 10.1007/978-3-658-15418-9_9
163
9
164
M. Graichen et al.
tion Detection”, which aimed at investigating driving behaviour when approaching intersections under real environmental conditions. Specifically, research objectives (amongst others) entailed: a. exploration of on-road driving behaviour when approaching different kinds of intersections, b. identification of indicators for predicting turning manoeuvres, c. establishing a data base as basis for the algorithm implementation. After briefly outlining the research rationale and the data collection process, the data handling challenges and possible solutions are described.
9.2 Driving at Urban Intersections Driving in urban areas, one encounters a great variety and dynamic of traffic participants in diverse scenarios. Additionally, urban traffic scenarios vary with regard to many different environmental features which are likely to affect driving behaviour. Environmental conditions can rapidly change as the driver moves along the blocks and stretches of roads with variations in geometry or other architectural features and constraints which influence the general visibility of traffic participants and other potentially safety-relevan
Data Loading...