bazilinskyy / truck-driver-profile

Modeling the profile of a truck driver based on big data.

Stability in Truck Driving Behaviour: A Geo-Specific Analysis

Table of contents

• Table of contents
• Introduction
• Installation
• Dataset Description
• Code Description
• Getting help

Introduction

This repository is part of an MSc thesis project, which can be found in the TU Delft repository. As part of the project geo-specific analysis has been conducted for a Dutch Field Operational Test(FOT), known as the Anti-ongevalsystemen AOS-FOT. In the current study, an analysis of driving data from 1,727 trucks was recorded over one year between September 2008 – May 2009 (across Europe). The aim is to explore stability in truck driving behaviour, focusing on time and location (urban areas and motorways).

Installation

All the required libraries can be installed using requirement.txt, which can be found in the root directory.

1. Open a terminal or command prompt.
2. Navigate to the folder with your requirements.txt.
3. Run the following command:

pip install -r requirements.txt

4. You are done installing dependencies.

Dataset Description

The dataset used in this study was recorded as part of a large-scale FOT aimed at assessing accident prevention systems by the Dutch Ministry of Infrastructure and Water Management. Connekt conducted a large-scale field operational test for trucks with active driver assistance systems, also known as accident prevention systems (APS). Five different accident prevention systems and a registration system were tested on Dutch highways over eight months. The test’s purpose was to understand better the extent to which accident prevention systems can contribute to traffic safety and traffic flow on the Dutch road network. Until now, the contribution of these systems has only been examined to a limited extent [1].

The dataset consists of two types of data - Orderly Use data (Trip data) and Main AOS Data (AOS Events data).

1. TRIP: TRIP folder consists of SUMMARY and DETAIL files. The detailed folder contains the orderly use data described below. Features were recorded every 2 km.

Feature	Description
Numberplate	License plate number used to identify vehicles
Point_date-stamp	Date per point
Point_time-stamp	Time per point
Latitude	GPS: 3 digits to denote the integer part + 6 decimal digits
Longitude	GPS: 3 digits to denote the integer part + 6 decimal digits
Meters_travelled	m: Meters travelled from previous point
Time_elapsed	sec: Seconds elapsed from previous point
Point_speed	km/h
Road_type	According to TeleAtlas: -3=not available, 0=urban, 1=motorway, 2=extra-urban, 3=other
Speed_restriction	According to TeleAtlas, km/h
Admin_area	According to TeleAtlas, standard name for Order 8 area
TNO_Time-stamp	Control variable

Fig. below depicts trips recorded across Europe during the field operational test. The data was recorded for over 96 $\times$ 106 kilometres.

2. Main AOS Data: Recorded using Mobile Eye. Contrary to Trip data, this data produced a warning when an event is triggered (the driver receives a real-time alert). The events include:
   1. Braking Events
   2. Headway Warnings: (Level I/II/III-HW)
   3. Right and Left Indicator: (R/L-I)
   4. Right and Left Lane Departure Warnings: (R/L-LDW)

The figure below visualizes the different events recorded in the Netherlands.

Code Description

In order to explore stability, correlation analysis has been used to examine if a linear relationship exists between variables. Clustering analysis has been used to understand further how stable; different ranges are for a particular metric. The spatial stability analysis has been split based on the environment, i.e. urban areas and motorways.

All the jupyter notebooks can be found in tdstability folder.

File	Description
eda	Used to perform exploratory data analysis for both Trip and AOS data. It also contains temporal stability analysis.
extract_trips_two_locs_motorway	Code to extract trips between two locations.
urban_area_clustering	Correlation and clustering analysis for urban areas
motorway_clusternig	Correlation and clustering analysis for motorways
discussion	Exploring effect of vehicle characteristics on stability

Getting help

The dataset used in this study is not publicly available. You can contact Dr. Joost de Winter or Dr. Dimitra Dodou for further details.

[1] AOS FOT (Wiki of Field Operational Tests)” Apr 2015. [Online]. Available: https://wiki.fot-net.eu/index.php/AOS