Title: Data-driven Modelling on Powered Two Wheelers using Machine Learning
Subject:
Level: ej valt
Description: Problem description
According to the European Commission (ec.europa.eu), in 2015, 15% motorcycle riders and 3% moped (and similar powered two-wheelers) riders road fatalities. As it can be found in [1-2], there are around 90% of road-traffic crashes caused by driver error (i.e. as inattention, loss of vigilance, mental under/overload) and unsafe behavior (i.e. inadequate training or lack of experience). Improving road safety includes understanding the individual, collective and interaction behaviour of riders. This work proposes a development of a methodology for riding patterns classification by using machine learning techniques. The riding pattern classification problem will be formulated as a classification problem aiming to identify the class of the riding situation by using inertial sensor data. This inertial sensor data was collected from three accelerometer and three-gyroscope sensors mounted on the motorcycle. These measurements constitute experimental database which was valuable to analyze Powered Two Wheelers (PTW) rider behavior. In a previous work [3], the obtained results based on the raw 3D inertial measurements (accelerometers / gyroscopes) data shown the effectiveness of such approach.

The project work is subdivided as follows:
1. Literature study and state-of-the-art
This task requires a systematic literature review to identify the features, time and frequency domain analysis approaches for feature extractions, approaches for feature selection/ranking, and machine learning approaches for classification for PTW riding manoeuvring. Student requires presenting an analytical summary of the state-of-the-art based on the literature study.

2. Implementation
Analysing the PTW's riding data of various events for a given dataset. Student also requires developing an approach using machine-learning algorithms for detection of different riding patterns based on the riding events and sensory dataset.

3. Evaluation
Student should evaluate the proposed approach and learning algorithm for detecting the patterns. Make a comparison between different feature extraction and selection methods on the data set that will be given.

4. Report writing
It is expected that student provide a report as a completion of the project work. Report consists of background, problem formulation, state-of-the-art, methods, evaluation, and discussion.

Reference:
1. Elander, J., West, R, French D, "Behavioural correlates of individual differences in road-traffic crash risk: An examination of methods and findings", Psychological bulletin 113.2 (1993): 279
2. Feyer, A.M., Williamson, A. & Friswell, R., "Balancing work and rest to combat driver fatigue: an investigation of two-up driving in Australia". Accident Analysis and Prevention 1997 Jul, 29, 541-553.
3. F. Attal, A. Boubezoul, L. Oukhellou and S. EspiƩ, "Riding patterns recognition for Powered two-wheelers users' behaviors analysis," 16th International IEEE Conference on Intelligent Transportation Systems (ITSC 2013), The Hague, 2013, pp. 2033-2038.
Start date: 2018-03-26
End date: 2018-06-30
Prerequisites:
IDT supervisors: Mobyen Uddin Ahmed
Examiner:
Comments: