PhD, EngD and MSc by research theses (SATM)

Permanent URI for this collection

https://dspace.lib.cranfield.ac.uk/handle/1826/8739

Browse

Browsing PhD, EngD and MSc by research theses (SATM) by Course name "PhD in Transport"

Now showing 1 - 2 of 2
  • Thumbnail Image
    ItemOpen Access
    Driver lane change intention inference using machine learning methods.
    (2018-04) Xing, Yang; Cao, Dongpu; Velenis, Efstathios
    Lane changing manoeuvre on highway is a highly interactive task for human drivers. The intelligent vehicles and the advanced driver assistance systems (ADAS) need to have proper awareness of the traffic context as well as the driver. The ADAS also need to understand the driver potential intent correctly since it shares the control authority with the human driver. This study provides a research on the driver intention inference, particular focus on the lane change manoeuvre on highways. This report is organised in a paper basis, where each chapter corresponding to a publication, which is submitted or to be submitted. Part Ⅰ introduce the motivation and general methodology framework for this thesis. Part Ⅱ includes the literature survey and the state-of-art of driver intention inference. Part Ⅲ contains the techniques for traffic context perception that focus on the lane detection. A literature review on lane detection techniques and its integration with parallel driving framework is proposed. Next, a novel integrated lane detection system is designed. Part Ⅳ contains two parts, which provides the driver behaviour monitoring system for normal driving and secondary tasks detection. The first part is based on the conventional feature selection methods while the second part introduces an end-to-end deep learning framework. The design and analysis of driver lane change intention inference system for the lane change manoeuvre is proposed in Part Ⅴ. Finally, discussions and conclusions are made in Part Ⅵ. A major contribution of this project is to propose novel algorithms which accurately model the driver intention inference process. Lane change intention will be recognised based on machine learning (ML) methods due to its good reasoning and generalizing characteristics. Sensors in the vehicle are used to capture context traffic information, vehicle dynamics, and driver behaviours information. Machine learning and image processing are the techniques to recognise human driver behaviour.
  • Thumbnail Image
    ItemOpen Access
    Safety culture: a legal standard for commercial aviation.
    (2017-10) Lawrenson, Anthony James; Braithwaite, Graham R.
    Although a link between organisational safety culture and human behaviour is well established within academic literature, ambiguity about the actual nature of the causal relationship has inhibited its practical application. This thesis aims to establish a legal standard of safety culture by producing a model which describes the relationship between organisational safety culture and potential corporate liability. The model, called d3SC, attempts to promote a defence of due diligence to potential prosecution by improving an organisation’s safety culture. The thesis consists of three sequential studies. The first study comprises of twenty-six accident case studies from which data is developed into a prototype model through a process of grounded theory. The subsequent studies then take the emergent model from a construct to a risk management tool that was applied and tested against a real-world data from commercial aviation and law. In attempting to develop a model, d3SC, the thesis has adopted a predominantly functionalist approach. However, it is recognised that the complexities of culture and causation are not sufficiently represented without adopting some methods of real world analysis. This recognition of the need to dig deeper into organisational dynamics is manifest in the use of qualitative methods in the thesis to triangulate the output of the d3SC process. It is also represented in the units of measurement or case studies from which safety culture is frequently described. The quality of safety culture is often described in terms of organisational performance yet a consistent theme in both the literature and the data collated in these studies, shows that aggregating organisational safety culture as a singular measurement can be misleading. Contrasting the data from different departments and hierarchical levels within an organisation gives a much deeper and contextual understanding of internal dynamics and influences. This is of particular relevance to corporate liability in the aftermath of an accident. Prosecuting agencies will not focus their investigation on the adequacy of overall metrics of organisational assessments, but on the perceived causal links between an accident and the weaker areas of organisational safety culture. By improving the visibility and understanding of the causal links between corporate liability and corporate culture it is hoped that this research can contribute to enhancing safety standards in commercial aviation.