Phillip Taylor
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Phillip Taylor has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
- Nathan Griffiths
University of Warwick, Coventry, UK
5 shared publications Thomas Popham
Jaguar and Land Rover Research, Coventry, UK
4 shared publicationsNathan Griffiths
The University of Warwick
3 shared publicationsAbhir Bhalerao
Department of Computer Science, University of Warwick, Coventry, UK
2 shared publicationsAdam Gelencser
Jaguar and Land Rover Research, Coventry, UK
2 shared publicationsAlain Dunoyer
Jaguar Land Rover Research
2 shared publications
Latest Publications
- Open Access
- 2 September 2021
- Published by Wiley in Mobile Information Systems
- Vol. 2021, 1-20
Classifying Vehicle Activity to Improve Point of Interest Extraction
Knowledge of drivers’ mobility patterns is useful for enabling context-aware intelligent vehicle functionality, such as route suggestions, cabin preconditioning, and power management for electric vehicles. Such patterns are often described in terms of the Points of Interest (PoIs) visited by an individual. However, existing PoI extraction methods are general purpose and typically rely on detecting periods of low mobility, meaning that when they are applied to vehicle data, they often extract a large number of false PoIs (for example, incorrectly extracting PoIs due to stopping in traffic), reducing their usefulness. To reduce the number of false PoIs that are extracted, we propose using features derived from vehicle signals, such as the selected gear and status of doors, to classify candidate PoIs and filter out those that are irrelevant. In this paper, we (i) present Activity-based Vehicle PoI Extraction (AVPE), a wrapper method around existing PoI extraction methods, that utilizes a postclustering classification stage to filter out false PoIs, (ii) evaluate the benefits of AVPE compared to three state-of-the-art general purpose PoI extraction algorithms, and (iii) demonstrate the effectiveness of AVPE when applied to real-world driving data.
- 2 September 2021
- Published by Wiley in Mobile Information Systems
- Vol. 2021, 1-20
Open AccessClassifying Vehicle Activity to Improve Point of Interest Extraction
Knowledge of drivers’ mobility patterns is useful for enabling context-aware intelligent vehicle functionality, such as route suggestions, cabin preconditioning, and power management for electric vehicles. Such patterns are often described in terms of the Points of Interest (PoIs) visited by an individual. However, existing PoI extraction methods are general purpose and typically rely on detecting periods of low mobility, meaning that when they are applied to vehicle data, they often extract a large number of false PoIs (for example, incorrectly extracting PoIs due to stopping in traffic), reducing their usefulness. To reduce the number of false PoIs that are extracted, we propose using features derived from vehicle signals, such as the selected gear and status of doors, to classify candidate PoIs and filter out those that are irrelevant. In this paper, we (i) present Activity-based Vehicle PoI Extraction (AVPE), a wrapper method around existing PoI extraction methods, that utilizes a postclustering classification stage to filter out false PoIs, (ii) evaluate the benefits of AVPE compared to three state-of-the-art general purpose PoI extraction algorithms, and (iii) demonstrate the effectiveness of AVPE when applied to real-world driving data.
- Open Access
- 17 November 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 207465-207484
Privacy Challenges With Protecting Live Vehicular Location Context
Future Intelligent Transport Systems (ITS) will require that vehicles are equipped with Dedicated Short Range Communications (DSRC). With these DSRC capabilities, new privacy threats are emerging that can be taken advantage of by threat actors with little experience and cheap components. However, the origins of these privacy threats are not limited to the vehicle and its communications, but extend to non-vehicular devices carried by the driver and passengers. A shortcoming of existing work is that it tends to focus on a specific aspect of privacy leakage when attempting to protect location privacy. In doing so, interactions between privacy threats are not considered. In this work, we investigate the privacy surface of a vehicle by considering the many different ways in which location privacy can be leaked. Following this, we identify techniques to protect privacy and that it is insufficient to provide location privacy against a single threat vector. A methodology to calculate the interactions of privacy preserving techniques is used to highlight the need to consider the wider threat landscape and for techniques to collaborate to ensure location privacy is provided against multiple sources of privacy threats where possible.
- 17 November 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 207465-207484
Open AccessPrivacy Challenges With Protecting Live Vehicular Location Context
Future Intelligent Transport Systems (ITS) will require that vehicles are equipped with Dedicated Short Range Communications (DSRC). With these DSRC capabilities, new privacy threats are emerging that can be taken advantage of by threat actors with little experience and cheap components. However, the origins of these privacy threats are not limited to the vehicle and its communications, but extend to non-vehicular devices carried by the driver and passengers. A shortcoming of existing work is that it tends to focus on a specific aspect of privacy leakage when attempting to protect location privacy. In doing so, interactions between privacy threats are not considered. In this work, we investigate the privacy surface of a vehicle by considering the many different ways in which location privacy can be leaked. Following this, we identify techniques to protect privacy and that it is insufficient to provide location privacy against a single threat vector. A methodology to calculate the interactions of privacy preserving techniques is used to highlight the need to consider the wider threat landscape and for techniques to collaborate to ensure location privacy is provided against multiple sources of privacy threats where possible.
- 12 June 2018
- Published by Maximum Academic Press in The Knowledge Engineering Review
- Vol. 33, 1-28
Reputation assessment: a review and unifying abstraction
Trust and reputation allow agents to make informed decisions about potential interactions. Trust in an agent is derived from direct experience with that agent, while reputation is determined by the experiences reported by other witness agents with potentially differing viewpoints. These experiences are typically aggregated in a trust and reputation model, of which there are several types that focus on different aspects. Such aspects include handling subjective perspectives of witnesses, dishonesty, or assessing the reputation of new agents. In this paper, we distil reputation systems into their fundamental aspects, discussing first how trust and reputation information is represented and second how it is disseminated among agents. Based on these discussions, a unifying abstraction is presented for trust and reputation systems, which is demonstrated by instantiating it with a broad range of reputation systems found in the literature. The abstraction is then instantiated to combine the range of capabilities of existing reputation systems in the Machine Learning Reputation System, which is evaluated using a marketplace simulation.
- 12 June 2018
- Published by Maximum Academic Press in The Knowledge Engineering Review
- Vol. 33, 1-28
Reputation assessment: a review and unifying abstraction
Trust and reputation allow agents to make informed decisions about potential interactions. Trust in an agent is derived from direct experience with that agent, while reputation is determined by the experiences reported by other witness agents with potentially differing viewpoints. These experiences are typically aggregated in a trust and reputation model, of which there are several types that focus on different aspects. Such aspects include handling subjective perspectives of witnesses, dishonesty, or assessing the reputation of new agents. In this paper, we distil reputation systems into their fundamental aspects, discussing first how trust and reputation information is represented and second how it is disseminated among agents. Based on these discussions, a unifying abstraction is presented for trust and reputation systems, which is demonstrated by instantiating it with a broad range of reputation systems found in the literature. The abstraction is then instantiated to combine the range of capabilities of existing reputation systems in the Machine Learning Reputation System, which is evaluated using a marketplace simulation.
- 25 November 2017
- Published by Springer Nature
- p. 84-102
Stereotype Reputation with Limited Observability
- 25 November 2017
- Published by Springer Nature
- p. 84-102
Stereotype Reputation with Limited Observability
- 1 July 2017
- Published by IGI Global Scientific Publishing in International Journal of Mobile Human Computer Interaction
- Vol. 9 (3) , 54-72
Investigating the Feasibility of Vehicle Telemetry Data as a Means of Predicting Driver Workload
Driving is a safety critical task that requires a high level of attention from the driver. Although drivers have limited attentional resources, they often perform secondary tasks such as eating or using a mobile phone. When performing multiple tasks in the vehicle, the driver can become overloaded and the risk of a crash is increased. If a vehicle is aware that the driver is currently under high workload, the vehicle functionality can be changed in order to minimise any further demand. Traditionally, workload is measured using physiological sensors that require often intrusive and expensive equipment. Another approach may be to use vehicle telemetry data as a performance measure for workload. In this paper, the authors present the Warwick-JLR Driver Monitoring Dataset (DMD) and analyse it to investigate the feasibility of using vehicle telemetry data for determining the driver workload. They perform a statistical analysis of subjective ratings, physiological data, and vehicle telemetry data collected during a track study. A data mining methodology is then presented to build predictive models using this data, for the driver workload monitoring problem.
- 1 July 2017
- Published by IGI Global Scientific Publishing in International Journal of Mobile Human Computer Interaction
- Vol. 9 (3) , 54-72
Investigating the Feasibility of Vehicle Telemetry Data as a Means of Predicting Driver Workload
Driving is a safety critical task that requires a high level of attention from the driver. Although drivers have limited attentional resources, they often perform secondary tasks such as eating or using a mobile phone. When performing multiple tasks in the vehicle, the driver can become overloaded and the risk of a crash is increased. If a vehicle is aware that the driver is currently under high workload, the vehicle functionality can be changed in order to minimise any further demand. Traditionally, workload is measured using physiological sensors that require often intrusive and expensive equipment. Another approach may be to use vehicle telemetry data as a performance measure for workload. In this paper, the authors present the Warwick-JLR Driver Monitoring Dataset (DMD) and analyse it to investigate the feasibility of using vehicle telemetry data for determining the driver workload. They perform a statistical analysis of subjective ratings, physiological data, and vehicle telemetry data collected during a track study. A data mining methodology is then presented to build predictive models using this data, for the driver workload monitoring problem.
- Open Access
- 15 March 2016
- Published by Taylor & Francis in Applied Artificial Intelligence
- Vol. 30 (3) , 233-256
Data Mining for Vehicle Telemetry
This article presents a data mining methodology for driving-condition monitoring via CAN-bus data that is based on the general data mining process. The approach is applicable to many driving condition problems, and the example of road type classification without the use of location information is investigated. Location information from Global Positioning Satellites and related map data are often not available (for business reasons), or cannot represent the full dynamics of road conditions. In this work, Controller Area Network (CAN)-bus signals are used instead as inputs to models produced by machine learning algorithms. Road type classification is formulated as two related labeling problems: Road Type (A, B, C, and Motorway) and Carriageway Type (Single or Dual). An investigation is presented into preprocessing steps required prior to applying machine learning algorithms, that is, signal selection, feature extraction, and feature selection. The selection methods used include principal components analysis (PCA) and mutual information (MI), which are used to determine the relevance and redundancy of extracted features and are performed in various combinations. Finally, because there is an inherent bias toward certain road and carriageway labelings, the issue of class imbalance in classification is explained and investigated. A system is produced, which is demonstrated to successfully ascertain road type from CAN-bus data, and it is shown that the classification correlates well with input signals such as vehicle speed, steering wheel angle, and suspension height.
- 15 March 2016
- Published by Taylor & Francis in Applied Artificial Intelligence
- Vol. 30 (3) , 233-256
Open AccessData Mining for Vehicle Telemetry
This article presents a data mining methodology for driving-condition monitoring via CAN-bus data that is based on the general data mining process. The approach is applicable to many driving condition problems, and the example of road type classification without the use of location information is investigated. Location information from Global Positioning Satellites and related map data are often not available (for business reasons), or cannot represent the full dynamics of road conditions. In this work, Controller Area Network (CAN)-bus signals are used instead as inputs to models produced by machine learning algorithms. Road type classification is formulated as two related labeling problems: Road Type (A, B, C, and Motorway) and Carriageway Type (Single or Dual). An investigation is presented into preprocessing steps required prior to applying machine learning algorithms, that is, signal selection, feature extraction, and feature selection. The selection methods used include principal components analysis (PCA) and mutual information (MI), which are used to determine the relevance and redundancy of extracted features and are performed in various combinations. Finally, because there is an inherent bias toward certain road and carriageway labelings, the issue of class imbalance in classification is explained and investigated. A system is produced, which is demonstrated to successfully ascertain road type from CAN-bus data, and it is shown that the classification correlates well with input signals such as vehicle speed, steering wheel angle, and suspension height.
- 25 November 2015
- Published by Springer Nature
- p. 53-67
Context-Driven Assessment of Provider Reputation in Composite Provision Scenarios
- 25 November 2015
- Published by Springer Nature
- p. 53-67
Context-Driven Assessment of Provider Reputation in Composite Provision Scenarios
- Open Access
- 17 October 2012
- Published by Association for Computing Machinery (ACM)
- p. 233-240
Road type classification through data mining
In this paper we investigate data mining approaches to road type classification based on CAN (controller area network) bus data collected from vehicles on UK roads. We consider three related classification problems: road type (A, B, C and Motorway), signage (None, White, Green and Blue) and carriageway type (Single or Double). Knowledge of these classifications has a number of uses, including tuning the engine and adapting the user interface according to the situation. Furthermore, the current road type and surrounding area gives an indication of the driver's workload. In a residential area the driver is likely to be overloaded, while they may be under stimulated on a highway. Several data mining and temporal analysis techniques are investigated, along with selected ensemble classifiers and initial attempts to deal with a class imbalance present in the data. We find that the Random Forest ensemble algorithm has the best performance, with an AUC of 0.89 when used with a wavelet-Gaussian summary of the previous 2.5 seconds of speed and steering wheel angle recordings. We show that this technique is at least as good as a model-based solution that was manually created using domain expertise.
- 17 October 2012
- Published by Association for Computing Machinery (ACM)
- p. 233-240
Open AccessRoad type classification through data mining
In this paper we investigate data mining approaches to road type classification based on CAN (controller area network) bus data collected from vehicles on UK roads. We consider three related classification problems: road type (A, B, C and Motorway), signage (None, White, Green and Blue) and carriageway type (Single or Double). Knowledge of these classifications has a number of uses, including tuning the engine and adapting the user interface according to the situation. Furthermore, the current road type and surrounding area gives an indication of the driver's workload. In a residential area the driver is likely to be overloaded, while they may be under stimulated on a highway. Several data mining and temporal analysis techniques are investigated, along with selected ensemble classifiers and initial attempts to deal with a class imbalance present in the data. We find that the Random Forest ensemble algorithm has the best performance, with an AUC of 0.89 when used with a wavelet-Gaussian summary of the previous 2.5 seconds of speed and steering wheel angle recordings. We show that this technique is at least as good as a model-based solution that was manually created using domain expertise.