Context-aware intelligent decisions: online assessment of heavy goods vehicle driving risk
There is a growing interest in assessing the impact of drivers' actions and behaviours on road safety due to the numerous road fatalities and costs attributed to them. For Heavy Goods Vehicle (HGV) drivers, assessing the road safety risks of their behaviours is a subject of interest for researc...
| Main Author: | |
|---|---|
| Format: | Thesis (University of Nottingham only) |
| Language: | English |
| Published: |
2023
|
| Subjects: | |
| Online Access: | https://eprints.nottingham.ac.uk/73680/ |
| _version_ | 1848800802449129472 |
|---|---|
| author | Mafeni Mase, Jimiama Mosima |
| author_facet | Mafeni Mase, Jimiama Mosima |
| author_sort | Mafeni Mase, Jimiama Mosima |
| building | Nottingham Research Data Repository |
| collection | Online Access |
| description | There is a growing interest in assessing the impact of drivers' actions and behaviours on road safety due to the numerous road fatalities and costs attributed to them. For Heavy Goods Vehicle (HGV) drivers, assessing the road safety risks of their behaviours is a subject of interest for researchers, governments and transport companies, as nations rely on HGVs for the delivery of goods and services. However, HGV driving is a complex, dynamic, uncertain and multifaceted task, mostly influenced by individual traits and external contextual factors. Advanced computational and artificial intelligence (AI) methods have provided promising solutions to automatically characterise the manner by which drivers operate vehicle controls and assess their impact on road safety. However, several challenges and limitations are faced by the current intelligence-supported driving risk assessment approaches proposed by researchers, such as: (1) the lack of comprehensive driving risk datasets; (2) information about the impact of inevitable contextual factors on HGV drivers' responses is not considered, such as drivers' physical and mental states, weather conditions, traffic conditions, road geometry, road types, and work schedules; (3) ambiguity in the definition of driving behaviours is not considered; and (4) imprecision of AI models, and variability in experts' subjective views are not considered.
To overcome the aforementioned challenges and limitations, this multidisciplinary research aims at exploring multiple sources of data including information about the impact of contextual factors captured from crucial stakeholders in the HGV sector to develop a reliable context-aware driving risk assessment framework. To achieve this aim, AI methods are explored to accurately detect drivers' driving styles, affective states and driving postures using telematics data, facial images, and driver posture images respectively. Subsequently, due to the lack of comprehensive driving risk datasets, fuzzy expert systems (FESs) are explored to fuse detected driving behaviours and perceived external factors using knowledge from domain experts.
The key findings of this research are: (1) recurrent neural networks are effective in capturing the temporal dynamics and differences between the different types of driver distraction postures and affective states; (2) there is a trade-off between efficiency and privacy in processing facial images using AI approaches; (3) the fusion of driver behaviours and external factors using FESs produces realistic, reliable and fair driving risk assessments; and (4) a hierarchical representation of a decision-making process simplifies reasoning compared to flat representations. |
| first_indexed | 2025-11-14T20:57:21Z |
| format | Thesis (University of Nottingham only) |
| id | nottingham-73680 |
| institution | University of Nottingham Malaysia Campus |
| institution_category | Local University |
| language | English |
| last_indexed | 2025-11-14T20:57:21Z |
| publishDate | 2023 |
| recordtype | eprints |
| repository_type | Digital Repository |
| spelling | nottingham-736802023-07-26T04:40:35Z https://eprints.nottingham.ac.uk/73680/ Context-aware intelligent decisions: online assessment of heavy goods vehicle driving risk Mafeni Mase, Jimiama Mosima There is a growing interest in assessing the impact of drivers' actions and behaviours on road safety due to the numerous road fatalities and costs attributed to them. For Heavy Goods Vehicle (HGV) drivers, assessing the road safety risks of their behaviours is a subject of interest for researchers, governments and transport companies, as nations rely on HGVs for the delivery of goods and services. However, HGV driving is a complex, dynamic, uncertain and multifaceted task, mostly influenced by individual traits and external contextual factors. Advanced computational and artificial intelligence (AI) methods have provided promising solutions to automatically characterise the manner by which drivers operate vehicle controls and assess their impact on road safety. However, several challenges and limitations are faced by the current intelligence-supported driving risk assessment approaches proposed by researchers, such as: (1) the lack of comprehensive driving risk datasets; (2) information about the impact of inevitable contextual factors on HGV drivers' responses is not considered, such as drivers' physical and mental states, weather conditions, traffic conditions, road geometry, road types, and work schedules; (3) ambiguity in the definition of driving behaviours is not considered; and (4) imprecision of AI models, and variability in experts' subjective views are not considered. To overcome the aforementioned challenges and limitations, this multidisciplinary research aims at exploring multiple sources of data including information about the impact of contextual factors captured from crucial stakeholders in the HGV sector to develop a reliable context-aware driving risk assessment framework. To achieve this aim, AI methods are explored to accurately detect drivers' driving styles, affective states and driving postures using telematics data, facial images, and driver posture images respectively. Subsequently, due to the lack of comprehensive driving risk datasets, fuzzy expert systems (FESs) are explored to fuse detected driving behaviours and perceived external factors using knowledge from domain experts. The key findings of this research are: (1) recurrent neural networks are effective in capturing the temporal dynamics and differences between the different types of driver distraction postures and affective states; (2) there is a trade-off between efficiency and privacy in processing facial images using AI approaches; (3) the fusion of driver behaviours and external factors using FESs produces realistic, reliable and fair driving risk assessments; and (4) a hierarchical representation of a decision-making process simplifies reasoning compared to flat representations. 2023-07-26 Thesis (University of Nottingham only) NonPeerReviewed application/pdf en cc_by https://eprints.nottingham.ac.uk/73680/1/Jim_s_PhD_thesis_final.pdf Mafeni Mase, Jimiama Mosima (2023) Context-aware intelligent decisions: online assessment of heavy goods vehicle driving risk. PhD thesis, University of Nottingham. Artificial Intelligence Fuzzy Expert Systems Risk Assessment Information Fusion Explainable AI Machine Learning Truck Driving |
| spellingShingle | Artificial Intelligence Fuzzy Expert Systems Risk Assessment Information Fusion Explainable AI Machine Learning Truck Driving Mafeni Mase, Jimiama Mosima Context-aware intelligent decisions: online assessment of heavy goods vehicle driving risk |
| title | Context-aware intelligent decisions: online assessment of heavy goods vehicle driving risk |
| title_full | Context-aware intelligent decisions: online assessment of heavy goods vehicle driving risk |
| title_fullStr | Context-aware intelligent decisions: online assessment of heavy goods vehicle driving risk |
| title_full_unstemmed | Context-aware intelligent decisions: online assessment of heavy goods vehicle driving risk |
| title_short | Context-aware intelligent decisions: online assessment of heavy goods vehicle driving risk |
| title_sort | context-aware intelligent decisions: online assessment of heavy goods vehicle driving risk |
| topic | Artificial Intelligence Fuzzy Expert Systems Risk Assessment Information Fusion Explainable AI Machine Learning Truck Driving |
| url | https://eprints.nottingham.ac.uk/73680/ |