Gravel roads play a vital role in transportation, particularly in rural and remote regions. They support essential sectors such as agriculture, forestry, tourism, and emergency services and can serve as access routes to remote natural resources or large onshore windmill farms. However, they deteriorate rapidly and rely on subjective, inconsistent and labour-intensive visual and manual condition assessment, often resulting in delayed maintenance interventions and costly maintenance actions. This dissertation addresses the need for improved condition assessment of gravel roads by integrating participatory approaches with objective data-driven methods. Overcoming the limitations of traditional assessment practices would enable broader network coverage, enhance the efficiency and effectiveness of defect detection, and support sustainable maintenance planning and informed decision-making.

The research is guided by Design Research Methodology, synthesising the six appended interrelated studies. These studies progress from problem identification and conceptual development to a case study and empirical field experiments to answer the three research questions (RQs): (1) How can objective data-driven approaches address the limitations of traditional approaches to enable efficient and effective assessment of gravel road conditions? (2) How can data-driven approaches and multi-source information enable reliable assessment of gravel road conditions and support maintenance planning and decision-making? (3) How can participatory and adaptive gravel road condition data collection enable sustainable condition assessment to support predictive and data-driven maintenance planning and decision-making?

Vehicle vibration response data (VVR) collected using Integrated Electronics Piezo-Electric accelerometers and smartphone sensors are analysed using statistical, spectral, and subband filtering techniques. The International Roughness Index (IRI) measurements, as well as videos and images of the gravel road condition, are also collected from Roadroid, an Android-based smartphone application, to complement the VVR and verify the actual defects on the gravel roads. A rule-based classification approach is developed and empirically validated using the collected field experiment data, enabling the detection of surface defects such as potholes, corrugations, and loose gravel. Multi-sensor integration and the sequential probability ratio test further enhance the robustness of the objective fault detection and condition classification for gravel roads. The findings from the Roadroid IRI measurements led to the refining of IRI thresholds for a four-class condition classification system.

The research also proposes a participatory and adaptive approach to collecting condition data. By involving regular gravel road users in collecting gravel road condition data, the proposed participatory data-driven approach to assessing gravel road conditions would increase the spatial coverage of the assessed gravel roads, which are mainly located in remote areas. At the same time, adaptive sampling optimises the data that is captured and stored without compromising accuracy, thus enabling more reliable, objective and sustainable gravel road condition assessments that would support predictive maintenance planning.

In conclusion, the dissertation demonstrates how objective, data-driven methods and the participatory approach can complement and gradually replace traditional subjective assessments, providing gravel road associations, municipalities, and road agencies with condition assessment approaches that enable broader assessment coverage, are efficient and effective, and are socially inclusive. By combining multi-source sensor data with stakeholder engagement, the research lays the foundation for sustainable, predictive, proactive and resilient gravel road management systems. It ultimately contributes to theory, method and practice in gravel road asset management.