Contractors are moving from traditional maintenance to preventive maintenance. To apply preventive maintenance, up-to-date data is required which is unavailable. The goal is to research the applicability of crowdsensing to detect road pavement distresses, and to investigate how the positions of these detections can be used to gain information on a road pavement section.

The construction sector is currently moving from traditional maintenance contracts, where maintenance was a pay per bill business, to contracts focused on the performance and availability of assets. These types of contracts shift the risk during the maintenance phase from the government to the contractor. The contractor receives a fine if either the minimum performance or the availability is not met. This fine in combination with the risk of unforeseen maintenance costs creates a demand for more efficient maintenance methods like preventive maintenance. However, in the field of road asset management, there is a knowledge gap in how road pavement distresses develop over time. To fill this knowledge gap, up-to-date information is required. The up-to-date information cannot be provided using the current high-quality methods due to the limited availability of equipment.

In this research, raw data is gathered on the state of the road through crowdsourcing with smartphones. This data is then preprocessed by signal analysis, positioning analysis and event detection to retain only the information which describes the occurrence of a road pavement distress. Next, a database analysis consisting of clustering, asset linking and averaging is performed to gather information on the measurements of an instance of road pavement distress. The position of the instance of road pavement distress can be seen as the position of a virtual sensor. A virtual sensor groups events together from multiple smartphones. The location of the virtual sensor is linked to the location of the nearest road pavement asset, allowing asset managers, the people who plan maintenance tasks at contractors, to send their maintenance crew to the correct location.

The results from the experiments show that the detection of road pavement distress is possible by using crowdsensing. The position of the virtual sensor increases in accuracy when more data is used. However, it is possible that two different virtual sensors can be merged due to positioning errors.

It is also possible to connect the position of the virtual sensor to a real world asset location by using the Dutch national road sections and the hectometer posts. The resulting information can be used to get an indication of the status of different assets, giving asset managers insight into which areas and assets they need to focus on.

Professionals in the Operations and Maintenance phase of national road
infrastructure projects are making decisions with large consequences based on low frequency measurements and subjectivity prone expert observations. A solution is expected in sensor innovations, IoT and User Generated Data to generate more frequent measurements and less subjective observations. The use of these methods has been researched and proven, however the main focus of these studies was often improvement of the technological capabilities or implementation in current practises with limited research into their contribution and effects on professionals in the construction sector. This research describes an experiment that tested the effects of more data and more diverse data on the decision-making of professionals in the construction industry. An attempt to answer this question is made by modelling different data sources into a Serious Game and testing the assumptions. After analysis of the gaming data, the questionnaires and the debriefing it can be concluded that in this experiment there was a correlation between better assessments and higher scores. However experts did not assess damage differently when presented with extra information, nor did they make significantly different decisions. From the qualitative section of the experiment the explanation was found that the extra information proved too much and experts were able to extrapolate with the marginal data that represents the current industry practise. This suggests that new information requires training, as our built environment gets richer in terms of data, the assessment of this built environment becomes too much for humans to cope with and solutions can be sought in the application smart algorithms, Machine Learning and Artificial Intelligence.