The manual road condition survey method used in New Zealand (road asset maintenance management ((RAMM)) surveys) was developed during the early 1980s with the primary purpose of feeding into the treatment selection algorithm. For more than 20 years the rating system was adequate for this purpose but as more sophisticated asset management evolved into deterioration modelling and advanced trend monitoring, the data quality from the manual surveys came under scrutiny. Attempts to improve the robustness of the rating system included increasing the recommended sampling size from 10% to 20% of the treatment length plus increasing the requirements for accreditation during the training of raters. Yet, these steps still fall short in increasing the overall usability and repeatability of rated data for the new demands of asset management processes. Automated defect data collection has been undertaken since the mid-1990s with early technology relying on photographic imaging and processing of road surface data. The technology was particularly popular for application on busy asphalt and concrete motorways in the northern hemisphere but failed to deliver acceptable robustness on chipseal surfaces. This situation changed with the arrival of laser scanning technology, which has overcome the limitations of photo-imaging technology. The measurements now solely depend on laser scanning at a high resolution, which gives a comprehensive 3D image of the road profile. Any defects such as cracks, potholes or surface defects can be identified on the image. The benefits this technology offers to the sector include: • surveys of 100% of the road are possible • all aspects of the condition of the surface are captured simultaneously • the measurements take place at high speed (60 to 80km/h), providing significant safety and traffic management benefits • ‘removing’ the human element from the measuring allows for more repeat measurements. Despite the accuracy of the measurement, the constraining factor for the technology is the algorithms that interpret the digital image to identify and quantify specific defects. This has resulted in the main question posed for this project — is the measurement sufficiently robust and is the sector ready to adopt this technology on a wide scale? The project’s aim was to focus mainly on the impact assessment with the assumption that the three defect types (cracking, ravelling and shoving) are being accurately collected using the laser scanners. A number of international research projects have confirmed the accuracy of the measurements. During the research, more work was required to validate the results and introduce a new algorithm for the detection of shoving. The ultimate focus of the research changed slightly to how ready is the technology in its current state for wider adoption in New Zealand. The research findings were consistent with most international research projects that have studied the accuracy and repeatability of the laser technology. These findings were: • The laser technology’s measurements are accurate and can be repeated — the benefit of having a 100% road length covered by the surveys is particularly attractive for the intended data use. • The comparison of the laser technology with existing practices remains a challenge and the results from such comparisons should be analysed with care. The comparison between laser scanning and RAMM surveys will never yield ideal results because: — The two assessment methods differ fundamentally in the way they define the extent of the defect — one-to-one comparisons are therefore not possible. — RAMM surveys only cover a percentile of inspection lengths and a sample outcome will most likely differ from the full-length survey. — Ensuring the two assessment types reference the exact same location is difficult and there is a lack of confidence that comparisons are being made between the same road sections. • The laser technology has identified defects successfully but has also identified a number of false positives. A more detailed investigation during this research into the shoving measurements has identified a number of road features that appear to trigger shoving according to the defined algorithm, but in reality identify a completely different road feature as a shove. The study has also confirmed a number of instances where the rating simply ‘missed the shove’ as it was not very apparent for a number of reasons. The implementation plan • Using laser scanning for detecting road defects should be adopted by all road agencies. This recommendation is made on the basis of the significant benefits that can be realised from: • more accurate assessment • better repeatability between surveys from consecutive years • greater coverage of the road network, i.e. more roads are being surveyed for 100% of the length. The laser technology, despite its accuracy, cannot be applied as a 100% automated process. The computer algorithms that analyse the data still need significant ‘learning’ that can only be achieved if the technology is supplemented by manual validation of the outcome. Someone needs to work through the digital images to find erroneous identifications and feed this knowledge back to the algorithms. Once this is completed, business as normal survey contracts should include calibration procedures, validation and quality assurance protocols. (Author/publisher)
Abstract