In 2021, the United Nations launched its Global Plan for the Decade of Action for Road Safety 2021-2030, with the overarching aim to reduce road deaths and serious injuries by 50 percent by 2030.
Recent research undertaken by John Hopkins University showed that using iRAP methodology, road safety infrastructure changes and safer speeds have prevented almost 700,000 deaths and serious injuries in 74 countries since 2016.
Driving Change, the Road Safety Strategy for Western Australia 2020-2030 has set an ambitious road safety target to reduce the number of people killed or seriously injured by 50 to 70 per cent by 2030. In line with Australia’s Vision Zero, Main Roads Western Australia’s vision is that death and life-changing injury on the roads will be eliminated by 2050.
Part of the strategy to achieve this target is to generate AusRAP Star Ratings for roads that carry 80% of traffic by 2025. AusRAP star ratings are measured on a scale from 1 to 5 stars and provide a clear indication of a road's safety performance, with the level of risk halving with each additional star. A 5-star rating represents the highest level of safety, while a 1-star rating indicates a higher risk.
AusRAP is based on iRAP Star Ratings which have been adopted globally by United Nations and its member states and provide an objective and proactive way to assess a road’s safety performance based on the physical characteristics of the road environment, as well as road user flows and speeds.
Historically, iRAP Star Ratings have been undertaken by an individual manually coding video and street images in accordance with iRAP specifications. With this method, identification of and ranking of safety attributes is based on the judgement and expertise of the person doing the coding with the coder not the process being accredited by iRAP.
In 2019, iRAP announced the introduction of AiRAP (Accelerated and Intelligent Road Assessment Program) for data collection, capture and coding. AiRAP coding seeks to use a wide range of data including existing stores of readily available commercial and open-source data (i.e. LiDAR, video and satellite data) and automated data analytics techniques including machine learning and artificial intelligence to deliver ‘accelerated and intelligent’ RAP data for Australia and the world to use.
The implementation of AiRAP, which is focusses on utilising accredited processes rather than individuals in identifying road safety attributes, has the potential to reduce error, time and effort required to undertake reliable, repeatable, and scalable road safety assessments.
Main Roads and Anditi recognise this potential and have both been key players globally in the development and implementation of AiRAP. Anditi’s RoadViewer technology is currently the only AiRAP accredited inspection system globally, leading to Main Roads appointing Anditi to execute this project.
This project included generating attributes to be used as part of AiRAP coding for Star Rating of approximately 19,164 km of State Road network. Star Rating attributes were generated using a range of data sources including mobile LiDAR and 360-degree imagery that were analysed using Accelerated and Intelligent automated feature extraction technologies.
The locations of the roads that were assessed are shown on Figure 1.
AiRAP assessment of the road network entails:
• Accessing, analysing and indexing data from a range of data sources for ~20,000 kms of state and main roads. Data sources used included mobile LiDAR and 360-degree imagery, satellite and aerial imagery, shapefiles depicting school zones, land use, area type, railway crossings.
• Dividing the road into approximately 210,000 road segments with the location of the start and finish of each nominal 100 m long segment being linked via an XML file to Main Roads IRIS data base.
• Detecting, recording and tracking for each 100 m segment the 78 road safety attributes and approximately 330 attribute sub-categories that are required for Star Rating the road network.
For the Project approximately 69.3 million attributes and attribute sub-categories were identified, recorded and tracked to enable Star Ratings to be generated for the road network. Examples of these include Road Severity Object and Distance, Intersections, roundabouts and rotaries, Road Condition and Skid Resistance assessment, Sight Distance assessment, and Sidewalk, Bicycle facilities and Shared Pathways
The Project used a range of remotes sensing and other data sets. This included mobile LiDAR and 360-degree imagery of the State Road network that has been captured as part of the Network LiDAR Project (NLP).
The data was captured by the following surveying & mapping companies: BCE Surveying, Jacobs, Land Surveys, MNG, Veris, TomTom. An example of the coloured LiDAR point cloud generated from the NLP surveys using mobile LiDAR and imagery is shown on Figure 2.
Mobile LiDAR is not well suited for use on unsealed roads due to dust generated by the capture vehicle or other vehicles using the road network potentially impacting on detection of the road surface attribute and road corridor assets and features.
Unsealed roads were captured using a helicopter fitted a LiDAR scanner and cameras to provided aerial imagery. An example of the coloured point cloud generated from helicopter capture is shown on Figure 3.
Remote sensing based AiRAP coding enables the use of a wide range of data sources to supplement the attribute and asset information extracted from mobile LiDAR and 360-degree imagery. This includes:
Anditi’s RoadViewer Inspection System was the first system globally to be AiRAP accredited to use video/imagery and LiDAR. This system utilises input data for coding from a range of sources including LiDAR, 360-degree imagery, aerial/satellite imagery, shapefiles, open-source databases and GIS layers.
These technologies utilise a range of Signal Processing, Optimisation and Artificial Intelligence analytical techniques including:
As part of its RoadViewer technology, Anditi has developed a suite of analytical tools specifically designed for the automated and semi-automated extraction of road assets and road safety attributes from a range of remote sensing data sources (Figure 4).
The workflows involved in this project include:
• Quality control of input data - checks the completeness, consistency and quality of the input data used for the project
• iRAP segment generation - automatically generates the approximately 100 m long road segments as provided from Main Roads IRIS database that were used for AiRAP coding
• Data organisation and tracking - organises the MLS data for every road segment so that the automated algorithms and processes can be utilised to analyse the data. This process is also used to track which road segments and attributes have been coded for each road segment.
• Feature and attribute identification - processes the input data sources using the following workflows:
o Automated and semi-automated extraction and characterisation of attributes from satellite and aerial imagery such as Area Type, Property Access and Service Roads
o Automated identification and extraction of attribute and attribute locations from aerial LiDAR, mobile LiDAR and 360-degree imagery (Figure 5).
o Semi-automated and manual coding of attributes that have been flagged as not being able to be reliably detected using the input data available. This occurs where features may be screened (i.e. grass growing around a wire safety barrier) or where there is not enough data for the feature to be detected (i.e. where the sealed road edge is covered in dirt). The QA process uses Anditi’s desktop RoadCoder QA tool (see Figure 6) which enables the people undertaking the Quality Control to access a range of data sources and jump from one flagged attribute location to another.
As part of the NLP project, Main Roads has effectively generated a contemporary, high resolution 3D point cloud model of the state road network that is readily accessible to a broad range of State users via a web portal. This 3D model contains a lot of information that can be used for many purposes in addition to iRAP Star Rating. Information available from the 3D model for the road corridor includes:
An example of output able to be generated from NLP data is shown on Figure 8.
The range of uses of the data generated as part of this project and the Network LiDAR Project, significantly add to the overall value of this data. These include:
Improvements could include accurate species-specific digital trees and analysis of shading throughout growth stages. Scaling could use aerial LiDAR data to prioritise planting areas with low shading.
This project which assessed almost 20,000 km's of national highways and state road network across 10 regions of Western Australia, demonstrates Safety Rating AiRAP coding can successfully be undertaken at scale using Accelerated and Intelligent analytic techniques to extract road safety attributes from a combination of remote sensing data sources. These include mobile LiDAR and 360-degree imagery, satellite and aerial imagery, aerial LiDAR and a range of database and GIS data sources.
AusRAP and iRAP have been actively involved in providing guidance and clarity in regard to AiRAP over the course of this Project and have provided the following comments:
Main Roads WA Network LiDAR Project has clearly demonstrated the future of road assessments using AiRAP methods. The partnership between Anditi and MRWA not only delivered the planned road assessments but provided new and more efficient business opportunities for MRWA.
The NLP has established innovative procedures and methodologies that set a clear path for other AusRAP members and indeed, the rest of the world to follow. Both MRWA and Anditi are to be congratulated on this game changing result.
As a charity, iRAP’s vision is for a world free of high-risk roads for all road users. AiRAP is an exciting global partnership that will help transform the scale and frequency of Star Rating assessments worldwide that will ultimately help road authorities, and their partners save lives. Main Roads Western Australia is a global leader in this regard, undertaking the world’s largest AiRAP assessment as part of the Austroads-led AusRAP programme and leveraging the full potential of the Network LiDAR Project to deliver benefits across the whole organisation.
As with any innovative project you need a team of experts who lead the way, explore the new horizons and have the tenacity and dedication to deliver. Anditi is truly the global champion of AiRAP through their exploration of the source data, technology, processes and quality controls needed to make AiRAP a success. This world leading project is a true game-changer that sets the benchmark for the future. A future that will fulfil iRAP’s dream to Star Rate every road on earth and help deliver the life-saving social, community, financial and economic benefits of 3-star or better roads worldwide.
.svg){kind=small}
.svg)
.svg)
