Disruptive technologies like RoadAI can reduce manual labour tasks and make data collection more efficient and accurate.
Disruptive technologies present opportunities to make laborious manual tasks faster and easier to perform. From Artificial Intelligence (AI) to blockchain, these technologies can be deployed in many practical applications, freeing up employees to focus on the important tasks, minimising human error and streamlining decision making processes. One such technology for pavement management is called computer vision. Vaisala, a Finnish company established in 1936 making the first radiosondes, is now keeping ahead of the game through the utilisation of computer vision technology. Vaisala’s enhanced capability RoadAI was launched early in June 2019, collecting data on the state of road surfaces in order to support transportation decision makers in maintaining roads in their communities.
The technology scans all areas of the roads that may require maintenance attention and can accomplish multiple scannings to the same route year round. The data collected is delivered in a simple, easy to understand map or files as actionable intelligence for decision makers. This saves time and money for decision makers, as the platform is up to four times faster and half the price of traditional pavement condition analysis. RoadAI supports road surface state data collection year round, from cracks in the pavement in the summer to ice covered roads in the winter (when paired with MD30 sensors).
We spoke with Markus Melander, head of business development and computer vision research and development at Vaisala, to discuss this new technology and its contribution to improved road maintenance efficiency.
How does this computer technology work and how can it contribute to improved road maintenance efficiency?
RoadAI is a system that enables us to collect information in video format and process it with an algorithm. It provides a user interface to browse and export the data to be used in other existing asset and payment management systems. Computer vision is the most prominent branch of Artificial Intelligence at the moment because it is easy for us to understand how it can replace or assist the human in tiring and repetitive work. For example, pavement condition inventories or analysing thousands of kilometres of road – these tasks can be intensive.
The computer analyses each individual frame in a video, working like the human eye. Tens of thousands of images have been carefully selected in order to make the computer the best kind of road inspector. Transportation assets and roads are owned mainly by governmental bodies – counties, municipalities and state organisations – these are the natural major beneficiaries. Organisations which work for these entities like contractors, consultants and road inspector companies can also benefit very quickly. Asset owners can cut their costs and keep roads in better conditions as they can analyse the road conditions better and faster – making everything more effective and cost efficient.
Where is the data collected by RoadAI stored and how can it be accessed?
The video data that is recorded and manage it for the customer – hosting the data and enabling the computer vision to work for them. If Transport England would like to have its own server, for example, we could provide it. The key question for the end customer is that if they want to have their own hosted system, it is more expensive. It is much easier to manage a centralised system, but we can also set it up so customers have their own hardware infrastructure as well.
Does visual data collection raise privacy concerns for people on the roads?
This system is really focusing on pavement and road sign data capture and not on collecting data in other ways. When it comes to congestion and other similar issues, we are not aiming to provide information. Anonymisation of visual data is executed automatically to make the system 100% aligned with GDPR by masking people and vehicles with colour coating. Anonymisation has proven to be so powerful at protecting privacy that people in the automotive industry are using this piece of RoadAI where video data from roads is needed.
What are the benefits of visual data for road maintenance?
Automated pavement condition data is naturally the core of the system but storing visual data has also proven to be an important and valued part of RoadAI. The existence of up-to-date road views is important because it enables validation of computer vision results and helps in detailed planning.
It creates a kind of showroom and when you zoom into the map it colour codes the network in a pavement condition index. The further in you zoom, the more detailed and accurate the data becomes. The different types of damage are colour coded, so this can speed up understanding of problems and decision making.
How does RoadAI technology compare with other data collecting technologies such as MD30 sensors?
Every system has its strengths and weaknesses and you need to pick the tool to meet your demands – computer vision replaces manual visual inventory. If you want to, for example, add the magnitude of rutting on the road, this is something that you do not inspect visually. For this you need a laser, so you need different technologies for different roles. This application of the technology is not a silver bullet, but it is a fast, simple and easy way to get accurate information that is visually identifiable on the network.
What is the importance of data collection for road maintenance and how can it help us build new roads and infrastructure in the future?
The data in pavement management specifically is so essential – if you don’t have information, how do you know where to invest? According to the International Road Federation more than €80m is spent on road maintenance in Europe every year, which includes building new roads and maintaining old ones. On average hundreds of millions of euros are invested in paying to fix roads and keep them in good condition. The planning process, fixing the roads, which methodologies to use, to do the intervention, and so forth, all of these steps are becoming much easier.
Looking at this from the technology perspective this is just one application of the computer vision technology, and now the same kind of technologies are being taking into new industry areas. The sports car industry is very interested in these kinds of issues, as sports cars are hard to drive and you can feel everything underneath – they are high maintenance and it is not nice to drive them if the road is very jumpy.
The early adaptors who take up this technology are the ones who will be followed. In the near future, we are going to see more maintenance organisations using RoadAI and similar technologies to enhance their operations. Currently, RoadAI technology is being used in multiple European countries. In Finland and Norway, commercial trucks and other vehicles are providing computer vision data collection for maintenance operations. Because of its ease in working with established fleets and staff, some regions of the UK have piloted RoadAI and are actively working to integrate it into their processes to make their work more efficient and effective.
Markus Melander
Head of Business Development and Computer Vision Research and Development
Vaisala
