Artificial Intelligence and the Built Infrastructure
In a world divided into haves and have-nots, digital literacy should be a given and not just accorded to countries that can afford digital solutions at the expense of those that cannot
With growing urbanisation in the world, the demand for built infrastructure such as roads and bridges is increasing. On top of this buoyant demand, the need to maintain ageing infrastructure is also paramount. How can AI serve the needs of the public to ensure that infrastructure systems as essential as water services are safe and reliable?
Maintaining infrastructure involves handling large amounts of data. The starting point for this lies in gathering on-site data. This often requires arduous and repetitive inspections, sometimes in dangerous and confined spaces. The use of automated sensors allows the gathering of data. For instance, using images creates simplicity, and these images can be taken regularly from different points close to a structure from different angles, removing the need for constant human vigilance. There is also the possibility that these images can be gathered from moving vehicles or drones to replace human effort further. Robot dogs, like one developed by Boston Dynamics, create the potential to remove the need for inspectors to enter confined spaces for inspections.
The combination of images and intelligent algorithms allows for fast bridge damage identification for bridge structures. However, how the image data is collected means that such methods can only identify bridge surface damage. Detecting internal damage to bridge elements remains a challenge. Still, using thermal imaging technology, the heat distribution data of a target object can be converted into an image by measuring the infrared radiation of the object, allowing humans to see beyond the visual barrier to the temperature distribution on the surface of an object. When damage occurs inside a bridge member, the internal voids are often filled with air or water; hence, infrared thermography is an alternative imaging method for detecting them and presents a significantly different thermal image than when there is damage or no damage.
AI’s Application Across Water Systems
The application of AI in road maintenance centres is based on deep learning and image recognition techniques whereby a detailed assessment of a road's current condition can be made through various data sources, including past road surveys. Computational tools process this large volume of data, highlighting areas where defects have occurred and where maintenance is required, providing information on the severity and type of road defects. This assists in improved decision-making and the prioritisation of repairs, ensuring maintenance programs are focused on areas of the road network that would benefit most and return the best value for life cycle performance.
Machine learning algorithms are crucial to this evaluation process because they automate the detection and recording of road defects and provide results clearly and intuitively to all stakeholders.
At the network level, AI paired with sensors can speed up the development of new infrastructure and efficiently manage ageing critical assets. For instance, water network leakage, which currently is at 45 billion litres of potable water per day in developing countries, would benefit from a combination of IoT devices— intelligent toilets, taps, and smart meters— and AI could create a similar impact at a community level. Unnecessary consumption in the agriculture sector and households could be cut.
AI Research and Studies
Research on AI is currently being undertaken at the University of Nottingham, funded by National Highways in the UK, to manage bridge structures without human intervention. The challenge is to detect structural deterioration before it is too late. The key to this is to understand not just the state of a structure at a point in time but how that condition changes over time. National Highways aims to reach a point where no unplanned bridge closures exist. The research already has 25,000 images for use as data training sets for the AI tool and is based on neural networks to classify defects. A neural network is an interconnected group of nodes similar to neurons in a brain. For this project, the network was trained on datasets of images depicting various surface blemishes with corresponding ground truth labels, such as “crack”, “spalling”, and “exposed reinforcement”.
The benefits of AI are apparent as above, but we must be mindful of the ethics of AI. In a world divided into haves and have-nots, digital literacy should be a given and not just accorded to countries that can afford digital solutions at the expense of those that cannot.
AI has been proposed as the latest technological innovation to help address water system deficiencies. However, access to technology is critical to solving water supply and wastewater disposal problems -and this is not always available to developing countries. Over 1.6 million people are dying annually from unsafe and inaccessible drinking water, stormwater and sewerage services.
Lastly, the security of AI must be considered. Breaches in IT systems, such as ransomware attacks, present a significant challenge to cybersecurity. In 2021, a water treatment plant management system in Florida was remotely accessed by an unknown entity that released a large amount of sodium hydroxide into the public water supply, intending to harm people.
In conclusion, AI yields many benefits to humanity and builds infrastructure, and these benefits can only increase as computing power grows exponentially over time. The modernisation of our societies relies on robust and reliable infrastructure. As the ravages of time take their toll on structures, we must be vigilant to avoid future breakdowns. AI, if used wisely, can help us.
