Public infrastructure in the U.S. could use improvement
Roads riddled with potholes, cracking bridges and leaking pipes are so common that many Americans don’t think twice about them. AI-driven infrastructure repair could be the solution to these persistent problems.
The crumbling infrastructure issue is more severe than many people may realize. A 2017 study found that 39% of American bridges are more than 50 years old, despite not being designed for such long lifespans. Almost 10% are structurally deficient, yet there are 188 million trips across them every day.
Similarly, the U.S. Department of Transportation (DOT) estimates that more than two-thirds of roads in the nation are in “dire need” of repair. One of the reasons this problem is so out of hand is because infrastructure repair is expensive and challenging. Here’s how AI could change that.
Finding Areas in Need of Repair
Since such a significant portion of the nation’s infrastructure needs maintenance, finding problem areas isn’t always straightforward. In a sea of potholes, a dangerously damaged road won’t stick out as much as one might think. AI technology can find and highlight these areas so crews can start planning their repair.
Drones can fly over roads, buildings and bridges, gathering footage of potentially damaged infrastructure far faster than an in-person survey. Machine learning software can then analyze this footage and point out areas that need maintenance. These automated surveys save local DOTs and other authorities hours of time in the planning phase, in turn saving money.
Connected cars can gather data about crumbling roads as people drive them in their daily commutes. Using this data to highlight maintenance issues doesn’t require much action beyond what people are already doing.
Prioritizing Infrastructure Repair Needs
Infrastructure repair is an expensive process. The American Society of Civil Engineers estimates it would take $4.6 trillion to repair all of the nation’s infrastructure sufficiently. The country can’t afford to spend all of that at once, so it needs to determine which projects are the most pressing or cost-effective.
RoadBotics, a startup from Carnegie Mellon, uses AI to analyze road conditions and prioritize which sections should see repair first. The AI categorizes sections of the road into green, yellow, and red areas, depending on how deteriorated they are. It then emphasizes the yellow, moderately damaged sections since repairing these is the most cost-effective long-term solution.
It’s not always clear which repairs will yield the best results. AI is much better than people at considering multiple factors and predicting outcomes, so it’s better suited to these tasks. By using AI to prioritize repair projects, local governments can build better infrastructure for less money.
Optimizing Infrastructure Repair Operations
Construction and maintenance projects can be disruptive. Sediment runoff from these sites can be 10 to 20 times higher than agricultural land, leading to erosion, which further harms nearby infrastructure. AI algorithms can help these projects minimize their impact, ensuring they do more good than harm.
AI software can analyze a worksite and predict the impact that various actions and methods will have. These analytics can then inform what the best, least disruptive course of action is. Repair crews can use this information to ensure that they don’t create the potential for future problems in their current work.
Similar algorithms can also help keep costs to a minimum. Machine learning programs can predict how many materials a project will need or which materials would be the most cost-efficient. Teams can then prevent waste and perform the most cost-effective repairs.
Predicting Future Maintenance Needs
As local authorities use AI to repair public infrastructure, they have an opportunity to go a step further. Introducing AI technology into structures themselves can help optimize maintenance and repair efforts in the future. AI-driven infrastructure like this is already starting to pop up across the country, and it’s showing promising results.
Fracta Inc., a startup in Redwood, California, uses AI software and smart sensors to detect weak points in the city’s water systems. This solution saves construction companies between 30 and 40% of their long-term repair costs, as it pinpoints problem areas before teams break ground. This improved accuracy also shortens the time it takes to perform maintenance.
Similarly, Purdue SMART Lab has started placing sensors in concrete to monitor wear and tear in roads. These sensors convert vibrations into electrical signals that communicate how a road is deteriorating. AI can then take this data and predict when a section will need maintenance, reducing related time and expenses.
AI-Driven Infrastructure Reduces Costs and Improves Safety
Better infrastructure will improve safety, enable more efficient travel and raise the standard of living. AI-driven infrastructure repair can help achieve these goals with minimal cost and disruption.
The shift to AI-driven infrastructure will be a slow but crucial change. If DOTs and other authorities across the country can embrace these technologies, the U.S. can start to fix its persistent repair problem.
Emily Newton is the Editor-in-Chief of Revolutionized, a magazine exploring how innovations change our world. She has over 3 years experience writing articles in the industrial and tech sectors.