UT Dallas AI Team Helps NEXCO Build Pothole Prioritization System

Artificial intelligence is now being deployed to answer a question that has frustrated city managers for decades: which pothole gets fixed first? Researchers from The University of Texas at Dallas have partnered with NEXCO-Central, a Japanese infrastructure company, to build an automated decision-making system for municipal road repair prioritization.

The collaboration centers on the Center for Applied AI & Machine Learning (CAIML) at UT Dallas. The team worked with NEXCO Highway Solutions of America Inc., the company's Irving, Texas-based subsidiary, to advance existing pavement assessment technology. The system combines AI algorithms with video footage captured from mobile cameras mounted on vehicles. This creates a networkwide view of road conditions across municipal infrastructure.

Dr. Gopal Gupta, CAIML director and professor of computer science in the Erik Jonsson School of Engineering and Computer Science, described the core function of the new system. "The new system emulates the mind of a city manager who has to decide the priority for fixing various road segments," Gupta said in the official UT Dallas announcement.

That's a fair description, but it glosses over the actual mechanics. The software doesn't just identify damage—it assigns scores to road segments based on multiple variables. These include pavement condition, traffic volume, budget constraints, and competing repair needs. The output is a ranked list that tells municipal engineers where to deploy crews first.

From a user perspective, this changes the workflow significantly. Instead of city staff manually reviewing thousands of road segments or relying on citizen complaints (which tend to cluster around vocal neighborhoods), the system processes visual data continuously. Mobile cameras capture footage during regular patrols. The AI analyzes the video for cracks, potholes, and surface degradation. The scoring system then integrates this data with budget parameters.

Koshiro Mori, a developer at NEXCO-Central, emphasized the financial optimization aspect. "Our technology aims to optimize the complex decision-making to determine which roads are most in need of repairs, the predicted financial investment and prioritizing who gets the money and when." The tool also explains the factors behind each recommendation, which is critical for transparency when public funds are involved.

The project involved UT Dallas computer science doctoral students Abhiramon Rajasekharan and Keegan Kimbrell, who contributed to the technical development. The collaboration was structured through an Intellectual Property Assignment/Sponsored Research Agreement. This arrangement allows NEXCO to retain ownership of the resulting intellectual property—a common model for university-industry partnerships.

Atsushi Onishi, vice president of NEXCO Highway Solutions of America, noted the budget constraints that drive this technology. "It is important to have the technologies to determine which segment has to be done within the budget and how much should be spent on specific road types." Municipal budgets are finite. Cities cannot fix every damaged road simultaneously. The AI system forces explicit tradeoffs.

NEXCO-Central initially researched academic institutions online and identified CAIML as a potential partner. "AI and machine learning are the core technologies that we use in our business," Mori said. "We saw a collaboration opportunity, and we're very happy with how the team has handled this project." The company serves clients mainly in the North Texas area through its Irving subsidiary.

CAIML's mission extends beyond this single project. The center applies artificial intelligence and machine learning technologies to solve problems for industry partners. It houses more than a dozen researchers with expertise in deep learning, computer vision, automated reasoning, natural language processing, constraint optimization, and statistical relational learning. Most projects result in the partner companies owning the intellectual property.

Gupta positioned the collaboration as a model for university-industry relationships. "We think of ourselves as the research and development center for companies that do not have an R&D arm," he said. "NEXCO collaborated with us in this way and created a phenomenal product." This framing highlights a growing trend: universities filling the R&D gap for companies that lack internal research capacity.

The physical reality of this technology matters. City crews still drive the roads. They still pour asphalt. They still deal with traffic delays and angry commuters. What changes is the decision-making layer. Instead of reactive repairs based on complaints or scheduled inspections, the system enables proactive prioritization. The cameras run continuously. The AI processes the footage. The scoring system updates as conditions change.

There's a practical benefit here that goes beyond efficiency. The system creates an audit trail. When a city decides to repair Road A instead of Road B, the AI's scoring rationale is documented. This matters for accountability. Taxpayers can see why certain roads received funding while others did not. The tool explains the factors behind each recommendation, which helps defend budget decisions.

The Office of Technology Commercialization in UT Dallas's Office of Research and Innovation assists researchers and external partners in bringing innovations to market. This project represents one such commercialization effort. The resulting technology has been integrated into NEXCO's software suite and is now available to municipal clients.

Whether this actually saves money remains to be seen. The system optimizes allocation, but it doesn't reduce the underlying cost of road repair. Asphalt prices fluctuate. Labor costs rise. Weather damages infrastructure faster than ever in some regions. The AI can prioritize, but it cannot fix the fundamental economics of municipal infrastructure maintenance.

That said, the technology addresses a real bottleneck. City managers face impossible choices daily. Limited budgets, competing priorities, and aging infrastructure create a perfect storm. An automated scoring system doesn't eliminate these problems, but it does make the decision process more transparent and data-driven. That's worth something, even if it doesn't solve everything.

The project was announced on April 29, 2026, according to UT Dallas media relations. Whether other cities adopt this technology depends on cost, integration complexity, and whether municipal budgets can absorb the investment. The AI doesn't fix potholes. It just tells you which ones to fix first. Whether that's enough to satisfy taxpayers is another question entirely.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn