Indiana students build an AI robot to map the sidewalks their city can't afford to audit

By 14 News (WFIE) · October 1, 2026.

A group of students at the University of Evansville (Indiana) has built a robot equipped with GPS and AI vision capable of identifying curbs, ramps, crosswalks and obstacles on the city's sidewalks. The project, dubbed 'Sidewalks 4 All', is the work of computer science and engineering students and aims to gradually build a complete map of the condition of Evansville's sidewalks, as reported by 14 News (WFIE).

The stated goal is twofold, as Jake Schwaderer, a data science student at the university, explained to the station: it serves as out-of-classroom practice for students and, at the same time, hands the city data it cannot afford to collect on its own. 'It helps us learn things that aren't normally taught here at the school, and it also helps the city, because they get data that would be difficult for them to collect, given that they have very limited resources when it comes to sidewalks,' Schwaderer summed up. The rollout starts in Evansville's Promise Neighborhood and is rounded out by a 'walk audit' mobile app that lets residents themselves report sidewalks in poor condition.

Broadly, the underlying problem is neither new nor unique to Evansville: almost any mid-sized US city has accessibility obligations under the ADA (the Americans with Disabilities Act, the federal law requiring that public spaces be accessible to people with disabilities), but auditing an entire street network on foot —curb by curb, ramp by ramp— is slow, costly manual work that few municipal budgets cover regularly. Academic initiatives such as Project Sidewalk, launched years ago at the University of Washington, had already explored delegating that inventory to volunteers who label Street View images; what changes now is that a student team, on a campus budget rather than a corporate lab's, can build its own street robot with computer vision to try to do it automatically.

That shift matters because it points to where applied AI is actually landing, far from the media spotlight on large language models: in the systematic collection of physical data that until now required city inspectors walking street by street. It is the same pattern already visible in pothole detection with dashcams or in street furniture inventories, and it fits what we have been seeing in our sector-by-sector coverage of employment: administrative work and manual data collection are by far the most exposed to automation, while judgment —deciding what to do with that map, prioritizing which sidewalk gets fixed first— remains human work.

Our reading is that projects like this are the humblest, and at the same time the most representative, face of the underlying case for AI: it does not replace a city's judgment about which sidewalk to repair first, but it removes the cost barrier that today keeps it from even knowing how many of its sidewalks are broken, lack a curb cut or have no wheelchair ramp. In the short term it remains a university pilot for which the source gives no details on funding or large-scale validation: we do not know how many false positives the robot produces, whether Evansville will commit real budget to acting on the map once it exists, or whether the model generalizes well to cracked, leaf-covered or poorly lit sidewalks outside the test environment. But if this kind of cheap, automated mapping spreads to more cities, it reduces one of those small, quiet frictions that today limit independent living for older people or those with reduced mobility, and it fits exactly the abundance we argue for over the long run: resources rationed today by budget, such as inspecting a sidewalk, become routine and almost free, freeing public funds for what genuinely requires human judgment.