New Police AI Could Help Catch Fugitives Faster and Cheaper

When police try to catch someone running through a city, the problem isn't just manpower — it's timing. Roads fill up, suspects move, and officers can't be everywhere at once. A team of researchers from Japan and India has built a computer tool that plans these chases much the way a chess engine plans a game. It's called MLN-EIGS, and it works by mapping a city's roads hour by hour so it can see how both the fleeing person and the officers move over time.

The tool belongs to a family of methods called Stackelberg games, which simply means one side moves first and the other side responds. Here, police place their checkpoints first, and the suspect chooses a route knowing that. The software then searches for the best possible arrangement — the one that gives officers the highest chance of catching the person, given that the suspect will try to slip through the gaps. Plainly put: it's a smart way to cover the exits without wasting officers on streets nobody will use.

The clever part is speed. Calculating the suspect's best escape route could be painfully slow, so the team used a mathematical trick to turn it into a simple shortest-path search — the same kind of math your phone's maps app uses to find the fastest route. They tested their method against a slower, exact approach on a large real-world transportation network. The new tool matched that approach's results closely while taking far less computer time, which is what makes it practical for a real operations room rather than an overnight calculation.

The catch: this is still a research paper, tested in simulations rather than live policing, and better prediction tools raise real fairness questions — including whether a system like this would spread police unevenly across neighborhoods. Its biggest near-term value may be for planners and emergency responders rather than street-level officers.