Imagine a hiker lost in the wild mountains of Scotland. Search and rescue teams need to find them fast, but the area is huge, and they don’t have enough people or time to check everywhere. That’s where a smart computer program, called an AI, comes in. This AI, named J2, acts like a super detective to guess where the lost hiker might be. Let’s break down how it works in simple terms.
J2 is like a video game where a pretend hiker, called an “agent,” wanders through a digital map of the real world. This map includes details like rivers, paths, forests, and hills, based on real places. The AI doesn’t just guess randomly—it uses clues from past rescue missions to figure out how hikers behave when they’re lost. For example, some hikers head toward water, others look for paths, and some try to find buildings or trees for shelter. J2 creates four types of pretend hikers, each following one of these habits, like choosing a path or heading to a river.
To make its guesses, J2 runs thousands of these pretend hikes, starting from where the hiker was last seen. It uses a math trick called Monte Carlo to try lots of different routes, like rolling dice to see where the hiker might go. The AI also thinks about how tired a hiker gets, using data to decide how far they could walk in a few hours. By combining all these pretend journeys, J2 makes a map showing the most likely spots to find the lost person. This map helps rescue teams focus on the best areas first.
The cool part? J2 doesn’t need tons of information to work. Even if there’s only a little data about past rescues, it uses a clever tool called a Gaussian Process to fill in the gaps. It’s like drawing a smooth picture from a few dots, helping the AI make smart guesses even in new places without needing to learn everything from scratch.
J2 was tested on the Isle of Arran, a rugged island in Scotland. It made maps that matched real rescue data pretty well, especially for open fields and roads. While it’s not perfect yet—sometimes it struggles with forests or buildings—it’s a big step toward saving lives faster.
You can read the research paper here.
Business Applications
This AI isn’t just for rescues—it could help businesses too! Companies that deliver packages could use J2’s mapping skills to plan the fastest routes through busy cities or remote areas, saving time and gas. Stores could predict where customers are most likely to go in a mall, helping them place ads or products smarter. Even farmers could adapt J2 to track livestock across fields, using its ability to guess movement patterns. By tweaking the AI to fit their needs, businesses could work faster, cut costs, and make customers happier.