You can't spot a hedgehog from space, but you might be able to find where they live by looking for brambles. That's the premise behind ongoing research at the University of Cambridge, where scientists are using satellite imagery and AI models to map potential hedgehog habitats across the UK by first identifying their favorite hiding spots: bramble patches. European hedgehog populations have declined by roughly 30 to 50 percent over the past decade, so tracking these nocturnal creatures across large areas remains difficult and expensive. Rather than searching for the hedgehogs directly, researcher Gabriel Mahler developed an AI model that identifies brambles, which are thorny shrubs that hedgehogs use for shelter and foraging, from satellite data. These small mammals rely on this type of dense vegetation for daytime shelter, nesting sites, and protection from predators. Brambles also attract insects and provide berries, supporting the invertebrate populations that hedgehogs eat. Traditional hedgehog surveys require extensive nighttime fieldwork, specialized equipment, or citizen scientists reporting sightings. Those are methods that don't scale well for national conservation planning. By contrast, satellite imagery covers vast areas continuously, and if AI models could reliably identify key habitat features like brambles, conservationists might gain a powerful tool for large-scale habitat assessment. From satellites to shrubs While AI is a popular buzzword these days, it's worth noting that the Cambridge team's detector is not based on a large language model like ChatGPT. Instead, the model uses relatively simple machine-learning techniques: a combination of logistic regression and k-nearest neighbors classification. Mahler's bramble detector also combines TESSERA earth representation embeddings, which process imagery from the European Space Agency's Sentinel satellites, with ground-truth observations from iNaturalist, a citizen science platform. But does it actually work? To find out, Mahler and colleagues Sadiq Jaffer, Anil Madhavapeddy, and Shane Weisz spent a day walking around Cambridge with smartphones and GPS devices, checking whether the model's predictions matched reality.