MIT solved one of hardest problems in underwater robotics AI that works with human divers

Added: 2026-05-05

[00:00:00]MIT just solved one of the hardest problems in underwater robotics by building an AI that doesn't replace the human diver, but actually works with one. Meet Maline Miller, who is developing hardware and algorithms that let autonomous underwater vehicles and human divers operate as a team on real military missions. And that's where underwater AI stopped being a solo operation. Divers have dexterity and object recognition. AUVs have processing power, speed, and endurance. The problem is underwater radio doesn't work. GPS doesn't reach and sound communication is extremely limited. Miller's team built an AI classifier that processes sonar and optical data mid-m mission. When it detects something with low confidence, it sends compressed data to the diver asking for confirmation. The diver corrects it. The system updates. They tested this in the Atlantic Ocean and the Great Lakes with actual divers swimming alongside the vehicle. Here's where it gets insane. The modern world runs on undersea cables. Repairs today require large-scale inspection operations involving ships and remotely operated vehicles. This system is designed to let a robot map the problem and guide a diver to the location that needs attention. But the craziest part, we spent decades trying to replace humans underwater with robots. MIT just figured out the answer was never replacement. It was teameing. If you want to stay updated with the latest tech news, follow for