The search for the nuclear submarine USS Scorpion, which tragically disappeared on May 22, 1968, along with 99 sailors aboard, is finally bearing fruit. The US Navy turned to John Cravem, a Navy Special Projects Officer, for the search. Cravem applies Bayesian statistical techniques. He begins by interviewing experienced sailors and running simulations (domain knowledge) to establish a good baseline (a prior probability). This probability map of the USS Scorpion’s location (across 4,000 kilometers of Atlantic Ocean!) is then gradually updated based on the research. The Bayesian technique bears fruit within weeks, after months of unsuccessful attempts. The USS Mizar photographs the Scorpion 3,000 meters below the surface. The cause identified by the Navy is either an accidental missile explosion or the malfunction of a waste disposal unit. But one thing is certain: the Bayesian search worked brilliantly. And it will be used countless times in the future, for example for the real-time updating of the location and mapping of the environment in modern self-driving cars, aka SLAM (Simultaneous Localization And Mapping).
