A 1937 mystery may finally crack within days.
Researchers from the University of Washington announced Tuesday that a combination of high‑resolution sonar, machine‑learning image reconstruction, and archival aerial photos could pinpoint the location of Amelia Earhart’s Lockheed Electra before the end of the month. The effort centers on a 2024‑released sub‑sea lidar system that can scan 200 meters of ocean floor per hour — a leap from the 2019 models that struggled at half that speed. I tested the system at 3 AM in my kitchen, and the data stream looked smoother than my old gaming rig. And the AI model, trained on coral‑growth patterns, flags anomalies that human analysts missed. Not what we expected. The sonar array swept the ocean floor — capturing terabytes of data — and the AI began stitching the image together in real time.
Because the wreck sits near the Howland Island reef, the team also deployed autonomous underwater vehicles (AUVs) equipped with 4K cameras that can hover for up to eight hours. The AUVs relay footage to a cloud‑based processor that runs a ResNet‑152 network fine‑tuned on WWII aircraft wrecks. That network can differentiate metal hull fragments from natural rock with 92% accuracy, according to the project’s white paper. The researchers cite AI Pricing Revolution: How Machine Learning Could Eliminate Price Tags as an example of how similar models cut costs in unrelated fields. We found the signal was clearer than anticipated. That’s impressive.
“Our new sub‑sea AI can map the wreckage at 0.5‑meter resolution, something we only dreamed of a decade ago,” said Dr. Lena Ortiz.
But the biggest hurdle remains funding. The project relies on a $2.3 million grant from the National Science Foundation, and the next phase — deploying a larger fleet of AUVs — won’t launch until the fall. I’m hopeful, yet cautious (the ocean has a habit of swallowing tech). For historical context, see the National Park Service Earhart archive. And if you’re curious about AI’s role in other cutting‑edge research, check out McKinsey Tests AI Client Simulation for Entry‑Level Interviews.
Key takeaway
- New lidar can scan 200 m/hr, cutting search time by 60%.
- AI distinguishes metal from coral with 92% accuracy.
- Project funded by a $2.3 M NSF grant.
- Full deployment slated for fall 2024.
More from gnewsx Tech
For more in-depth tech coverage, visit gnewsx.com — your daily source for tech news, reviews, and analysis.
