Artificial Intelligence Trained to Find Shipwrecks

Summary

In Michigan's Great Lakes, a team led by Professor Katie Skinner is developing advanced technology using AI and autonomous vehicles to explore shipwrecks. With over 1,500 known wrecks in the area, the team is using sonar data to autonomously identify sites of interest, improving the efficiency of underwater searches. Their work focuses on detecting wrecks and debris fields, aiming to refine search methods for both known and undiscovered shipwrecks. This innovative approach could revolutionize maritime archaeology and search-and-rescue operations, including efforts like locating missing aircraft such as Malaysian Flight 370.

Takeaways

• 😀 Michigan waters contain around 1,500 known shipwrecks, mostly from the 19th century, including schooners, brigs, and steamers.
• 😀 Less than 1% of the estimated 3 million shipwrecks worldwide have been discovered, highlighting the vast potential for underwater exploration.
• 😀 The University of Michigan's team, led by Professor Katie Skinner, is developing AI technology to aid in the search for shipwrecks in the Great Lakes.
• 😀 AI technology processes sonar data to quickly identify areas of interest, increasing the efficiency of wreck searches.
• 😀 Initial tests of the AI system are taking place in Lake Huron, particularly near Thunder Bay National Marine Sanctuary, which has over 100 known shipwreck sites.
• 😀 The AI-driven exploration uses autonomous vehicles like the IVER, which follow pre-programmed survey patterns to map the seabed.
• 😀 The survey patterns are designed to detect clues or patterns in debris, which can help identify shipwrecks more efficiently.
• 😀 AI training includes analyzing large amounts of sonar data to teach the system how to differentiate shipwrecks from natural features on the seabed.
• 😀 The AI system can autonomously recognize shipwrecks and other archaeological sites of interest, even if they are previously unknown.
• 😀 This AI-based approach could also improve search efforts for other missing objects, such as the wreckage of Malaysian Flight 370, by narrowing down search areas.
• 😀 The combination of AI and underwater robotics represents a significant leap forward in both shipwreck detection and our ability to explore and understand the ocean and lake floors.

Q & A

What is the estimated number of shipwrecks in Michigan waters?

There are approximately 1,500 known shipwrecks in Michigan waters, primarily around the Great Lakes.

What types of ships were involved in the majority of shipwrecks in Michigan?

Most of the shipwrecks in Michigan involve 19th-century vessels such as schooners, barks, and brigs. Later wrecks also include steamers and propeller freighters from the early 1900s.

How many shipwrecks are estimated to be in oceans and lakes worldwide?

It is estimated that 3 million shipwrecks lie at the bottom of oceans and lakes worldwide, with less than 1% of these wrecks discovered.

What is the main challenge faced by researchers in finding shipwrecks?

The main challenge is the vastness of the areas to search, with experts like Guy Meadows mentioning that most of the time involves reviewing images without finding any wrecks.

How is AI being used to aid in the search for shipwrecks?

AI is used to analyze sonar data, helping to quickly identify areas of interest on the sea floor that may indicate the presence of shipwrecks.

Where are the initial tests of the AI technology being conducted?

Initial tests of the AI technology are taking place on Lake Huron, specifically around the Thunder Bay National Marine Sanctuary in Northern Michigan.

What is the purpose of the 'lawnmower' pattern in the underwater survey missions?

The 'lawnmower' pattern involves the underwater vehicle performing multiple back-and-forth passes across the sea floor, allowing it to cover the area methodically and gather sonar data efficiently.

What specific wreck site is the team studying during the mission in Thunder Bay?

The team is studying the wreck of the Lucinda Van Valkenburg schooner, which sank in Thunder Bay after a collision, and they are analyzing the debris field around the wreck.

What is the broader challenge that the AI system is trying to solve, according to Guy Meadows?

The AI system aims to help refine large-scale search operations, like the search for Malaysia Airlines Flight 370, by narrowing down search areas to those most likely to contain wrecks or clues.

What does the AI system do with existing sonar data?

The AI system analyzes existing sonar data, which includes both known shipwrecks and natural seabed formations, to learn patterns and features that can help identify new wrecks or archaeological sites.