bg
Extractive industry
14:54, 17 November 2026
views
1

Russian Researchers Enable Underwater Vehicle to “See” Ore Deposits at a Depth of Four Kilometers

Marine mineral exploration is considered one of the most challenging fields in the mining industry. Operations in the open ocean or on the Arctic shelf require specialized technologies capable of withstanding extreme conditions. Researchers at St. Petersburg State University have developed one such technology.

The hardware and software system has been tested during an expedition to the Mid-Atlantic Ridge. A camera housed in a pressure-resistant enclosure operated at a depth of four kilometers. Onboard software equipped with computer vision capabilities took on a task previously handled only by geologists: identifying ore formations in video footage.

Advanced Technology Underwater

Researchers manually labeled more than 50,000 video frames recorded during previous expeditions. Using these annotations, the algorithm learned to recognize patterns: what ore looks like, including its shape, color and texture, and how to distinguish it from ordinary rock and underwater interference. Now, when the camera surveys a new location, the system points researchers to areas where ore deposits may be found. Instead of receiving only a raw video feed, geologists get footage in which the software has already highlighted the areas of interest.

This technology helped researchers refine the boundaries of the Semenov ore zone during the expedition to the Mid-Atlantic Ridge. In other words, they were able to establish more precisely where the mineral deposits lie and how far the ore reserves extend. The system consists of a camera inside a rugged, pressure-resistant housing and a cable tether designed to supply power and transmit data under extreme conditions. Its core component is the software that analyzes video footage and identifies objects.

A Broad Horizon for Research

According to the GEBCO project, 28.7% of the world's ocean floor had been mapped to modern standards by early 2026, up from 27.3% at the beginning of 2025. Progress is being made, but much remains to be explored. More than 70% of the seabed still lacks detailed, high-resolution maps. And while mapping uses echo sounders to survey the seafloor remotely, direct observation – using submersibles, cameras and sampling equipment – covers only a very small fraction of the ocean floor. That makes technologies capable of providing a closer look at deep-sea mineral deposits particularly valuable.

In Russia, researchers at the St. Petersburg Federal Research Center of the Russian Academy of Sciences (SPb FRC RAS) have produced a detailed map of a section of the Pacific seabed in the Magellan Seamount region, an area rich in cobalt-rich crusts used as raw materials for electronics and energy technologies. The researchers combined two methods: acoustic surveying and underwater video recording. By cross-referencing sonar data with video footage, they created a geographically accurate map showing the location and boundaries of areas with potential for development.

Meanwhile, St. Petersburg Electrotechnical University “LETI” has developed an acoustic instrument described as a “new-generation echo sounder.” It sends sound waves through the water and uses their reflections from the seabed to build a three-dimensional picture of the underlying sediments and rock. The resulting data show the seafloor relief, layer thickness, density and structure. The software automatically classifies the seabed materials, distinguishing sand from silt and solid bedrock.

The Moscow Institute of Physics and Technology (MIPT) and the Shirshov Institute of Oceanology of the Russian Academy of Sciences have developed software that analyzes images of the sea surface taken by drones, aircraft, ships and satellites. It automatically identifies objects such as floating debris, pollution slicks and vessels. The researchers trained the neural network on a dataset of more than 500,000 images collected in the Arctic. Because the algorithms were trained to handle challenging weather and lighting conditions – including glare, waves, low-angle sunlight, fog and ice – these disturbances do not prevent the system from identifying objects under conditions in which conventional recognition systems fail.

From the Lab to the Atlantic Seafloor

The most obvious next step is to adapt the system for use in other parts of the Atlantic and Pacific oceans, where Russia is already exploring deposits of polymetallic sulfides, ferromanganese nodules and cobalt-rich crusts. But comprehensive geological exploration also requires specialized software. The algorithms could be further developed to map mineral deposits automatically, classify rock formations and correlate images with sonar data. With the right adaptations, the system could also be used to inspect subsea pipelines and marine infrastructure, as well as for environmental monitoring.

As for the new hardware and software system, further testing is likely in the coming years, along with expansion of its training dataset and integration with hydroacoustic systems. If these efforts prove successful, the technology could move toward serial production, paving the way for automated mapping of extensive deep-sea areas.

The future of digital resource management will not be determined by how many new datasets we accumulate. It will depend on whether we can turn fragmented data into connected, evidence-based knowledge that is readily available for decision-making. It is not about a single database – it is about creating a unified space of meaning.
quote

like
heart
fun
wow
sad
angry
Latest news
Important
Recommended
previous
next