The scientific process: machine learning on a quantum computer and gasoline out of thin air

Russian researchers have taught robots to navigate according to the principles that work in the human brain, developed a universal external pilot station that helps control drones of various types, and also applied machine learning on a quantum computer. In the USA, a device has been introduced that produces gasoline from the air. And China has created the world's largest 250-meter farm vessel for breeding salmon and other valuable fish species in the open sea. Read about the most interesting scientific achievements over the past week in the Izvestia collection.

Human memory algorithms for robot orientation

Scientists from the Institute of Computer Science and Management of the Russian Academy of Sciences and the Moscow Institute of Physics and Technology have developed an innovative method for robot orientation that allows them to dispense with global positioning systems and detailed terrain maps.

The new approach is based on algorithms similar to those used by the human brain. The program identifies the main landmarks on the ground and establishes the relationships between them. This allows the machines to save computing resources and at the same time plan the route more quickly.

—Experiments in virtual 3D environments and field tests have shown that the new method successfully builds accurate and coherent maps, even in the presence of sensor measurement errors," said Alexander Melekhin, an engineer at the Intelligent Transport Laboratory.

As the researchers explained, the development was called PRISM-TopoMap (Place Recognition and Integrated Scan Matching for Topological Mapping) — location recognition and integrated comparison of scan results for topological mapping.

The technology allows the machine control device to build and update the map directly while driving. In the future, the developers intend to train the system to determine the type of premises, distinguishing between workshops, warehouses, kitchens and corridors.

Universal "external pilot" for drones

The country's first and one of the first universal external pilot stations in the world was developed by specialists from the Aeromobility Directorate and the Unmanned Aircraft Systems Center at the Moscow Aviation Institute. The device allows you to control drones of aviation, copter and hybrid types.

The station communicates with the unmanned aircraft via a radio channel. The device allows you to build and monitor the flight path of the UAV and receive and manage data from payloads.

"Interaction with the drone begins even before takeoff. The operator needs to set a route, specify key points, altitude, speed, and additional tasks. For example, turning on the camera, shooting, stopping in the air. All this data is transmitted to the drone's control system. As soon as it takes off, a continuous exchange of information with the ground station begins," said Maxim Kalyagin, one of the developers and chief designer of the MAI's Unmanned Aerial Vehicles center.

Replace weapons: which strike systems can be replaced by drones

And what classes of UAVs are already used on the battlefield today?

The device includes a touch screen, a control panel, and a housing that is resistant to adverse external conditions. The battery is designed for five hours of active operation.

Machine learning on a quantum computer

Specialists from the Lebedev Physical Institute of the Russian Academy of Sciences (FIAN) and the Russian Quantum Center (RCC) demonstrated the solution of applied problems on a quantum computer. Moreover, to achieve this goal, they are one of the first in the world to use machine learning algorithms for quantum computing.

— Such technologies are being actively developed in all leading countries. At the moment, an important challenge is testing quantum computing methods in various applications. In particular, one of the main results of our work is the use of these algorithms in combination with machine learning technology," said Alexey Fedorov, head of the Quantum Information Technologies Group at the RCC.

According to him, during the experiment, the specialists used a processor based on ytterbium (Yb+) ions. With his help, they separated the handwritten images of zero and one and a number of mathematical objects — graphs. As a result, the quantum computer correctly identified all the numbers in both the training and test images. That is, I completed the task without errors.

Thus, scientists have shown that even small quantum processors can already solve simple but practically significant tasks, such as image classification. This is a step towards a future where quantum processors will start performing more complex calculations. For example, to verify DNA sequences or to model new molecular structures.

Gasoline will be extracted from the air

In the USA, a device has been introduced that produces gasoline from the air. It was developed by the Aircela startup team. The device extracts carbon dioxide (CO2) and water (H2O) from the surrounding space and, as a result of a number of chemical processes, converts them into high-octane fuel.

The installation receives the necessary energy using solar panels. According to the inventors, the development is suitable for use in ordinary living conditions.

— We use electrolysis to separate water into hydrogen and oxygen using environmentally friendly electricity. The latter is safely released, and the hydrogen and trapped Co₂ combine to produce methanol. It is then converted to gasoline using standard chemical processes. The result is a gasoline that does not contain fossil fuels and is fully compatible with existing engines and infrastructure," the company's website says.

The scientists reported that they are ready to start mass production of the plants, but, as experts suggest, at the moment the cost of energy and consumables will make the cost of the process too expensive. In the future, such devices will find their niche.

Scientific interest: the status of the clownfish and the mystery of the origin of lightning

Izvestia collected the top 5 research news for the week

A ship for growing 8 thousand tons of salmon per year

China has unveiled the world's largest 250-meter farm vessel for breeding salmon and other valuable fish species in the open sea. It was named Su Hai No. 1, which indicates a connection with Jiangsu Province. It is expected that the floating production will operate in the waters of the Yellow Sea, where conditions for salmon are optimal (low temperatures and good water circulation). They will grow up to 8 thousand tons of fish per year.

According to the shipowner of Jiangsu Lianshen Marine Technology, the vessel is equipped with artificial intelligence, water purification, automatic feeding and monitoring systems. This minimizes the impact on the environment.

The Su Hai No. 1 sea trials took place in April, and the vessel will be launched in June. An industrial line is planned to be launched in the autumn, and the first batch of salmon will be caught in 2026. In total, about 100 thousand tons of this fish are consumed in China. The new project will cover up to 8% of the demand. If it proves successful, a flotilla of similar vessels will be created in the country.