- Статьи
- Science and technology
- The eco-ecosystem has been raised: AI to clean up the Arctic and emotional robots
The eco-ecosystem has been raised: AI to clean up the Arctic and emotional robots
In Russia, "unkillable" mannequins will test driver assistance systems in self-driving cars, and an AI program will help clear the northern seas of debris. At the same time, Russian engineers figured out how to teach robots sign language, and their American colleagues created a machine with a human face that mastered facial expressions from YouTube videos. However, researchers from Brazil have found out why mosquitoes have fallen in love with humans. These and other science and technology news of the past week can be found in the Izvestia article.
The Arctic will be cleared of garbage with the help of AI
Scientists from the Institute of Oceanology of the Russian Academy of Sciences and the Moscow Institute of Physics and Technology (MIPT) have created a system that helps detect debris in the Arctic seas in real time. The development is based on artificial intelligence algorithms that process data from ship's cameras in a special way.
— The system identifies four types of objects — marine debris, birds, glare on the water and drops on the lens. The development is relevant for the Arctic region, where pollution from insoluble anthropogenic debris (industrial, household and plastic waste) poses a growing threat to the ecosystem," said Mikhail Krinitsky, head of the Machine Learning Laboratory in Geosciences at MIPT.
Photo of the scientific vessel and the side on which the camera was installed
According to him, more than half a million photos of the surface of the Barents and Kara Seas were processed during the study. At the same time, sea foam, glare from the sun and the rolling of ships were particularly difficult to analyze. These factors make it difficult to detect small objects on the surface of the water and at shallow depths.
Tests have shown that the development is many times superior to its global counterparts. Thus, the accuracy class of the Russian program was 0.4, while the closest competitor, the popular American YOLO algorithm, has a similar indicator of about 0.1. According to scientists, the new complex will allow creating up—to—date maps of garbage accumulations and promptly sending vessels to the desired area to clean up the waters.
Robots have been taught to understand sign language in Russia
Sirius University has introduced and patented a technology for remote control of robotic manipulators using hand gestures. The method allows you to dispense with remotes, joysticks, gloves and other special equipment.
According to the developers, the complex consists of a camera system and software. The main element is the depth camera, which helps to create a three-dimensional model of the operator's hands. Then algorithms recognize his movements in real time and convert them into commands. As a result, the robot repeats human manipulations with a delay of only 50 milliseconds.
According to the scientists, the complex can be activated and switched between different modes using simple "manual" commands. The development also allows you to "train" the machine with sequences of actions. For example, to capture and assemble parts. As a result, the robot will perform operations on its own in the future.
— We see the most obvious scope of application in the framework of unpopulated production. Gesture control technology is a natural and safe way that allows operators to remotely train, configure and adjust the operation of industrial robots," Andrey Shipatov, one of the developers, associate professor of Mathematical Robotics at Sirius University, associate professor at the National Research University MIET, and CEO of Elmix, told Izvestia.
"Unkillable" test dummies for self-driving cars created in Russia
Engineers from the R.E. Alekseev Nizhny Novgorod State Technical University have developed and patented a device that helps a test dummy accurately simulate the gait of a pedestrian. The device was created to test driver assistance systems.
Unlike analogues, the new complex can be programmed to perform various actions. For example, running, walking, or stopping suddenly. In addition, an artificial pedestrian can reproduce the state of fright and other behaviors of people under stressful conditions. This is important for AI training for self-driving cars.
— The thruster was positioned in the pelvic area of the dummy. At the same time, his legs are connected to his body by magnets. Thus, in an accident, the limbs easily fly off to the side, and then they can simply be attached back. This allows us to quickly restore the test facility in order to continue research," said Anton Zolotov, head of the production department at the NSTU Industrial Robotics Development Center.
According to him, the device is protected by special shock-absorbing surfaces and can withstand collisions with cars at speeds of 80 km/h. This allows the device to be used repeatedly.
The robot has mastered facial expressions from a YouTube video
Scientists from Columbia University (USA) have created a robot that has mastered the ability to move its lips and express emotions on its face. According to the developers, the robot was equipped with 26 motors to control facial expressions and AI tools. He first studied the control of facial "muscles" in front of a mirror, and then analyzed YouTube videos, and as a result learned to synchronize facial movements with speech and singing.
—When the ability to synchronize lips is combined with conversational AI, such as ChatGPT or Gemini, the effect gives a completely new depth to the connection that the robot establishes with humans," explained project leader Johan Hu, a member of the Faculty of Mechanical Engineering at Columbia University.
In the future, scientists intend to achieve an integrated robot communication system based on large language models such as GPT chat. The researchers suggest that the more the robot observes people and learns, the better it will imitate subtle facial gestures, through which it will be able to establish an emotional connection with people.
The disappearance of forests has provoked the interest of mosquitoes in humans
Biologists from the Oswaldo Cruz Institute and the Federal University of Rio de Janeiro (Brazil) studied mosquitoes in the Atlantic Forest (an area in western South America) and using DNA analysis found out whose blood they feed on. The study showed that in the vast majority of cases, people became their victims.
— A total of 1,714 mosquitoes were caught, of which 145 females (6.98%) were full. The results revealed a tendency for mosquitoes to feed primarily on humans, the scientists reported in the publication.
They explained that the blood in the stomachs of insects in 18 of the 24 successfully deciphered cases was human.
According to the researchers, this is an important observation, as it shows how the preferences of mosquitoes change as a result of massive deforestation. In particular, the Brazilian Atlantic Forest is known for the fact that since the arrival of Europeans, more than 80% of its vegetation has been destroyed in favor of agriculture and urbanization.
Previously, the massif was home to a large number of animals — potential victims of mosquitoes, but as the forest shrank, the number of animals decreased, and mosquitoes changed their eating habits, switching to humans. As a result, the risk of infection with diseases caused by mosquitoes has increased. Scientists believe that similar processes occur wherever forests are destroyed.
Переведено сервисом «Яндекс Переводчик»
