Keys to the Swarm: Ant watching will teach robots to work together

Russian scientists have developed a technology for controlling groups of robots and swarms of drones, which allows several devices to act in concert with minimal information exchange, maintain system performance in case of failure of some mechanisms, and solve complex computational problems in real time using mathematical models. One of the stages of the project is the observation of ants. Based on the data obtained, the researchers plan to create experimental stands to find out how insects without central control coordinate actions when solving complex tasks. The technology can be used not only in drone systems, but also in microgrids of solar panels that independently balance the load, groups of autonomous trucks and other systems. For more information about the development and its natural origins, see the Izvestia article.

Borrowed a mechanism from nature

Specialists from the Scientific Center for Information Technology and Artificial Intelligence at Sirius University have proposed a new approach to decentralized robot management using biomimetic solutions, principles borrowed from wildlife.

Modern robotics, distributed energy and intelligent control systems are increasingly in need of a transition from centralized to decentralized architectures, the developers explained. If a single center makes decisions, the system becomes vulnerable: its failure can lead to the shutdown of the entire network. When each device solves local problems by itself and negotiates with its neighbors, the system becomes resilient, flexible and scalable.

Within the framework of the Multi-Agent Technologies in Complex Robotic and Distributed Computing Systems project, specialists have developed and tested a number of algorithms for managing groups of robots with efficient use of resources. One of the results is a logical time synchronization system on microcontrollers, inexpensive and common chips that are used in a wide variety of electronic devices, from motion sensors to household appliances. Tests have shown that the devices synchronize with microsecond accuracy and quickly reconnect after signal loss. This part of the development has already been tested on real hardware.

Another area of the project, studying the behavior of ants, is still in the preparatory stage. Scientists plan to create experimental stands to find out how insects solve complex tasks in a coordinated manner without central control. The revealed patterns will form the basis of new mathematical models and algorithms for robotics.

Cognition of machines: robots have been taught to "feel" space like a human

Embodied AI helps technology to see and understand environmental changes

— Observations of ants will help to understand how the coordinated work of the entire colony develops from the actions of individual individuals and what causes it to change its behavior when conditions change. We are particularly interested in how personal memory, insect-to-insect contact, and the footprints they leave together ensure the transition from food search to regular food delivery. Based on experimental data, we will be able to verify which mechanisms really determine this dynamics and refine mathematical models. It will also allow us to identify common principles for creating systems in which a group of robots independently rebuilds collaboration in response to environmental changes," Ivan Lomakin, deputy head of the Scientific Center for Information Technology and Artificial Intelligence at Sirius University of Science and Technology, told Izvestia.

Where swarms of robots and drones will find use

The technology can be used in a wide variety of fields. In mobile robotics, it will make it possible to coordinate swarms of drones and ground-based robots during search and rescue operations, monitoring territories and performing agricultural work. In the energy sector, it is important to manage microgrids with renewable energy sources and balance the load. In transport and logistics, the goal is to coordinate autonomous vehicles, optimize their loading and routes. For ordinary people, the introduction of such solutions can mean increased safety of autonomous transport, greater reliability of energy supply and faster work of rescue services.

Group control of autonomous devices is one of the areas that will largely determine the development of robotics in the coming years, Artyom Bogatyrev, market expert at NTI Aeronet, told Izvestia.

— The more devices work in one system, the more difficult it is to manage them from a single center: the load on communication channels increases, and a command post failure can stop the entire group. A decentralized approach allows us to solve this problem," the expert noted.

According to him, the first such solutions can be implemented in the fields of unmanned aircraft, autonomous transport and distributed energy. For example, a swarm of drones is able to explore a vast area faster when searching for people, monitoring forests, or inspecting power lines.

Whale of the season: The secret of fish anatomy will make underwater drones super energy efficient

Biosimilar devices will find application in environmental monitoring and covert surveillance

— The next stage will be testing on real equipment in conditions of interference and unstable communications. Their results will show how quickly the development can be implemented in industry," said Artem Bogatyrev.

At the same time, one of the main difficulties in evaluating developments in the field of group control of drones is the lack of a specific scenario for their use, said Andrei Boyko, Candidate of Physico—Mathematical Sciences, senior researcher at the ITMO University Faculty of Information Technology Security.

Without understanding the conditions and practical tasks under which the proposed algorithm will be used, it is difficult to objectively assess the scientific novelty and practical significance of the results, the expert emphasized. The logical time system, despite its advantages, is suitable only for a certain range of tasks. In addition, there are already a number of technical solutions for synchronizing timelines. It is much more difficult to determine the relative spatial position of the devices.

"It is important to take into account physical limitations here: when exchanging signals between devices, the accuracy of time measurements is affected not only by the clock error, but also by the signal propagation time, which depends on the distance between the devices," the expert explained.

Thus, in order to fully evaluate the development, it is necessary to specify a specific scenario in which the proposed approach really provides significant advantages — for example, due to simplicity of implementation, compactness and low cost.