They develop software that would make robots work in a ‘swarm’

The collective work carried out by certain groups of insects served as the basis for trying to answer this question, based on automatic software design. The idea came from the electronic engineer, David Garzón Ramos, Master in Engineering – Industrial Automation, from the National University of Colombia (UNAL) Manizales Campus, who thinks that in the future the robots of consumption, that is, those who are used in daily activities such as housework, must learn to work in a group.

Even from the United States Army Research Laboratory, Harvard University or the Massachusetts Institute of Technology (MIT), studies are being carried out with robotic swarm systems that could be used for the search and rescue of hostages after terrorist attacks, in support of humanitarian missions in conflict areas, monitoring borders and locating environmental disasters, tasks in which they fit the objectives raised.

“As a society we must solve needs such as building roads, buildings or carrying out agricultural activities, and for this we need to work with large groups of robots, such as robot swarms”added David Garzon. In addition, he indicated that based on an empirical study, it was evaluated whether existing methods of manual and automatic design of collective behaviors are effective in producing control software for one swarm that coordinates another.

The research was developed with simulations executed in ARGoS3, a simulator specialized in swarm robotics. The experiments considered two groups of robots: an “active swarm” made up of five robots that are in charge of coordinating, and a “passive swarm” made up of ten robots, which need to be guided by the active swarm to carry out a series of tasks. . To create coordination strategies between the groups, Automatic and manual software design methods for robots were explored and tested in four tasks in which they interact with each other.

One challenge was that the passive robots, which were scattered, were brought together by the active robots. Another task was dispersion, in which the passive robots were grouped together and the active robots had to separate them, in order to maximize the relative distance between them. The third challenge was to carry out a herding in which active robots had to direct the passive robots to a certain place. Finally, in the fourth activity, segregation, where the active robots sought to separate the passive robots into small groups.

In this regard, the researcher points out that “he used optimization methods that seek, combine and tune different modular behaviors (exploration, attraction, repulsion, among others)”. “These behaviors were inspired by the observation of social animals such as ants, bees, fish, birds; and they were useful for studying collective robotics in the past.”

Modular automatic design is capable of creating complex behaviors from the combination of simpler behaviors. For example, exploration (random movement through the environment) and phototaxis (directed movement towards a light source) modules can be combined to help robots search for and pick up objects in their environment, and bring them to a certain location.

The results showed that it is possible to make simple robots with passive behaviors be adapted with automatic design methods so that other (active) robots can coordinate them. This research was developed with the support of UNAL, the Université Libre de Bruxelles and IRIDIA. Electronic Engineering students from the University of Nariño participated. In addition, it received funding from the DEMIURGE project and Minciencias.

Another example of how technology takes nature as an example occurred in a research carried out by scientists from the University of Zhejiang (China), who created an innovative system that can plan microdrones flight routes so that they do not appear. collisions in the process. The method is inspired by how animals can flexibly coordinate their collective flight paths, identifying obstacles and sizing shared spaces.

The experiment is designed to demonstrate swarm navigation with full autonomy in a very dense environment.that is, a bamboo forest, without harming themselves or the plants, said the study, published in the journal Science Robotics. One of the objectives is to help rescuers in search and rescue operations, also support animal and plant researchers or even propose a new package delivery system by means of drones which are planned for the near future. As a result, By following the planned trajectories, the swarm flies through the forest while staying in proper formation and not having accidents or colliding with each other.

Newspaper La Nacion / Argentina / GDA