Teaching robots to be team players with nature

Algae bloom, birds swarm and insects swarm. This en masse behavior of individual organisms can provide separate and collective benefits such as: B. improving the chances of successful mating propagation or providing security. Now, researchers have harnessed the self-assembly skills needed to take advantage of natural swarms for robotic applications in artificial intelligence, computing, search and rescue, and more.

They published their method on August 3rd intelligent computing.

“Designing a set of rules that, once executed by a swarm of robots, will lead to a specific desired behavior is a particular challenge,” said corresponding author Marco Dorigo, professor at the Université’s IRIDIA laboratory for artificial intelligence Liber de Bruxelles. Belgium. “The behavior of the swarm is not a one-to-one map with simple rules executed by individual robots, but results from the complex interactions of many robots executing the same rules.”

In other words, the robots must work together to reach the sum goal of discrete contributions. According to Dorigo and his co-authors Dr. Valentini and Prof. Hamann, the problem is that traditional single unit design to achieve a common goal is bottom-up, requiring refinement through trial and error that can be costly.

“To address this challenge, we propose a novel global-to-local design approach,” said Dorigo. “Our key idea is to assemble a heterogeneous swarm from groups of agents with different behaviors, such that the resulting swarm behavior corresponds to a user input that represents the behavior of the entire swarm.”

This assembly involves selecting individual agents with predetermined behaviors that researchers know will work together to achieve the desired collective behavior. You lose the ability to program individual units locally, but Valentini, Hamann and Dorigo say it’s worth the tradeoff. They pointed to an example of a surveillance task where a swarm may need to monitor a facility that requires more internal surveillance during the day and more external surveillance at night.

“The user provides a description of the desired swarm assignments as a probability distribution over the space of all possible swarm assignments – more agents inside during the day, more outside at night or vice versa,” says Valentini.

The user would define the target behavior by changing the number and position of distribution modes, each mode corresponding to a specific mapping, such as B. 80% of agents inside, 20% outside during the day and 30% inside, 70% outside at night. This allows the swarm to change its behavior periodically and autonomously, dictated by the set modes, as circumstances change.

“While finding the precise robotic control rules to make the swarm behave as we desire is difficult, desired swarm behavior can be achieved by combining different sets of control rules that we already understand,” Dorigo said. “The swarm behavior can be designed macroscopically by mixing robots with different predefined rule sets.”

This isn’t the first time Dorigo has turned to nature to improve approaches to computer science. Previously, he developed the ant colony optimization algorithm, which relies on how ants navigate between their colonies and food sources to solve difficult computational problems involving finding a good approximation of an optimal path on a chart.

While Dorigo first proposed this approach for a relatively simple problem, it has since evolved into a means of solving a wide variety of problems. Dorigo said he plans to take the swarm methodology in a similar direction.

“Our immediate next step is to demonstrate the validity of our methodology for a larger set of swarming behaviors and go beyond task assignment,” Dorigo said. “Our ultimate goal is to understand what makes this possible and to formalize a generic theory that will allow researchers and engineers to design swarm behaviors without having to go through the tedious process of trial and error.”

Provided by Intelligent Computing