Honeybees may hold a surprisingly useful lesson for the future of robotics: sometimes the best response to conflicting information is to wait before making a decision.
An international team of researchers has developed a simple decision-making method inspired by honeybee colonies that helps groups of robots reach agreement quickly, even when some robots are receiving or sharing unreliable information.
The study, led by computer scientist Andreagiovanni Reina from the University of Konstanz, was published in Nature Communications.
Robot swarms could eventually perform jobs that are difficult or dangerous for humans. Groups of relatively simple robots might search disaster zones for survivors, monitor sensitive ecosystems or investigate chemical spills.
Instead of relying on a person to control every machine, the robots would work together and make decisions as a group.
But this creates a problem. Robots need to exchange information to make good collective decisions, and that information isn’t always reliable.
Sensors can make mistakes, machines can malfunction, communication systems can introduce errors, and messages could even be deliberately manipulated in a cyberattack.
The researchers compared two simple ways robots could respond when they receive information that disagrees with what they currently believe.
In one approach, called “direct-switch,” a robot immediately changes its opinion to match the new information. This is simple and requires little computing power, but unreliable or conflicting messages can cause robots to repeatedly change their minds. As a result, the swarm may struggle to reach a strong agreement.
The second approach, called “cross-inhibition,” adds a short period of uncertainty. Instead of immediately accepting conflicting information, a robot temporarily becomes undecided. It then waits for additional evidence before choosing an option.
This behavior was inspired partly by how honeybee colonies select new nesting sites. Bees supporting one potential home can send signals that discourage other bees from promoting competing locations. Over time, this helps the colony settle on one destination rather than remaining divided.
When the researchers introduced unreliable information into robot swarms, this bee-inspired strategy generally produced faster and clearer decisions than the direct-switch approach. It continued to work well as swarms became larger and when robots had to choose between several options.
The researchers also discovered something unexpected: a moderate amount of unreliable information could sometimes improve the swarm’s decisions.
A small amount of disturbance could prevent robots from settling too quickly on a poor choice, giving the swarm more opportunity to identify the better option. This suggests that removing every source of noise isn’t necessarily the best strategy for designing reliable collective systems.
Similar decision-making patterns are found elsewhere in nature, including networks of brain cells and the molecular systems that control activity inside cells. In these systems, competing signals can suppress one another until one clear outcome emerges.
The researchers believe this may represent a powerful general principle that engineers can use to build more reliable autonomous machines.
In the future, bee-inspired robot swarms could make rapid collective decisions about where to search after a disaster, which environmental threat needs attention first or where limited resources should be sent—all while remaining surprisingly resilient when some of the information they receive is wrong.
