Cheating for food, robots are capable of it

Evolution in silicon has shown that machines are capable of self-improvement. Generation after generation selection and selection allow to debug and select the most adapted samples. Now, robots show scientists how strategies of animal behavior in natural conditions could develop and improve. Search for food, fight with competitors, natural signals, how they affect everyday life and how they were formed. In the Swedish laboratory of the Laboratory of Intelligent Systems, a group of 10 robots competed for food. What came of it?

For the experiment, s-bots were chosen, with a diameter of 12 cm, a height of 15 cm, and having 2 lithium-ion batteries, giving about an hour of autonomous existence. The robots are equipped with a 400 MHz Xscale processor, 64 MB of RAM and 32 MB of flash memory used for data processing, as well as 12 PIC microcontrollers for low-level processing. The operating system uses self-assembly Linux, and communication with the central station is via WiFi. The sensory capabilities of the robot include infrared sensors (15 around the robot and 4 under the robot), power and speed sensors, humidity and temperature sensors, as well as 8 light sensors, a camera with all-round visibility and 4 microphones.

The robots were programmed to search for a “food source,” which was a lightly luminous ring at one end of the arena. They could “see” this source only at close range with the help of their sensors. At the other end of the arena was a darker ring, which is considered "poisoned." Robots received points depending on how much time they spent near food sources and / or poison, thereby demonstrating how successful their artificial lives were.
And they could "communicate" with each other. Each robot could project the blue light that others could define with their cameras, and thus indicate the position of the food. Thus, the light for robots is a carrier of information. However, after several generations, robots have evolved and taught their descendants not only to communicate with the light, but also to deceive rivals.

Robots were able to evolve, due to the fact that each robot was equipped with an artificial neural network controlled by a binary "genome." The network consists of 11 neurons that are connected to sensors of robots and 3 neurons that control their two wheels and blue color. Neurons are connected using 33 synapses, and the power of the signals of each synapse was controlled by one eight-bit gene. Thus, each robot has a 264-bit genome that controls how it will react to information from the sensors.
Artificial evolution took place in the Enki simulation environment, where both robots and their sensors were modeled. Then there was the use of the evolutionary framework of robots Teem for the evolution of the best controllers, which were then transferred to real robots.
')
In this experiment, each round consisted of 100 groups of 10 robots, each group competing for food in a separate arena. The two hundred robots with the most points, the most adapted from the population, moved on to the next round. Their 33 genes were randomly mutated with a probability of 1/100 that any bit would change, and the robots were “merged” with each other to mix genomes. The result is a new generation of robots, whose behavior was inherited from the most successful representatives of the previous one.

In the initial experiment, robots emitted light randomly. However, after generations, the robots were getting better ... The light began to carry more and more information and bots began to focus on the light only after 9 generations.
But the situation with robots was similar to the behavior of real animals, because it is not always in the interests of the robot (animal) to transmit information about the location of food. The ring with food has only 8 places for robots, that is, if everyone gathers around a food source, they will have to physically push each other. The effect of the struggle increased when the experimenters allowed the genes responsible for the emission of blue light to evolve.

What was the result? If the robots initially emitted light at random and when they gathered around the food, the light produced the location of the food. That after the evolution of the robots have become more secretive. By the 50th generation, they began to emit light around food with much less likelihood than anywhere else in the arena, and the light itself became not such an important source of information, making it less attractive to robots.
However, the light did not become completely useless. As robots became more cunning and less reliant on light, individuals that glowed near food could be a kind of protection, because light could be regarded as true or false depending on the perception of another robot. Therefore, an evolutionary pressure calling for no signal at all was diluted by the aforementioned factor and because of this light was still used after generations.

It also means that the robots were diverse in their behavior. With free natural selection, such processes as genetic drift (when genes change randomly) - genetic diversity easily arises, which in turn creates a diverse behavior of individuals. After 500 generations of evolution, about 60% of robots never emitted light around food, but about 10% emitted it there most of the time. And some robots began to emit light around the poison, luring other robots. Some robots attracted a little blue light, but a third of the robots, he attracted very strongly, and another third did not react to it.
Experimenters think that similar processes occur in nature. When animals move, for example, for food, they thereby inadvertently signal this to other animals. This creates a conflict of interest, and natural selection will allow a better life for those who can suppress or modify this information, for example, using disguise, secrecy, creating interference or false signals. As in the experiment with robots, these processes can help to understand and explain the huge variety of behavioral strategies in the natural world.

This study gives every reason to believe that robots can be applied far beyond the technical sphere. As one of the researchers who conducted this experiment said: “Robots can be quite useful for studying and better understanding the mechanisms of interaction between living organisms. This is only the beginning, but we are confident that robots will be used for research in biology, psychology or medicine. ”

Video 6 robots looking for food (14 Mb)
Schematic video of the fight of robots for food (20 Mb)