AI AND NATURE
NATURAL MODELS
18.09.2020
Drones, autonomous cars, unmanned long-distance trains – robots have long since made their way into everyday life. When it comes to the question of the rules by which these vehicles will move and communicate with each other in the future, we can learn a lot from nature.
Ants are known for their unimaginable power and energy output in relation to their size, as well as for their architectural skills. It is not so widely known that they serve as a model for the traffic of the future. The representatives of the Formicidae family from the order of hymenoptera have special abilities when it comes to finding their orientation on unknown terrain. Anyone who has ever admired an ant trail has no doubt asked themselves how the insects manage to find the shortest route between their home and a source of food. To do this, the temperate climate species uses scents above all: in their search for food, they swarm out in all directions, marking their paths with pheromones. Once they have found food, they retrace their steps and, in this way, strengthen the spoor. The next ants follow the strongest pheromone trail – and within a very short time a densely frequented path is formed. The use of scents in road traffic does not promise too much success, however. But if phosphorescent dyes were used instead of scents, the situation would be quite different: a route could be defined for autonomous vehicles in this way. An approach that could play a role in the traffic planning of the future.
ON HOT SAND
The relatives of the European ants who live in the African desert have to find other solutions, as fragrances evaporate immediately in the heat. They rely on the sky for orientation, but above all on their memory and algorithms. When a desert ant leaves its home in search of food, it remembers every step and every turn on its journey. It counts its steps and thus unerringly finds the small hole in the ground, its front door, from a distance of up to several hundred meters. To find out how orientation works under changed conditions, researchers in an experiment lifted the swarming ants out of the trail and set them down again a little further on to confuse them. The insects began to work their way back – using an extremely efficient search algorithm aimed at searching the widest area of terrain possible while covering the shortest distances possible, thus saving energy. Within a very short time, the ants found their way back to their homes despite the adverse conditions. These findings were used, among other things, in the programming of Mars probes. The ants’ tracks therefore not only lead to the roads, but also into space. But the orientation skills of other animals are no less impressive.
SWARM INTELLIGENCE
Up till now, the relationship between cars and birds has largely been limited to vehicle paintwork being damaged by their droppings. The age of autonomous, self-driving cars will add a whole new chapter to this common history. The reason: The proponents of autonomous driving will not be able to avoid dealing with flocks of birds. It is no coincidence that connected cars are also referred to as swarming robots. But what exactly can a group of wild geese, for example, teach us about the autonomization of road traffic?
Flocks of birds are not only fascinatingly beautiful, but also astonishing in terms of their consistency. At least to the layman. Experts, of course, know how it is possible for hundreds of animals to fly apparently synchronously with each other, adjusting direction, altitude and speed without colliding with each other. An appealing scenario, especially in road traffic. To do this, the swarm only has to follow two rules:
1. Someone needs to know where to go.
2. The distance to the nearest neighbor must always remain the same.
In regard to the first rule, it was found that even a very small percentage of “informed” birds is sufficient to lead a flock to its destination. Researchers and developers are working intensively on the question of how this information can be shared between self-driving cars. Must all the information be communicated? Is just a part of it sufficient? How can it be transmitted to obtain a “safe swarm”? The question is, therefore, about how information recorded by a vehicle equipped with networked communication technology can be passed on to other vehicles in the vicinity, long before a driver notices anything. Cars can thus share their collective knowledge and disseminate information almost in real time to make it easier for human drivers to take action to avoid problems before they occur, for example, in a traffic jam that is potentially kilometers long. A heavily loaded truck knows when it is approaching a hill thanks to digital maps and satellite navigation. The truck’s AI can consequently calculate when the vehicle will slow down. This information on the traffic status is shared with other road users, whereupon they automatically adjust the distance to the vehicle in front, thus complying with the second rule of swarm safety. To avoid being unjust to anyone at this point, it should be mentioned that the behavior of fish and honey bees has also played and continues to play an important role in the development of AI for communication between drones and self-driving cars. And who knows what else we will be able to learn from other creatures? In the field of bioacoustics, for example, researchers have discovered that plants can communicate with each other via cracking sounds produced in their roots. Bumblebees, which are considered the only animals with mathematical abilities, solve the traveling salesman problem, i.e. the task of keeping the distance as small as possible on a return tour of a certain number of stations, in a way that would take days for today’s supercomputers. Dolphins communicate by ultrasound, the elephantnose fish uses electrical sensing to find its orientation in murky waters, the black firebeetle can detect forest fires at a distance of 80 km with its infrared sensory organs – fire detectors have already been modeled on it. So there is no lack of role models.
GROW AND PROSPER
Advances are constantly made in leaps and bounds thanks, among other things, to insights gained from the animal and plant world. Nevertheless, however impressive the progress of the new technologies might be, the experts agree that more research is required. Dr. Joseph Reger, CTO of Fujitsu Germany, believes that “Artificial intelligence is still a delicate blossom, but it will mature quickly, and become a powerful tool.” And this too, with the help of Mother Nature.