How Is Swarm Intelligence Implemented in Swarm Robotics?
If you enjoy playing online slot machines, you also know what to do to win: you have to place symbols side by side and use them together. In games such as the Reactoonz slot machine, you need to create clusters with identical symbols. In other words, you have to put together things that are alike and used for the same purpose. Swarm intelligence technology actually works this way. So what is this technology and how is it used in the robotic industry? What can we expect from it in the near future? We answer all these questions for you below.
What Is Swarm Intelligence?
Let’s start with the basics first. Swarm intelligence is a concept used in artificial intelligence studies and means collective behaviour of self-organized systems. When Gerardo Beni and Jing Wan started working on cellular robotic systems in 1989, they realized that it was not possible to give direct orders to such small robots: they had to be able to organize themselves.
However, what we are talking about here is not artificial intelligence. It is possible to say that it is a form of behaviour that is mostly based on instinct. When you look at nature, you see birds flying together: this is an instinctive behaviour and is not done consciously. There is also no central authority (a leader) that tells them what to do. However, thanks to this behavioural model, it is possible for creatures that remain in a group to react as a group (for example, to escape from danger or to counter it). Creatures that can act independently can come together for a specific purpose without having to take orders from anyone, and they can scatter again when that goal is achieved.
This is exactly what Beni and Wan wanted to apply to their robots: cellular robots would be able to come together for a specific purpose (for example, to attack a cancer cell) and then disband when they were done. Likewise, when passing through a capillary, for example, they would create the most appropriate formation on their own and achieve their goals without having to take separate commands. The goal was simply this: in nature, when independent creatures come together and create a herd (swarm), they instinctively form a common intelligence and can move together in synchrony. Is it possible to get this artificially?
Swarm Technology in Robotics
First of all, let’s say that swarm intelligence technology is still in a theoretical stage: no one has yet managed to develop anything that fits the above definition. But most elements of the swarm intelligence idea are currently used by the US military and NASA. The military uses the ideas on which this technology is based on unmanned vehicles, while NASA uses it for planetary mapping. Although swarm technology is an idea that can be used in many different industries, it is predictably best suited to the robotic industry: the idea of robots that can self-orient and analyze a problem and figure out how to solve it as a group is really exciting. Robots that use this technology are really small, and even when they come together, they cannot be noticed. Currently, swarm intelligence is used in the robotic industry for:
- Simultaneous Localization and Mapping (SLAM): A group of robots is used to map and localize a specific region. This is a tougher task than you might think. Robots are indeed very small, and they have to decide for themselves what geometric shape they need to create to counter the wind, for example. In addition, they should be able to combine and analyze information using range and visual sensors and identify what stationary landmarks are. The goal of this technology is to unleash a group of robots in unexplored places (such as caves) and make them return with a detailed three-dimensional map within minutes.
- Observation and Determination: In this scenario, robots are not used to create maps, but to detect and identify objects in a particular environment. They can be used, for example, to detect a specific document or weapon type: drop robots in a room through a small crack, and they’ll come back in a few minutes and tell you how many guns (and types) are in the room. Like the SLAM scenario, this scenario has many challenges: robots have to come together to observe an object (they literally take the shape of an eye) and have access to some kind of database to be able to classify the image – meaning they have to stay online all the time.
These examples look like military purposes, right? That’s right: the army is most interested in swarm technology, and therefore they determine the intended use. However, the information obtained from these studies can be used in all other sectors. In the near future, robots can be developed that will stay in our bodies for long periods of time and repair the damage they see. The collective intelligence of robots could one day become real artificial intelligence. We may be witnessing the beginning of a revolution.
No Comments