Robots that behave like a swarm. A swarm whose one or more robots detect something and then collectively take an action. Future music? Scary?
We talk on the Campus of RDM Rotterdam with Wilco Braam, senior lecturer in Research & Innovation at Hogeschool Rotterdam, about so-called 'swarming robots'. "Imagine," he says, "a central water drone, an aquabot, which has a swarm of other aquabots like the chicks around it like a mother duck. Together, they navigate a river, mapping, for example, the soil conditions of that river. If a member of the swarm, whether it is the mother duck or one of the chicks, notices something, the rest are notified."
The above is a simplified representation of the concept of swarming robots. The idea comes directly from nature, where social insects and birds can exhibit a kind of controlled swarming behaviour. The central insect or bird, and in this case the central aquabot, points the way to the other insects, birds or aquabots. In the beginning, the 'chicks' work semi-autonomously, but over time they can do their work completely independently. So they are self-learning.
Applying tricks from nature in everyday life, so-called bio-mimicry, is of course nothing new. Think Velcro (based on the hooks on the ends of the leaves of the - no joke - Klit plant) and the Fastskin shark suit worn by several swimmers at the 2000 Olympics. Inspired by the skin of a shark and also not allowed after those Games. Swimmers who wore such a suit were swimming too fast!
On land, sea and air
The applications of swarming robots are many. Apart from mapping river bottom conditions, swarming robots are perfectly suited, for instance, to be 'released' into a river and fish them out again a few kilometres down the road. In those few kilometres, they have gathered an awful lot of formation about things like flow and water pollution.
Another application of swarming robots is their use in disaster areas. Robots can get to places where rescue workers' lives are not safe. They can use infrared sensors to determine whether there are any living victims at the scene. With the information they collect, rescuers can determine the risks for a rescue operation.
In the air, robot swarms can measure air quality. Or be used in burglary prevention. And on the road, there are even plans to run autonomous trucks in Rotterdam Europoort, between the First and Second Maasvlakte and the terminals, using the platooning principle.
Of course, there are also some more controversial ways to use swarming robots. For example, they can form an autonomous army. The US Navy has tested a swarm of autonomously sailing robots that can steer and take offensive action, without outside intervention. These kinds of swarming robots can be armed to deter and possibly destroy enemy ships.
In all examples, the same applies: The constant flow of information between the members of the swarm builds up a huge amount of Big Data.
What does Wilco Braam actually think about the discussion on the danger of self-learning robots? "This discussion cannot start early enough. Humans must remain important to control and master technology. The processors and algorithms of self-learning robots have enormous power and speed, it is increasing. Man cannot handle that power. As an example, a few weeks ago, a self-learning algorithm beat the best go-players in the world."
He continues: "At our campus, it's about developing fairly 'dumb' robots that can perform simple tasks, but with great impact. We are also very much into circular economy: what is the impact of robots on the environment and what does it cost if something goes wrong."
By: Walter Kooren
