Why Robot Swarms Look Set To Roam The Earth

In 25 years, swarm robots could do everything from taking out the garbage to cleaning your arteries, says Antonio Espingardeiro

Swarm robotics is a recent concept, inspired by nature. It involves the coordination of large numbers of simple, physical robots. With the help of clever software, a desired collective behaviour can emerge from the interactions between the robots and their environment. Swarms are self-organising and self-assembling artefacts that can communicate and learn, much like biological organisms.

For a quick introduction, here is a video by ITN explaining the basics of swarm robotics. And here, you can find a magnificent rendition of Monty Norman’s James Bond Theme, as performed by the team of tiny quadrotors.

TechWeekEurope got the independent robotics expert and IEEE (Institute of Electrical and Electronics Engineers) member Antonio Espingardeiro to explain what swarm robots are and why this branch of robotics matters.

Heart of the swarm  

What made you interested in robots in the first place?
Since my childhood, I had a fascination with machines, and ways to control them. I was a software engineer, but later I decided to apply for a Master’s degree in robotics. I ended up becoming an independent robotics researcher, and set up my own lab in 2010. Currently, I am working on social assistance robots. These could be used in elderly care and hospitals to help deal with huge demographic challenges that we will face over the next few decades.   

What are swarm robots, in a nutshell?
It is a very recent topic of research. It involves creating artificial intelligence systems that can communicate, share information and ultimately make decisions among robots. This is an emergent collective behaviour, for machines.

Here’s an example: imagine that you have ten little robots exploring a room, where there are several obstacles. Let’s say that one of the robots has found an obstacle in the corner of the room. He will immediately tell the others he has found an obstacle, transmit coordinates and tell them to stay away. In a short time, all of the robots will know where all of the obstacles are.  

What recent developments have advanced swarm robotics?
Over the last few years we have made huge progress. Sensors have evolved, and become cheaper. They are still not cheap enough. Also, the necessary computing power became affordable. But there are still a lot of things that need to be done in terms of hardware. At the moment, swarm robots are interesting to look at, but they can’t really do anything useful.  

How did the concept of swarm robots emerge?
If you look at biological systems in nature – flocks of birds, schools of fish, ant colonies – they organise themselves in certain ways, to look for food or defend themselves. Researchers were inspired by this and thought they could do the same with machines. When you are struggling with a certain problem, it would be easier if you could sub-divide it, and deal with smaller problems, one at a time.

Swarm robotics is also inspired by neuroscience and the way we learn. Scientists are trying to convert learning into an algorithm and use it to program robots. It is closely connected to the AI research.  

Isn’t that a little bit dangerous?
We are not talking about true intelligence, but the ability to perform certain tasks more efficiently. Overall, you can expect more and more autonomous systems in the future.  

Building the future  

Where could swarm robots be used?
Using these developments, cars in the future could communicate between themselves, share information and increase the safety of pedestrians and other drivers. Submarine swarms could scan the bottom of the ocean for debris, others could work in space. Then there are nano-robots that could be one day injected into our bodies to clean arteries and fight diseases.  

Could swarm robots adapt to different environments?
It is only a concept at the moment. In biology, it is called morphogenesis, and it could allow robots to acquire different shapes. They could assemble in a stiff line to form a bridge, or become flexible to climb a wall. Researchers have been creating swarm robots that can do both. In the next 20 to 25 years, we could see these systems being used in our everyday life.  

TechWeekEurope recently wrote about the DARPA (Defense Advanced Research Projects Agency) competition that is looking for robots to work in disaster zones. Could a swarm robot design be a winner?
I don’t think that will be the case. At the moment, all the funding and effort is spent on hardware. The first assessment will take place in December 2013, followed by another one in 2014. Right now, the participating teams are focusing on building the platform – hydraulics, pneumatics, etc. Once a good platform is built, they have to prove it works, and deal with the challenges. The challenges include walking across rubble, driving a utility vehicle, opening and closing valves, dealing with lack of communication and the like.

By 2014, the competition will be more focused on developing the brain of the robot, making it think.  

So when can we expect robots to start walking the streets?
I give it 20-25 years. Communication is a big problem. You have to imbue the machine with enough intelligence, so if you lose communication, it would stop working or return to a docking station, and definitely not harm anyone. We need even more computing power, more miniaturisation. Besides, there’s the problem of social acceptance: people are afraid of robots. That needs to change.

With swarm robotics, your imagination is the limit. Because when you get something working, you can apply it to different areas. Imagine robots monitoring pollution, or taking care of your recycling. We just need to keep pushing the boundaries.

Think you know patents? Test yourself with our quiz!