Czech Technical University team working on animal inspired drone swarms
The Multi-robot Systems group (MRS) at the Czech Technical University in Prague has been busy testing drone swarms in the deserts of the United Arab Emirates over the past few weeks. The project seeks to explore the possible application of such massed drones in humanitarian search and rescue operations. To find out more, I spoke to the man in charge of the MRS group, Dr. Martin Saska.
“We were doing two projects there. The first one was focused on the swarming of micro-aerial vehicles. The task was to coordinate robots flying fully autonomously over the desert. They were tasked with exploring the area and trying to find some objects in the desert. They were pre-trained to the neural network that was carried on board of the vehicle.
“We tried to miniaturise the drones, because if you have a fleet of really small micro-sized vehicles… they are less of a danger to people.”
“The second task was to explore a multi-floored building. The purpose of this research is to provide a technology that would be able to provide search and rescue fast in a tall building. The advantage of drones is that they can reach the location really fast when compared to people, ground robots, or land vehicles. They can fly into the nearest window, enter the building and then explore its rooms, either finding humans there, or help extinguish fires for example.”
You also said that you were inspired by the behaviour of birds and fish for this research. How exactly?
“Yes. The motivation from nature was mainly taken in order to design the perceptions of the drones. Each fully autonomous drone was equipped with on board cameras that behaved in a similar way to the sense organs of fish and birds. They are able to detect their neighbours in a swarm and to react to this in a fully decentralised way. They may do the swarming or flying in a fully compact formation using only this local sensing, without communication, GPS or any other external resources.
“This makes the swarm very reliable and able to work in different conditions and environments independently to its external infrastructure.”
How big are these drones that we are talking about?
“They come in different sizes. We tried to miniaturise the drones, because, in my opinion, if you have a fleet of really small micro-sized vehicles, that means even less than 200 grams, they are less of a danger to people.
“For me it would be more acceptable to see these really small drones flying in cities, for example measuring the environment, temperature, or whatever you need. This is better than seeing a really huge drone, fully equipped with perfect sensors but weighing as much as 10 kilograms, because this is really dangerous.”
200 grams. I guess that’s about the size of a small bird, right?
“Exactly. This is our target, I would say. Even the European Commission now decided that a drone below the limit of 250 grams, if I recall correctly, would be considered less dangerous and would find it easier to get permission to fly autonomously.
“The drones that we tested in the desert were quite bigger, about three kilograms. However, as I said, our intention is to get to smaller dimensions in the future.”
You said that your research could also be followed up on by Czech companies. How does the Czech Republic compare internationally when it comes to businesses developing and applying the use of drones?
“Each fully autonomous drone was equipped with an on board cameras that behaved in a similar way to the sense organs of fish and birds.”
“I believe Czech companies are well known in the worldwide drone business. We have quite a lot of companies involved in drones.
“I hope that our contribution to this effort is also quite significant. We already have quite a large group of researchers at our university. We also closely cooperate with other companies and try to apply the fundamental results of our research into applications. This also highlights the results of the Czech Technical University worldwide, because its products will already be quite well known.”
There is much talk in military circles about the use of swarm drones in future combat. Now, as your research shows, there are also attempts to explore its use in humanitarian efforts. Do you see any other areas where such swarms could be useful?
“Yes. You are right that military application of drones is the first thing that everyone things about. However, in my mind, the deployment of larger drone groups is useful in all kinds of human activity.
“As I said already, it is much more useful to deploy a large swarm of less equipped drones and try to replace larger vehicles. Search and rescue is perhaps a nice example, because you need to explore the area as fast as possible and you will get that much faster by deploying multiple vehicles, rather than one, well equipped drone.
“This can also be used in any other task where you need to take measurements. For example, imagine that you would need to measure pollutants around a factory, in a city, or after some sort of natural disaster. Deploying multiple vehicles in the field can provide you with much more information faster, measured in different locations in real time. Another example is if you want to track the motion of the target that you are addressing.
“We have a project where we measure radiation. Using swarms, we can detect radioactive objects even if they are moving, because we can measure the radiation from different positions instantaneously. This can help you estimate the movement of the vehicle. These are all areas where drone swarms are useful.”
How does one programme the drones to work in swarms?
“Let me try to focus on one aspect. As I said, our drones are fully autonomous. That means that they use on board sensors for their operation in the environment and are able to respond to their input. That means that they can also avoid dynamic obstacles, such as a predator that would attack the swarm.
“Deploying multiple vehicles in the field can give you much more information faster, measured in different locations in real time.”
“These actions are done without any human control. We do not use any operator. There may of course be some human in the loop if it is required by the mission. However, if not, these swarms really are fully autonomous and behave like a robot. That is the difference between our research results and those normal drones that you see on the market.”
How soon do you think that this can move from the research and testing phase to a phase when we will see these drones deployed in real search and rescue efforts?
“In the area of drone robotics we can see a really fast deployment of fundamental research technology. If I compare it with my experience with ground robots it really is much faster. Usually when I see some results of fundamental research at a conference, I know that it will be available even in toy shops in a space of two years. It is a really fast process.
“We are cooperating on the development of a drone that hunts other drones. This is to protect airports, prisons, or other parts of a country’s critical infrastructure.”
“We already have some projects where we are working on that in a real product. We are cooperating on the development of a drone that hunts other drones. This is to protect airports, prisons, or other parts of a country’s critical infrastructure. Some clients require the deployment of more drones simultaneously, because they may be the receiver of a simultaneous attack of multiple drones. They need multiple agents to counteract this. We are already applying the multi-robot cooperation and coordination.”
