The observation of swarms in nature, such as ants in colonies or birds in flocks, inspired the study of swarm intelligence which examines how individual agents following simple rules can develop a sophisticated global behaviour. In London last week, the ReWork Future of Technology Summit brought together engineers and scientists to discuss advances in swarm intelligence and robotics. We captured the highlights for those who may have missed the event:
What is swarm robotics?
Swarm robotics looks at how individually engineered robots, each behaving autonomously, can self-organize and achieve a goal, explained Roderich Gross of the University of Sheffield. Each individual robot is given a certain amount of information allowing it to interact with the others. By limiting each individual robot’s information, we are able to make a swarm aggregate, cluster and transport objects, ultimately to solve a larger problem.
Why is swarm robotics significant to medicine?
Nanoparticles are important in healthcare because they can leak out of blood vessels and go to target sites, explained Sabine Hauert of the University of Bristol. These particles are too small to program but they can be made to swarm by changing their coating, charge or size. If a group of nanoparticles can swarm, they can potentially travel to target cancer cells, contain a coating of medicine, be activated by external energy and destroy targeted tissue effectively. Whereas drugs simply diffuse into tissues, swarm nanoparticles would have more intelligence and could more effectively target diseased tissue.
What role does swarm robotics play in healthcare today?
At the moment, we are thinking about how to engineer a swarm system to interact with the individual agents, and between the agents and the environment, explained Gross. “How complex should these agents be? Do these agents need a brain? How do we bring four agents together, then 40? How can we solve these tasks in an environment that does not contain any clues?” asked Gross. So far, Gross has used binary sensing to bring agents together. If a robot is in the line of sight of another one, then a certain fixed motion pattern program is activated. If not, then another fixed motion pattern program is activated. In this way, Gross has clustered and transported objects with robots using very little energy. The benefit of this system in healthcare is its simplicity, reconfigurability and verifiability.
What still needs to happen for this technology to reach its full potential?
We need to study the design of the nanoparticles such as the materials and the drugs to bind to the nanoparticles, explained Hauert and Gross. The nanoparticles will need to sense each other, without communicating using language, which will require significant processing power. Additionally, the nanoparticles will need to be relatively inexpensive and scalable. Hauert tackles some of these problems using open source software, nanodoc.org, a site that allows the crowd to solve problems in swarm robotics.
We are seeing a lot of intelligence in artificial intelligence, what happens when we see intelligence in an entire swarm? Will it become unpredictable?
The truth is that we are still far away from giving swarms intelligence on an individual level; we want to see how intelligence on a large scale in a swarm can solve problems, stated Hauert. On a philosophical level, whenever we create a society, a new culture emerges, and this may eventually apply to technological systems as well.