Faculty of Engineering
Opening New Frontiers through Advancements in Swarm Robotics
Scientists have been fascinated by swarm behavior for decades, though it originated in the natural world. It is still incredible to think of the complexities of an ant colony, or the accurate migration patterns that birds or fish exhibit. More recently, there has been a growing trends to adapt these patterns to machines, especially robots. KUAS’ Dr. Hiroaki Fukushima has made swarm robotics his major field of research. He is aiming for his research to contribute to major societal applications across various fields.
The reason why this field has been garnering attention is that recent advancements in technology have completely transformed what a swarm of robots is able to accomplish. The trend is from centralization to decentralization. Currently, centralized algorithms are used to control multiple robots in most cases. In drone shows, for example, target trajectories of all drones are determined in advance, so that the swarm of drones generate animations given beforehand. However, in applications where target trajectories or deployment of robots need to be decided in real time according to circumstances, such centralized algorithms become more and more difficult to be applied due to computational and communication loads. This is because, in centralized algorithms, trajectories of all robots are decided on one central computer based on the information collected from all robots and sent to each robot through wireless communication. Looking back towards birds and fish, they obviously are not given any target swarm shape. They decide their own trajectories based on the local information such as positions and velocities of neighbors, which generates various shapes of swarms as a result. In that sense, they use a decentralized algorithm. Decentralization makes control algorithms of robotic swarm scalable. In other words, the total number of the robots has little effect on the computational and communication loads, since movements of each robot are affected by its neighbors only.
Dr. Fukushima currently studies on navigation algorithms of a group of robots to a target point through areas filled with obstacles. In order to move robots through obstacle environments, the shape of the swarm need to be changed so as to avoid getting stuck or collisions in narrow spaces. Decentralization is necessary to achieve this in real time. One immediate practical application of such decentralized navigation algorithms that comes to mind is transportation, which can range from industrial applications (e.g. in factories) to the provision of medical supplies in a disaster area. Indeed, the extreme environments are where the field of swarm robotics has the greatest potential. In other words, if robot swarms can operate in such environments, they will reduce human risk and save lives. Applications of swarm robots can include the aforementioned transportation, but also demining, search and rescue, sensing (e.g. of vital signs) as well as cleaning up toxic spills. To come up with solutions for these applications, which take place in inherently unstable and unpredictable environments, is exactly what Dr. Fukushima is working on as he continues his research. Without a doubt, swarm robotics is another field of engineering with profound real-world potential.