go to contents go to side menu go to main menu

main menu area

main menu area

Innovative University Changing the World through Convergence

Embed to SNS

Research News

One Stop Control System Detecting Down Bad Drones

  • 조회. 472
  • 등록일. 2017.05.30
  • 작성자. Administrator

One Stop Control System Detecting Down Bad Drones

-DGIST Convergence Research Team Interview 

"Drones will be utilized in our daily life very shortly. Despite its benefits, we should think about the ripple effect the pervasive usage of drone will cause. We can’t rule out the possibility of privacy invasion and terror attacks with the negative use of drones. Our goal is to complete a system defending against any negative usage built on the relevant technologies," said JeonIl Moon, director of the Collaborative Research Center.

Dr. Moon is working for convergence research along with Dr. Daegun Oh for creating a system to detect and control low flying drones. Dr. Moon said, "The system will be a control tower of drones, so that we are making a key module and algorithm in order to manage the activity of drones comprehensively.

Multiples technologies are necessary for building the system including radar technology that detects a little flying objects in far distance, deep learning in artificial intelligence (AI) technology to identify if the flying object is bird or drone, sonar signal processing, and robot technology.

Currently, the center led by Dr. Oh pull together to ensure the radar technology detecting low flying drones. Dr. Oh said, “It is difficult to detect the little drone flying in the urban areas fully packed with multiple radio waves. It is harder than to find the missile flying over the sky in which clouds are rising up. Radar technology is used to detect only the reflected radio wave after shooting the waves. However, it remains a big challenge to detect the signal from the drones in the city in which tremendous obstacles exist.

In a way to address the issue, Dr. Oh succeeded in developing a dual channel high resolution radar technology used to identify flying objects within one km. The research result has been published on the IEEE Transactions on Aerospace & Electronic System Spectrum on May 26. (doi:10.1109/TAES.2016.140282) "I have improved the resolution three times more than the existing one," Oh said. "In the next three years, I will catch up with the world"s best and overcome it."

AI Key to Detect Drones

What if a flying bird is detected mistakenly as drone? Dr. Moon said, “There exists an obvious difference between the flying drone and a bird. However the issue is how to figure it out and that’s the big challenge.” He continues, “Deep learning and AI technology might be solution to address the issue, but it is almost impossible to collect the every flying pattern of all bird species.”

So, his team plans to increase the probability for making more accurate detection using deep learning technology.

Dr. Oh also said, "We know we have little experience yet and we have to make more experiments with other living creatures to raise the probability." He added that the team will work with radar technology specialists in the US who have worked for long on the AI. In addition, he will contact the bird trainers directly for further advancing the technology.

The DaVinci Robot Surgical System, which has been introduced in many hospitals recently, is a "cooperative robot" that moves the robot arm precisely according to the direction of human. The system leaves final judgment to human; robots only perform according to the manipulation of the human. A similar technique is applied for controlling drones. Dr. Moon said, "If unauthorized drones appear in the sky, they will act as cooperative drone robots that track down them quickly. We will advance the technology to prevent future risks by incorporating all relevant technologies from detection to disposal."