Quest for the Holy Grail of artificial intelligence
“Robotics is a subject of intense scientific interest in the USA and Europe. In Finland we also need to boldly go where no man has gone before,” says Associate Professor Joni Kämäräinen.
By 2050, robots will have learned to communicate with each other and a team made up of robots will be able to beat the reigning FIFA World Cup champions. “But this will not be possible without the seamless integration of sophisticated vision systems and robotics,” says Associate Professor Joni Kämäräinen.
“There are major challenges that the field has yet to face. The visual comprehension and information processing capacity of robots must be improved before humanoid robots can become a reality,” says Joni Kämäräinen, Associate Professor in the Department of Signal Processing at TUT. His research interests focus on advanced computer vision and pattern recognition technologies.
“The information we receive through our eyes is hugely important to us people, and much of our brain is devoted to vision. If I want to pick up a cup off a table, the visual cortex in my brain tells me where the cup is. But we’ve got a long way to go before we understand everything that’s involved in picking up cups without spilling or dropping them. That’s one of the Holy Grails of artificial intelligence, and whoever finds the answers will have the world at their feet.”
WHO: Joni Kämäräinen
- Leads a team of nine researchers who develop computer vision, pattern recognition and Image Google.
- Studied information and computer science at Lappeenranta University of Technology
- Graduated with a master’s degree in 1999
- Completed a doctorate in 2003. His dissertation was titled “Feature Extraction Using Gabor Filters”.
- Has previously held a professorship in the field of computer vision and pattern recognition at Lappeenranta University of Technology.
- Currently Associate Professor of Signal Processing (tenure track) at TUT.
In ten years’ time robots will live in our houses
According to Joni Kämäräinen, robots will make their way into our homes in the next 10 years. The first domestic robots will be used for entertainment purposes, but they will gradually evolve into household helpers who can take care of children and the elderly. Eventually they will become members of the family. Robots are able to learn and practise different tasks, but teaching them must be easy.
“We must be able to ask our robot to come and watch us make coffee. Then we can go to work and let the robot practice using a coffee maker and, if necessary, ask for more instructions from other robots through their own, parallel Internet. Robots will start communicating with each other and ultimately sustain a community like the human population.”
Developing the “Image Google”
Kämäräinen’s research group is currently working on a four-year project funded by the Academy of Finland. They are developing technologies that allow robots to detect and recognize objects in images in a human-like fashion. Kämäräinen calls the project Image Google.
“Our major goal is to do everything with a camera that people can do with their eyes,” Kämäräinen says.
The project involves teaching a computer to identify at least 10,000 different objects in digital images. As a result, users can select any image found on the Internet, and the people and objects displayed in the photo and the location where the photo was taken are automatically identified by a computer. This will make it possible to immediately find out who the lady standing at the right edge of a photo is, what she does, and where she has been.
“At least Google, Facebook and Microsoft see definite business potential in Image Google. The combination of image data will enable the creation of detailed profiles of people based on their interests. That’s valuable market information.”