Engadget pal Dave C. recently kicked it at the Intelligent Robotics Research Center in Seoul, South Korea, and wrote up this brief field report of his visit for us:
I was in Seoul recently and was lucky enough to arrange a meeting with Dr. Beomjae Yoo, Director of Intelligent Robotics Research Center at the Korean Institute of Science and Technology (KIST). He showed me around the lab, where I spotted two robots recently discussed here on Engadget before, a "male" bot named Maru, which means "pinnacle" or "apex", and a "female" bot named Ahra, which means "recognition". Unfortunately, I wasn't able to to see the robots operational, but there was also another, much larger humanoid model hanging in the hallway as well. I also spotted a room straight out of Blade Runner that was stocked full of assorted robot limbs and torsos.
Dr. Yoo focused a great deal on the networked brain KIST has developed for each robot. Each brain actually consists of four to five computer stations connected to a server, one each for voice/speech recognition, face recognition, object recognition, motion and movement, and general AI features such as speech (as Engadget mentioned, one server setup can control both bots). I asked Dr. Yoo whether the robots could do deductive reasoning (for example, if you fed them the information that A=B and B=C, can it figure out that A=C), and he said that they would have that capability by the end of the year. He said that some of the most difficult aspects of developing the robots were related to kinesthetics and movement, but that the hardest of all was creating the ability to adapt to a moving, changing environment, both from the physical response side and from the pattern recognition/AI side.
These robots are part of wider a project to build an Ubiquitous Robotic Companion, or URC. The URC is intended to be a robot butler that combines all the functionality of a cellphone and a computer with the abilities of a human concierge, so to speak. The part about the URC being ?ubiqutous? refers to their goal of having these bots be a available to everybody regardless of income, so there will be a range of models available. Dr. Yoo said that he expected non-humanoid household bots in to be widespread within three years, but that humanoid bots will most likely take a decade or more before they?re ready for primetime. He predicted that humanoid bots would most likely enter the workforce in basic jobs such as waiting tables and working cash registers, as well as basic manual labor and household jobs.