UniversityOfTsukuba
Latest
Qumarion 3D modeling mannequin coming soon for $750, still won't play with your kid (video)
Trying to get convincing, natural poses out of 3D models can be tricky, so it's a relief that two Japanese universities' joint ventures, the University of Electro-Communications' ViVienne and the University of Tsukuba's SoftEther, are close to wrapping up work on their posable mannequin. Now called Qumarion, the model formerly known as QUMA uses 32 sensors across 16 body joints to translate the humanoid statue's pose to the computer screen simply by bending limbs, much like you would the legion of action figures you had when you were eight. Neither you nor your kids will be using Qumarion to storm Fort Barbie anytime soon, but the 120 frames per second sample rate over USB does mean that poses are mirrored in your modeling tools almost instantly. You also won't have much longer to wait to buy one for your fledgling anime production: the mannequin and custom modeling software from Celsys should be bundled together sometime within the summer for a comparatively frugal $750.
Cyclone Display exemplifies 'multi-colored expression,' totally heading to a nightclub near you (video)
Ever heard of Yoichi Ochiai? You have now. Hailing from Japan's University of Tsukuba, this whizkid was on hand here at SIGGRAPH to showcase one of his latest creations -- and it just so happened to be one of the trippiest yet. The Cyclone Display was a demonstration focused on visual stimulation; a projector shown above interacted with a plate of spinning disks. Underneath, a cadre of motors were controlled by a connected computer, and as the rotation and velocity changed, so did the perceived pixels and colors. The next step, according to Ochiai, would be to blow this up and shrink it down, mixing textures in with different lighting situations. With a little help, a drab nightclub could douse its walls in leopard print one night, or zebra fur another. Interactive clubbing never sounded so fun, eh? You know the drill -- gallery's below, video's a click beneath. %Gallery-130394%
Gesture-controlled robot arm enables civilization's most meta high five
This video, criminally, doesn't actually show any high fives, but we're sure the students at the University of Tsukuba have sustained endless LOLs over the past few months, pushing their gesture-driven robot arm system to the limits of human-robot high five interaction. The system itself is relatively simple: it uses two cameras to track a hand's movements, including specific finger gestures, which are then processed and translated into robotic movement in real time. The end result is basically the world's most elaborate claw machine game, as demonstrated above.
Baby simulator surfaces in Japan, Devo wants its mascot back (video)
When we first saw Yotaro baby simulator we thought: "we've seen this ugly robot before." And no, weren't thinking about Keiko, the robot patient currently seeing action in teaching hospitals in Japan. It just so happens that robot in question has more than a passing resemblance to Booji Boy, a character beloved to Devo fans world-wide. The robot, developed at the University of Tsukuba, is designed to teach people about childcare and parenting by portraying various facial expressions, movements, and physiological traits associated with babies, relying on a Rube Goldberg-esque array of water tanks, projectors, sensors, and sophisticated emotion control software to respond to your actions appropriately. If you think it looks strange above, wait until you see it in action! Check it out after the break.
CirculaFloor robot floor tiles keep you moving in virtual reality
One of the big problems facing VR is the issue of mobility -- how do you allow users unrestricted movement in virtual reality, while keeping them relatively static in real reality? Omni-directional treadmills have been tried in the past, and now researchers at the University of Tsukuba in Japan have developed something called CirculaFloor. The system uses four robotic tiles that constantly shift position, ensuring that there's always a tile in the direction you're headed. Additionally, the entire assembly moves slowly backwards, giving one the impression of movement while they're actually standing relatively still. The tiles also incorporate lifts, for simulating staircases and the like. While this research is promising, there's still plenty of work to be done -- for instance, the tiles still move awful slowly, and while they seem sturdy enough for a leisurely stroll, a Left 4 Dead implementation is not likely to happen any time soon. Video after the break.[Via Tech Digest]
