Intel shows devices getting smarter all the time with ability to learn
Herb finally can make it out of the kitchen. The personal robotic assistant that scientists at Intel Research Pittsburgh have worked on since 2006 is smarter and more confident now, and can do more than move cans around.
The machine that combines a Segway transporter with a robotic arm could maneuver through any cluttered family room to find a mug filled with coffee and take it to someone. And through technology that allows it to learn, “Herb knows you can’t wave the coffee around, or it will spill,” said Sidd Srinivasa, a senior research scientist at Intel Corp.’s research center in Oakland.
The company hosted its annual open house Thursday to show off 58 projects that its 20 researchers are working on with Carnegie Mellon University and University of Pittsburgh faculty, plus students and University of Pittsburgh Medical Center representatives.
The research center on CMU’s campus opened in 2002 to work on technology-based projects that could be five to 10 years from some form of actual use. Santa Clara, Calif.-based Intel is the world’s largest maker of microprocessors that are the brains in personal computers.
Srinivasa said Intel’s efforts to improve personal robots are like making building blocks for a variety of futuristic devices. Intel’s Seattle office, and CMU and University of Washington students are involved in the project.
“A lot of the technologies we are developing for Herb — the perception and face recognition, for example — you will see appear in your home much earlier than you think,” he said. TVs in a few years might recognize a child who enters a room, and automatically set parental controls, for example.
The robot uses a camera to recognize objects it’s seen before. It adds more things — or people — to its database as it makes its way around a house. It uses a spinning laser to create a three-dimensional picture of its surroundings.
Herb — short for home exploring robot butler — was the life of the party yesterday, taking drink orders through a Web interface from many of the 500 or so visitors. During a lull, it gathered up empty cans and tossed them into a recycling bin.
Intel’s guests had an opportunity to try Pin Point, a game built to demonstrate gesture recognition technology. Players moved their arms to pop balloons on a screen — no game controller needed.
The technology could be used someday to point to a light and turn it on, for example, researcher Rahul Sukthankar said.
CMU’s Seth Goldstein and Intel’s Jason Campbell demonstrated Blinky Blocks, Lego-like pieces that contain tiny computers.
They communicate and change color as they’re put together in different ways. They can be a learning and research tool for students learning to program distributed systems of computers, such as a power grid might contain.
“We want to make them inexpensive enough so that a school anywhere can buy a few hundred of them,” said Goldstein, an associate professor of computer science.
CMU senior Max Salley showed off a project designed to help confused restaurant customers. Using a phone with a camera and Internet access, he snapped a picture of a menu board, honed in on a listing for a chicken sandwich and sent it to the Intel lab.
Through image-matching, a rundown of the calories, fat and other nutritional details appears on the phone screen, along with a picture of the sandwich. “This allows us to make a much more informed choice,” Salley said.