IDF 2008: Intel shows robots that can clear cups and lift apples

If you want a robot that can help around the house or the office, one of the basic things it needs to be able to do is fetch things. That's why the robots that Intel's personal robotics team showed off at IDF concentrated on moving around and picking things up.

"The first challenge is navigation," said Intel research scientist Siddhartha Srinivasa. "Our homes are filled with clutter chairs, tables moving objects like people." Herb the Robot uses a blue laser to navigate and a pair of cameras in the "elbow" of its grasping arm that feed data to an eight-core processor running real-time Linux to find mugs to clear away.

Picking things up precisely without knocking them over is another challenge for robots according to Srinivasa. "We want robots to be able to open doors and cabinets, to pick up complex objects like keys or a coffee mug."

To let the robot pick up delicate objects, senior research scientist Joshua Smith fitted it with electric-field pretouch sensing, which is used by some fish to detect objects by generating a weak magnetic field and sensing interruptions in it. That lets the robot grasp an apple without dropping or crushing it Smith let the robot get a grip on his elbow as well; and because the robot can detect conductivity it knows when someone takes hold of the apple and hands it back.

That is how these latest robots differ from existing devices, said Srinivasa. "Look at the Roomba: it wanders around semi-stupidly and it's ok for it to bump into stuff. You certainly don't want an algorithm like that picking up your china. Personal robots in the $200-300 range are all mobile. If you want to actually affect the environment, that's been very hard." To get general purpose robots smarter, they need a better processor like Atom, said Smith. "Give me twice the processing and I can make the robot go faster, I can have better models of objects and do better sensing."

Whether you would trust it just yet with a steaming mug of tea is a different matter; luckily that's the next problem the personal robotics team is working on.