It’s still primitive, but the Kinect seems to be the best consumer-side telepresence and remote-operation device to hack right now. In fact, in a playful toy sort of way, we’ve just come across a video on YouTUBE that shows a user controlling an Aldebaran Nao humanoid robot with the Microsoft Kinect. I’ve mentioned before why remote operation interfaces are so important—and why NASA should be getting in on this already.

As you can also see, this is the expected outcome from the NAO robot developer program that Taylor Veltrop participates in and wrote the code to bring this to bear. Be sure to watch the video of Halit Berner Suay (in video) discussing his code.

Controlling Nao the humanoid robot using Microsoft Kinect depth camera. The main purpose is to control humanoid robots (Aldebaran Nao in this case) in the most natural way possible, without wearing any kind of device, using arm and leg gestures.

The second purpose is to make Human Robot interaction, and teaching new tasks to robots easier, without requiring and robotics knowledge by the user, which is a contribution to Learning from Demonstration field.

We do have a long way to go from being able to detect particular poses and mimicking human gestures and motions with tiny toy robots to something that could articulate actions in a waldo-and-chassis robot setup, but it’s a huge leap if garage engineers are bringing this into the world. Imagine how useful this might be for hazardous robot operations if a user could just step in front of a camera and remotely operate a robotic arm or remote vehicle just by moving their arms. No bulky interface-equipment to lug about, just a camera and a laptop and a space to stand or sit in.

For those interested, the code that runs this demonstration is available online.

Link, via Kinect Hacks.