Intel’s New RealSense Camera Adds Positional Tracking To Any Headset For $199

by David Heaney • January 24th, 2019

Intel today announced a stereo camera module which performs positional tracking. The RealSense Tracking Camera T265 can be preordered today for $199. It’s intended for robotics, drones, and head mounted displays.

Unlike all other RealSense products, the T265 is not a depth sensor. It contains two cameras with fisheye lenses and global shutter sensors. It also features an IMU containing an accelerometer and gyroscope. An onboard image recognition chip, the Myriad 2, runs Intel’s proprietary SLAM algorithm.

intel realsense slam camera

SLAM

Simultaneous Location And Mapping (SLAM) algorithms are the kind used for camera based inside-out tracking. They work by noticing unique static features in the room. By comparing the rotation and acceleration from the IMU to how these features appear to move, the position of the camera can be determined.

SLAM algorithms are used on the Windows MR headsets, Lenovo Mirage Solo, and HTC Vive Focus, as well as the upcoming Oculus Quest and Vive Cosmos.

intel realsense slam camera

The T265 is roughly the size of a USB stick. It weighs just 55 grams and consumes just 1.5 watts of power. This means even the USB port of a headset like the Oculus Go could power it.

The inclusion of the dedicated image processing chip is thanks to Intel’s 2016 acquisition of Movidius. The Movidius team makes ultra low power specialized image processing chips smaller than a coin. The Myriad 2 has also been used in products like Google Clips.

You may be wondering why the existing RealSense Depth cameras aren’t suitable for SLAM. Those cameras have significantly smaller field of view so can’t see much of the room. They’re designed for depth data. Furthermore, because they don’t have the built in Movidius chip the PC or headset would have to do the image processing.

The T265’s price and the wide availability of positional tracking today means it won’t have wide appeal in the consumer VR sphere. But for hobby projects and hardware prototyping this plug-and-play positional tracking solution will be helpful.

Tagged with: , , ,

What's your reaction?
Like
Wow
0%
LOL
0%
Dislike
0%