All News | Boards | Chips | Devices | Software | LinuxDevices.com Archive | About | Contact | Subscribe
Follow LinuxGizmos:
Twitter Facebook Pinterest RSS feed
*   get email updates   *

Latest RealSense camera adds tracking smarts to robots and drones

Jan 24, 2019 — by Eric Brown — 837 views

Intel’s Linux-compatible RealSense Tracking Camera T265 for autonomous robot and drone development is built on its Myriad 2 VPU. The dual-lens, 6DoF T265 camera requires no external sensors for V-SLAM localization.

Intel opened $199 pre-orders for its first Intel RealSense camera equipped with its Movidius Myriad 2 visual processing unit (VPU). Even more so than previous RealSense cameras, the $199 Intel RealSense Tracking Camera T265 is squarely aimed at robots and drones, although it can also be used for augmented reality gear.



Intel RealSense Tracking Camera T265, front and back
(click images to enlarge)

The Myriad 2 vision processing chip drives a dual-lens stereovision camera, but it lacks the usual depth sensing found on RealSense cameras. Instead, it enables inside-out tracking, which means it does not rely on external sensors or GPS to understand its environment and where it is located. This enables drones and robots that can operate autonomously in remote areas, including agricultural robots and medical supply drones, as well as low-cost robots that operate in warehouses without a good GPS signal.

Due to the low power draw and small footprint, the 108 x 25 x 13mm, 55-gram camera can be used with less sophisticated autonomous devices. “The only hardware requirements are sufficient non-volatile memory to boot the device and a USB 2.0 or 3.0 connection that provides 1.5 watts of power,” says Intel.

Intel claims the RealSense T265 is the only inside-out tracking solution with 6-degrees-of-freedom (6DoF). It gathers inputs from an IMU and two onboard fish-eye cameras with OV9282 imagers, each with a 163-degree range of view. A micro-USB 3.1 Gen 1 port and dual mounting sockets are also available.

The camera’s V-SLAM (Visual Simultaneous Localization and Mapping) algorithms construct and continually update maps of unknown environments and the location of the camera within that environment. All position calculations are performed directly on the device.



RealSense T265 dimensions
(click images to enlarge)

The RealSense T265 allows less than 1 percent “closed loop drift under intended use conditions,” says Intel. It’s also claimed to offer sub-6ms latency between movement and “reflection of movement in the pose.” The camera is said to perform well at light levels as low as 15 lux.

The lenses are equipped with infrared cut filters, which enables it to ignore the projected patterns from Intel’s most recent RealSense D400 series depth cameras. This enables developers to use both devices together for applications including occupancy mapping or collision avoidance and navigation in locations where GPS data isn’t available.

At launch, the camera’s SDK will support Ubuntu and Windows. Early customers have already used the open source Host API library to port to Android.

Intel recently announced a new version of its Myriad chip called the Myriad X, which is integrated in its Neural Compute Stick 2. The Myriad X adds a Neural Compute Engine and more cores for up to 8x greater performance.

 
Further information

The Intel RealSense Tracking Camera T265 is available for $199 pre-order, with shipments expected Mar. 4. More information may be found on the RealSense T265 announcement, product, and shopping pages.
 

(advertise here)


Print Friendly, PDF & Email
PLEASE COMMENT BELOW

Please comment here...