Sighthound Inc. has announced the release of Sighthound Sentry D, a system allowing drones to follow people or objects autonomously.
According to the company, the system is a version of the company’s Sentry computer vision engine but is specifically designed for unmanned aircraft systems (UAS).
Sighthound Sentry, deployed on embedded devices or computers or in the cloud, uses computer vision to interpret the video stream from a camera feed. The Sentry D variant has been specifically adapted to be able to detect, classify and track people and objects from moving cameras.
Sighthound says it has worked with Ambarella Inc. – a provider of low-power, high-definition video compression and image processing solutions – to deploy Sentry D on Ambarella’s A9SE 4k Ultra HD platform. As a result, Sentry D can process the video stream from a drone’s onboard camera to look for humans and objects and accurately track them.
The output of the computer vision algorithms is then fed to the UAS’ flight control systems so that the aircraft can autonomously follow the person or object being tracked. The result is the ability to follow and view the subject more closely, as well as with better camera centering than current GPS based approaches provide, in environments with little or no GPS signal, and with greater flexibility than hardware tethered alternatives, says Sighthound.
“The Sentry D computer vision engine enables highly accurate person and object tracking whilst the camera is moving,” explains Stephen Neish, CEO of Sighthound. “Combined with the advanced imaging and video capabilities of Ambarella’s latest Ultra HD A9SE system-on-chip, it will enable autonomous follow-me capability in the next generation of flying cameras.”