Vision processing and AI improve control of drones

November 03, 2015 // By Julien Happich
Chinese drone maker DJI has partnered with open source Ubuntu OS provider Canonical to release a computer embedded on a drone platform, helping developers turn their drones into intelligent, autonomous machines capable of performing complex computing tasks on their own.

Dubbed the DJI Manifold, the computing platform packs an NVIDIA Tegra K1's 4-Plus-1 Quad-core ARM Cortex-A15 Processor and 192 GPU CUDA cores with clock speeds of up to 2.2GHz, providing any equipped drone ample graphics computational power to process high resolution images in real time. The GPU cores support a throughput of 326 GFLOPS.

Mounted to the expansion bay of the company's Matrice 100 drone platform, the 110x110x26mm unit weighs less than 200g and could be deployed for use in artificial intelligence applications such as computer vision and deep learning for aerial monitoring applications.

Such computing capabilities could enable the drone to sense their surrounding environment but also identify objects and react accordingly.

DJI Manifold runs on Canonical’s Ubuntu operating system. It supports CUDA, OpenCV and ROS.

Related news:

Parrot takes instant 3D mapping to the sky

3D camera-computer does all the vision processing you want