DJI has just announced Manifold 2, an ultra-compact on-board supercomputer for DJI drones designed with the next generation of autonomous aerial robotics solutions in mind.

Manifold 2 replaces the original Manifold that DJI introduced back in 2015.
With the additional computational capability of Manifold 2, users can process complex image data collected on board and obtain results immediately. Drones can also be programmed to fly autonomously while identifying objects and avoiding obstacles.

Allowing the drone to receive and analyze data in real time, Manifold 2 opens new opportunities and features. It allows drones of the DJI Matrice 210 series equipped with Manifold 2 to fly independently where no one else has flown before, such as environments with prohibitive radio frequencies where most of the valuable and usable data reside.
By partnering with some of the industry’s leading processors and chip manufacturers, we are enabling DJI drones to become intelligent flying robots capable of performing complex processing tasks and advanced image processing literally on the fly.
Jan Gasparic, DJI Director of Strategic Partnerships at DJI
What does Manifold 2 consist of?

Manifold 2 provides powerful intelligence for a variety of applications: choose between an Intel Core i7-8550U CPU or an NVIDIA Jetson TX2 supercomputer on a borderline artificial intelligence module.

Both versions are compatible with DJI’s enterprise drones, including the Matrix 210 and Matrix 600 series. They can be integrated into the drone’s internal systems and sensors using the DJI software development kit; this allows for customization and integration into the DJI ecosystem of software and hardware solution providers.
Manifold 2 is compatible with the following DJI products:
Series Matrice 210
Series Matrice 600
N3 Flight Controller
Flight controller A3

Each Manifold 2 includes USB, UART, CAN and more ports, allowing developers to connect multiple Manifold 2 processors and stack them to turn DJI drones into specialized aerial robots for specific applications and industries.
Skynet?
While Manifold 2 is designed primarily for a myriad of different business purposes, such as on-site inspections, the technology could definitely be used for aerial shooting where a drone needs to navigate in very limited or hazardous areas.
Autonomous technology is moving very fast. Who knows how long it will be before drones start making their own decisions regarding the route or even what angle and position to shoot in, without the need for input from an operator.