Lets promote our nations innovations! Click here

Follow us for regular updates on Australian and NZ innovation  

Skydio’s self-flying AI drone is now open to app developers

Skydio’s self-flying AI drone is now open to app developers

 Photo by Vjeran Pavic / The Verge

Skydio, a Redwood City, California-based startup that makes a drone that flies and films 4K video footage entirely on its own, is announcing today that its introductory device, the Skydio R1, is now open to third-party app developers. The Skydio Autonomy Platform, as the company is calling its developer toolset, will let app makers and drone enthusiasts develop custom software that takes advantage of the device’s bevy of cameras and sensors, as well as its sophisticated computer vision software and machine learning algorithms.

The company is mainly providing two software development kits as part of this effort: one is a Skills SDK, while the other is a pair of iOS and Android mobile SDKs. The first will let developers access the Skydio R1’s “autonomy engine,” which is basically the blend of software and hardware that lets the drone understand objects and perceive its surroundings, avoid obstacles as it maneuvers the environment, and follow the commands it receives from the companion mobile app. The other SDK will let developers craft mobile apps that control the drone and any custom skills the developer is utilizing.

Video footage filmed using a review unit of the Skydio R1 in April 2018 that showcases the drone’s obstacle avoidance and target-tracking capabilities.

Because the R1 has 13 onboard cameras as well as depth and motion sensors, it’s able to take in data about its environment, map that environment in real time, and move through the world while avoiding obstacles and tracking subjects you want it to film.

All of this is done through proprietary vision algorithms Skydio has created using artificial intelligence training techniques, specifically the variety of machine learning and deep learning that has underpinned much of Silicon Valley’s advancements in object and image recognition in the last half-decade. (Skydio’s co-founders were former early members of Google’s drone delivery initiative, Project Wing, and both studied and helped create autonomous flight systems at the Massachusetts Institute of Technology with a focus on computer vision.)

With access to the autonomy engine, developers will be able to define the parameters for how Skydio takes off, tracks subjects, and 3D maps its surroundings. Developers will also get control of even granular data like takeoff-relative position, user location, battery levels, and temperature. Skydio envisions developers creating apps and skills that take advantage of certain cinematic styles, so the Skydio can become a proper autonomous floating skycam. In May, the company added vehicle tracking to the R1, opening up the possibility that more advertising agencies, production companies, and other businesses and commercial operators could start using the R1 to film professional video.

To show off its Skills SDK, Skydio created a set of new cinematic shooting modes for the R1, including a cable cam mode to make the R1 retrace a specific flight path between two points and a series of fancy single-shot action modes that Skydio says are designed to make shareable social media clips. The company has already sold out of its first run of R1 devices, which it manufactured in Redwood City and labeled its Frontier Edition. Now, the company says it’s shipping the mass-production version of the drone at a reduced price of $1,999, which is down from the Frontier Edition’s $2,499 price tag back in April.

                 Courtesy; www.theverge.com