*October 1, 2019, Skydio introduced its new smart product for realizing action shooting from the air. Learn more about Skydio 2 in our review.
About the developer
Skydio R1 is the result of long and painstaking work, a team that includes graduates from the Massachusetts Institute of Technology in the field of aviation and robotics, as well as world experts in the field of AI (Artificial Intelligence) and robotics, who managed to establish themselves in such large companies as Google, Apple, Tesla and Oculus.
- Skydio was founded by Adam Bry, Abe Bachrach, and Matt Donahoe.
- The company is supported by such investors as: a16z, Accel, IVP, Playground, NVIDIA, Kevin Durant, Naval Ravikant, Elad Gil, Padma Warrior, Justin Timberlake, Magic Johnson.
Before the advent of the R1 drone, the most practical obstacle avoidance system was equipped with UAVs from DJI Innovations and Yuneec. More recently, DJI has raised the bar in this direction with FlightAutonomy 2.0, a system first implemented on the Mavic Air quadcopter. It would seem, what could be better than what we already have? Skydio developers answered this question revealingly, presenting their advanced obstacle avoidance technology called the Skydio Autonomy Engine.
Skydio Autonomy Engine
Skydio R1 takes obstacle avoidance to a whole new level. The R1 uses 12 cameras built into the drone’s frame to create a 3D map of the world around it that includes trees, people, buildings and more. For operational real-time processing of incoming visual information, the top full-featured computing platform for artificial intelligence was used — Nvidia Jetson TX1, built on a 256-core graphics processor, the one used in modern cars with autopilot.
How it works?
The first step in autonomous flight is to get the drone to understand where it is in the environment and how it is moving. This is made possible by the 3D rendering, which uses highly textured areas as markers. Thanks to triangulation, it becomes possible to calculate the position of the drone.
*computer vision R1
The next step in processing is to create a 3D understanding of the world around the drone. To do this, Skydio R1 calculates the depth of the stereo image coming from each pair of cameras, which are then combined to create a dense three-dimensional understanding of the surrounding world.
According to co-founder Adam Bry, the drone uses a combination of deep neural networks to recognize the people around it. The drone knows what people look like and creates a visual identifier for each person individually so that they can later focus on the correct object.
The drone movement planning system turns all the received information about the environment into correct behavior, thus allowing you to avoid obstacles in a balanced way, keeping the object in the frame, capturing amazingly smooth video, as well as predict the movement of the drone (up to 4 seconds ahead) based on the movement of the object, for which are being followed.
*Top image: R1 computer vision in predictive mode. Bottom image: normal camera view.
Adam Bry noted that, despite the intelligence of the drone, there are still situations where the drone can run into difficulties. For example, thin wires or branches can be difficult to see, glass surfaces or crowds of people can also confuse the drone. If R1 cannot understand what to do next, he will immediately report this through the smartphone that controls it.
16min (two batteries included)
4K at 30 fps; 1080p at 30.60fps;
64GB built-in (1.5 hours for 4K and 4.5 hours for 1080p at 30fps)
2‑axis mechanical (1 axis flight control)
256-core Nvidia Jetson GPU
4‑core 64-bit ARM processor
12 cameras for omnidirectional vision
1 camera for custom shooting
Video link resolution:
WiFi Removal Distance:
90 meters (depending on smartphone)
Skydio R1 has 5 smart follow modes:
Skydio R1 received a 4K camera (4k at 30fps; 1080p at 30.60fps; FOV 150°) suspended on a 2‑axis mechanical gimbal.
According to the specification, the flight time is 16 minutes. By default, two batteries are included in the kit.
The R1 is made from aerospace-grade materials including lightweight aluminum and carbon fiber.
Full control over the drone can be realized through a smartphone through the Skydio application. For Apple Watch owners, the ability to issue commands to the R1 directly from the smartwatch is available, thereby eliminating the need to take out the smartphone every time.
The production and sale of the model has been discontinued. On October 1, 2019, Skydio introduced their new smart product for realizing action shooting from the air — Skydio 2.
Photo with Skydio R1 drone.
Video with Skydio R1 drone.