Zero Zero's Camera Drone Could Be a Robot Command Center in the Future

Hover Camera follows you by face recognition. Zero Zero Robotics says its camera drone may evolve into the eyes of a home robotics system

5 min read
Zero Zero Robotics' Hover Camera camera drone follows your face to take photos.
Hover Camera, Zero Zero Robotics' new camera drone, uses face and body recognition to follow you around.
Gif: Zero Zero Robotics

Startup Zero Zero Robotics just took the wraps off its eye in the sky, the Hover Camera. The company hasn’t set a price but expects the lightweight drone (it weighs in at 240 grams) to cost under US $600.

The flying camera is a relatively new type of gadget. It all started about a year ago, when startup Lily Camera came out of stealth with its $500 to $1000 camera drone and argued that it wasn’t so much a drone as a simple-to-use flying camera. This March, drone-maker DJI introduced the Phantom 4, with autonomous flying and tracking features that essentially make it that company’s first flying camera at $1400.

Flying cameras are drones designed for use by consumers that don’t want to learn how to fly a drone; they just want to take pictures. The cameras have tracking capabilities so they can keep a subject in sight, and can autonomously hover or circle, as well as take off and land on command without the user having to control the ascent or descent precisely.

People are betting big on these companies. Lily, with founders out of UC Berkeley, has $15 million in funding and $34 million in preorders. Zero Zero, with founders out of Stanford, has $25 million in funding.

One—or perhaps more—of these gadgets will catch on. In a few weeks, I’ll be attending my son’s high school graduation in Silicon Valley, with, I’m sure, my view obscured by parents using pads and phones and selfie-sticks to record the moment. By next spring, I’m betting at least a few of the selfie-sticks and tripods are going to be replaced by camera drones. I’m not sure if that’s going to be more or less annoying.

Here’s the scoop on how the Hover Camera works, as described by Zero Zero cofounder and CEO Meng Qiu Wang to me yesterday:

“It has two cameras. The front viewing camera is a 13-megapixel camera that records video, but also has Simultaneous Localization and Mapping (SLAM), an algorithm that allows it to determine where it is. It also has a down-facing video camera, running an algorithm called optical flow, that looks at ground at 60 frames per second, so the Hover knows when it moves and can correct itself. These visual sensors are giving inputs of actual position and speed, meanwhile, the accelerometer and gyroscope gives relative position. All these signals are fed into the flight control algorithm, so when I throw it up in air, it can just hover there.

“When I want it to follow me around, it is using facial and body recognition to follow me and make sure I’m in the frame. It can follow anybody I choose. In the final version, though not just yet, it will do a 360 scan around itself and pull out all the faces, they pop up on my phone, then I can choose which person to follow automatically. Or I can control it manually with swipes and other gestures.

“This approach differs from the Lily Camera and the Phantom 4. Lily does most of its tracking with GPS, so you have to wear a device on your wrist.

“The Phantom 4 is running a lot of visual computation, but it relies on motion tracking, that lets it follow a car, say. We are running body and face recognition.”

Meng Qui Wang, and cofounder Tony Zhang have feet in both Silicon Valley and China; they started their undergraduate engineering education in China, completed their PhD degrees at Stanford. (Wang spent two years as a software engineer at Twitter in between.) They started Zero Zero in March 2014, locking up $1.7 million in angel funding just days after Wang submitted his Ph.D. thesis, he recalls. The company is headquartered in Beijing with 80 employees there, and has a small office in San Francisco.

“The scope of the engineering problem we are trying to solve is way beyond just a few folks. I knew it would take a long time, and a significant team, and when I did the math, running it in Silicon Valley would be risky. In China, our burn rate for our two years in stealth mode has been ridiculously low, at least a quarter of what it would have been like in Silicon Valley. And the quality of engineering in China is pretty good, particularly in computer vision, which takes a lot of math. The problem with engineers in China comes when they are doing original research, from scratch; because of the education system there’s not so much creativity. But my cofounder and I are both Stanford engineering grads, so we have great ideas, and we know what to do, we just need engineers to build out our vision.

“Another advantage of basing a hardware startup in China is to be close to suppliers and manufacturers. A lot of the R&D involves just getting to know every vendor out there, every chip solution out there, and figure out what is best to use. If you are in Silicon Valley, you are handicapped in terms of your access to that information, compared to being in China.”

Zero Zero has 1000 preproduction models built; it’s using some for its own testing, but plans to give 200 out to beta testers, that it will select from applicants who commit to a purchase of a production unit down the line. It expects to ramp up production and start taking preorders in the summer, with the drones widely distributed by the 2016 holiday season.

Does it really take 80 engineers to build a flying camera? Not exactly. Wang has a number of those engineers working on a more ambitious goal: a line of personal robotics products, of which the Hover Camera is a small but essential step.

“I don’t think the future of personal robotics is one-device system. I think it will take multiple devices that function together to be useful. A big problem in the home environment for a robot is object avoidance—we have a lot of objects in our homes and offices. But most of them sit below 1.5 meters from the ground. A robot on the ground has to do a lot of object avoidance, but once it is airborne above 1.5 meters, there isn’t much except the people, so a flying robot solves the object avoidance problem. But a flying robot has to be small and light, which limits its ability to move, say, even a mug of coffee.

“What we think could work is have a flying robot be the eye in the sky for the ground robot; it could do the path finding, then send control signals to the ground robot that say come this way and move this thing. We can decouple the sensors from the actuator, and make a personal robotics system, with one robot in the air be the sensor and the other on the ground do the heavy lifting.”

I had one last question for Wang before we wrapped up. Why Zero Zero?

“We have a Chinese name that is two circles,” he says, “But that was rejected when they tried to register it in China as a company name. So we came up with Zero Zero. It resonated well with the team; we’re engineers, after all, so we like zeros and ones.

The Conversation (0)

How the U.S. Army Is Turning Robots Into Team Players

Engineers battle the limits of deep learning for battlefield bots

11 min read
Robot with threads near a fallen branch

RoMan, the Army Research Laboratory's robotic manipulator, considers the best way to grasp and move a tree branch at the Adelphi Laboratory Center, in Maryland.

Evan Ackerman
LightGreen

“I should probably not be standing this close," I think to myself, as the robot slowly approaches a large tree branch on the floor in front of me. It's not the size of the branch that makes me nervous—it's that the robot is operating autonomously, and that while I know what it's supposed to do, I'm not entirely sure what it will do. If everything works the way the roboticists at the U.S. Army Research Laboratory (ARL) in Adelphi, Md., expect, the robot will identify the branch, grasp it, and drag it out of the way. These folks know what they're doing, but I've spent enough time around robots that I take a small step backwards anyway.

This article is part of our special report on AI, “The Great AI Reckoning.”

The robot, named RoMan, for Robotic Manipulator, is about the size of a large lawn mower, with a tracked base that helps it handle most kinds of terrain. At the front, it has a squat torso equipped with cameras and depth sensors, as well as a pair of arms that were harvested from a prototype disaster-response robot originally developed at NASA's Jet Propulsion Laboratory for a DARPA robotics competition. RoMan's job today is roadway clearing, a multistep task that ARL wants the robot to complete as autonomously as possible. Instead of instructing the robot to grasp specific objects in specific ways and move them to specific places, the operators tell RoMan to "go clear a path." It's then up to the robot to make all the decisions necessary to achieve that objective.

Keep Reading ↓ Show less