April 15, 2015      

A new system for autonomous drone stabilization and landing could help usher in a new era of enhanced safety for commercial drone operators, according to the University of Zurich team behind the technology.

It enables drones to autonomously identify safe landing spots in case of systems. As an added bonus, the system also allows drones to be launched simply by throwing them in the air.

Existing drone-stabilization systems use either GPS or motion capture to achieve similar performance, said Davide Scaramuzza, an assistant professor of robotics at the University of Zurich (UZ) and an expert on computer vision applied to autonomous navigation of visually guided micro-flying robots.

“GPS doesn’t work indoors, and outdoors it’s not reliable when flying close to buildings, whereas [motion capture] requires the drone to fly within the space covered by the preinstalled and precalibrated motion-capture cameras,” Scaramuzza told Robotics Business Review. “However, in both these methods, the drones are unable to perceive their environment.”

“Our system allows drones to stabilize both indoors and outdoors regardless of any external infrastructure, such as GPS or motion-capture systems. Thus, it makes the drones truly autonomous,” he said. “Furthermore, through the use of an onboard camera, the drone is capable of perceiving its environment. This allows the drone to actively search for a safe landing spot without relying on an already existing and therefore outdated or imprecise map.”

Equipped with a single camera and acceleration sensors, the autonomous drone system allows drones to emulate the human visual system and sense of balance. When a toss or a failure situation is detected, machine vision software analyzes the scene to identify distinctive landmarks in the environment and restore balance.

An onboard smartphone processor handles drone control and image processing, which allows the drone to operate autonomously.

This allows drones to operate safely beyond the operator’s line of sight — a capability that could have huge commercial implications.

“The biggest hurdle to overcome when trying to use autonomous drones in commercial applications is to make them safe,” said Scaramuzza. “Only when the safety of drones can be guaranteed, they will be allowed to be used for commercial use. Our technology provides a first step towards safe drones and, hence, their commercial usage in the future.”

Scaramuzza’s group has an ongoing collaboration with Parrot-SenseFly, a Switzerland-based company that manufactures fixed-wing drones for mapping. Some of the new vision-based autonomous navigation algorithms developed by the UZ team will be used in a new quadrotor drone called the “eXom,” which is designed for remote-inspection tasks.

The eXom is expected to be released before the end of 2015.

Alhough the technology works, it may take one to two years for the new system to be widely incorporated into commercial drones, according to Scaramuzza.

“The technology … is already around. What we are missing is good legislation to ensure that people are safe when drones are operating,” said Scaramuzza, who said he expects to see drones using predefined flight corridors, to enhance safety.

Limited drone delivery programs have already been approved in Germany without such safety technology. Those programs are not premature, said Scaramuzza, but questions remain about their capabilities in case of loss of GPS.

“What happens if the drone loses the GPS signal? Do they have an alternative way to navigate?” he asked. “What technology do they plan to use? Do they have autonomous landing site detection and lading? If the drone plans to take off or land between buildings, where there is no GPS, how do they plan to do this?”

After some hesitation, the FAA granted Amazon limited rights to test its Prime Air service in U.S. airspace, although some industry observers doubt that drone delivery will catch on in the U.S. The FAA has already granted some exemptions for industrial drones.

Parrot takes flight with drone industry

Parrot will be the first company to use the technology, but it doesn’t have an exclusive license, and Scaramuzza’s team plans to license the technology to other companies as well.

A 2015 research report from WinterGreen Research predicts that the global unmanned aerial systems market, valued at around $609 million in 2014, will rise to $4.8 billion by 2021. Chinese drone maker SZ DJI Technology Co. is in talks with various investors to raise its valuation to $10 billion.

And Parrot is becoming an increasingly important player in the drone space.

Parrot’s fourth-quarter 2014 earnings report (download PDF) saw consolidated earnings rise 33% to €80.4 million ($85 million), with massive growth in the drone segment. In Q4 2013, drones accounted for 23% of Parrot Group revenues and was valued at €13.8 million ($15.5 million). By Q4 2014, drones accounted for 46% of group revenues with a value of €37.1 million ($39.2 million).

In 2014, 91% of Parrot’s drone revenues came from retail drones, which are aimed at hobbyists. Professional drones accounted the remaining 9%.

Parrot invested $2 million in MicaSense in 2014. Seattle-based MicaSense develops data gathering and analysis technologies that allow people to convert raw, multispectral image data into digital maps.

Also in 2014, the company invested $2.2 million in Airinov, a French provider of drone software and sensors for the agricultural market.

Parrot invested a further $1.05 million (33% of company capital) in EOS Innovation, which specializes in mobile robotics for the surveillance of industrial sites, logistics platforms, and warehouses.

Scaramuzza’s work is partly funded by NCCR Robotics — a 12-year joint effort among four universities, 19 professors, and more than 100 researchers funded by the Swiss National Science Foundation.

Founded in 1994, Parrot also operates in the consumer smartphone, tablet and automotive technology markets. The company has been listed on Euronext Paris since 2006 (FR0004038263 — PARRO) and is currently trading at €20.40 ($21.58).