Not that long ago, the industry was dominated by bold forecasts (and VC investment) projecting enormous adoption rates and a huge rise of the number of drones in operation.
In reality, the more ambitious projects have been slower to get going than some predicted. Amazon’s Prime delivery service is still more Youtube marketing than substance; Uber’s aerial taxi vision has barely got off the ground.
But that’s not to say they will never come to fruition. In fact, there is a range of technologies that could provide the foundations to expand and transform the use of drones in the future. Here are just a few of them…
Amid all the talk of BVLOS and other advanced operations, one of the ever-present challenges facing drone deliveries and urban aerial mobility is dealing with limited flight times. Some existing delivery solutions – like the one offered by Zipline – use fixed-wing drones to increase the range of their operations.
Others, like German UAM company Volocopter, plan to rely on several high capacity batteries to provide a rapid and more sustainable mode of transport.
However, both options still require recharging in between flights and/or spare batteries to be swapped in, reducing the amount of time they can spend actually doing what they are designed to do.
One of the futuristic solutions to the power and flight time problem is hydrogen. Powering drones with hydrogen comes with plenty of advantages. For starters, the element is the most abundant in the universe. It’s absolutely everywhere. It’s also emission-free and far more efficient than electric or fossil-fuel powered flight.
Hydrogen fuel cells rely heavily on air when creating a reaction to produce energy. So a hydrogen tank generates more energy than a regular LiPo battery of the same weight. In turn, this allows hydrogen fuel cells to offer superior flight times.
Part of that is down to weight. But it’s also down to the way hydrogen fuel cells produce energy: the reaction is on demand, rather than the stored energy of LiPo batteries.
There have been some promising examples of hydrogen being used to power drones in recent years.
In January this year, a UK research group flew a 20kg multi-rotor drone powered by a hydrogen fuel cell for over one hour. The project was a collaboration funded by Innovate UK that included engineering firm Productiv, UAS videography company BATCAM and fuel cell provider Intelligent Energy.
In February, South Korean drone company MetaVista used a fuel cell from Intelligent Energy (who also supplied the original project) to fly a far smaller drone for over ten hours.
There are plenty of barriers to wider hydrogen adoption, which we will explore in an upcoming post. But there’s no doubt that drone performance and capability could be greatly enhanced should the technology succeed and hit the mainstream.
Improving Computer Vision and Motion Planning
With all the talk of UTM, Urban Aerial Transport and flights BVLOS, a core component of these advanced, futuristic operations will surely be the improved ability for drones to recognise and react to their surroundings.
If we are going to trust them to fly with true autonomy, the AI systems to make that happen need to be devised. And it’s slowly happening.
In the past 18 months, several drone industry names have made significant progress in the fields of computer vision. Skydio’s R1 and the subsequent release of the startup’s developer platform look like an inflexion point for the industry’s relationship with AI.
The release of DJI’s latest drones – which once again showed progress in terms of sense and avoid capabilities – along with an interesting launch from ZeroZero Robotics – the Hover 2 – are clear signs that drones are getting smarter irrespective of their target market.
This year we’ll also see the joint contest from Lockheed Martin and DRL come to fruition, as university students battle it out to in the AlphaPilot Drone Innovation Challenge to build AI systems capable of navigating around drone racing courses at speed.
Last year, researchers at the Robotics and Perception Group at the University of Zurich’s Department of Informatics and ETH Zurich’s Department of Neuroinformatics built an AI system that pilots drones using a convolutional neural network and state-of-the-art path-planning.
They highlighted the system’s capability by using it to fly a tiny racing drone – usually the type of aircraft limited by how many sensors it can’t carry. But just imagine the technology applied to commercial applications on larger drones.
So you get the idea: there’s a lot going on. The boundaries of computer vision are being pushed from multiple angles. Drones that are capable of navigating complex, unfamiliar environments with 360º awareness and obstacle avoidance – all while tracking a moving target? Who needs pilots?
The Ability to Rest and Recharge
As living, breathing things, we all take for granted the ability to take five, sleep through until the morning and generally recharge our metaphorical batteries. Rest is a necessary part of life.
For drones, rest isn’t really an option. At least it’s not an option while they are on the job, doing the things we want them to do.
But there’s been some interesting R&D in this space. Essentially, a handful of researchers have been finding ways to make drones behave more like animals, and working out whether these changes in design can make them more energy efficient.
An international team of engineers this month published a paper exploring whether it is possible to design drones capable of ‘perching’ – maintaining altitude without expending energy or deviating from their data-gathering mission.
The team propose an adaptable landing gear that allows drones to attach themselves to a wide range of different structures, resting on street lights and the edges or corners of buildings to reduce power consumption, increase vision stability and preserve the scope of that vision.
It’s a fascinating concept and one that’s worth keeping an eye on.
There have been similar attempts in the past, most notably at Imperial College London. In 2017 a team of researchers from its Aerial Robotics Labs modified a DJI Matrice 100 – aka SpiderMAV – to shoot out synthetic material and attach itself to nearby walls. It was then able to reel in the thread to make it taut so it could shut off its motors to save power while hanging out for a bit.
There are two better known technologies that have gained ground while more sophisticated methods are developed: Docking stations and tethers. Both provide drone operations with a greater degree of persistence, whether it’s situational awareness support at the Super Bowl, an eye in the sky above an incident as it unfolds, or regular surveillance of a sensitive location.