fully automated drone flights

Wait, how many drone flights are actually automated?

By most definitions, drone means “fully autonomous.” But a surprising amount of drone flights are not actually considered mostly automated — and even fewer are considered fully automated. Whether by design or not, most drone flights these days are highly human-dependent.

Just over 50% of drone programs said their drone flights were mostly automated, according to theDroneAnalyst 2021 Drone Market Sector Report, which relies on self-reported information from more than 1,0000 business and agency users in the drone space.

What’s the holdup in fully automated drone flights?

A big reason why the drone industry struggles to get to fully autonomous levels is for one single holdup: data insights and analytics. Just 14% of drone programs say that they’re able to automated the step of data insights and analytics.

“This makes sense when considering that both the data itself can be hard to interpret and that without repeatable, automated flights advanced AI/ML techniques cannot yield great results,” according to a blog post from Drone Analyst interpreting the results of the survey.

While data gathering and data processing seems to be a fairly easy step to automate, the next step of actually interpreting the data and making decisions based off it still largely seems to require a human mind.

And we’ve certainly seen a proliferation of drone software that can automate the process of gathering and, well, processing data.

automated drone flights agriculture mapping
Chart courtesy of Drone Analyst.

Mission planning, data syncing/upload and data processing seem to have middling degrees of automation. That said, the actual drone flights themselves have high degree of automation In agriculture and mapping, more than 60% of the actual drone flights are fully automated.

If you want a job as a drone pilot, you have some hope at getting employment. But if you’re a software engineer able to build a data analytics and insights platform, there’s likely far more hope for you.

Which industries are best at automation?

Some drone applications are better at (or perhaps just more well-suited to) automation than others. Mapping has seen a particular proliferation of automation, perhaps because it is relatively simple to program a drone to fly a defined mission plan — particularly if that mission involves covering large, flat areas like construction sites or farmland. Thus, minimal onboard intelligence like obstacle avoidance sensors are needed.

Unsurprisingly, the agriculture industry has adopted more fully-automated drones more quickly than other industries.

Among the lowest: real estate and insurance. Again, unsurprisingly, most real estate agents are using drones to generate listing photos or videos where a high degree of artistic creativity is required (and thus a human touch is more welcome). The public safety industry lags behind.

“The potential for automated drone ops at scale has never seemed closer,” according to Drone Analyst. “But we’ll continue to see a divergence across industries as certain missions benefit from relying more on intelligent controls, but not necessarily full automation.”


  • Jeremy Bennington says:

    While not all automated flights will be BVLOS, all BVLOS flights will have significant automation. An often overlooked aspect of automation is reliable navigation, especially when a VLOS pilot cannot detect an anomaly.

    We often take for granted the dependency of UAS on GPS/GNSS for navigation, surveillance, airspace management, and deconfliction. The ASTM PNT BVLOS working group has been working on tackling these topics and the first recommendation will be released soon. Moreover most exemptions and waivers for BVLOS has specific language about ensuring GPS performance, especially near structures. Solutions to these challenges are now coming to market to enable safe automation and BVLOS.

  • George Claymoore says:

    This doesn’t take a look at anything on the DoD side but my educated guess is that the majority of automation is being done there. Shield AI’s Nova is a great example. Robotic Research’s Pegasus is another one. There are several others as well. They don’t require any pre-defined planning beyond setting a few limitations to keep them where you want them. These are mostly recon tools that provide a lay of the land before boots move in.

    Another thing to consider is all of the Skydio users. Most people who buy the Skydio never even touch the manual controls. Those flights are mostly recreational. Which is another category that wasn’t looked at here.

Leave a Reply