When it comes to the rules governing airspace, it’s a bit of a Wild West situation right now–the introduction of small unmanned aircraft systems (UAS) has left us in uncharted territory.

At this year’s conference for the Association for the Advancement of Artificial Intelligence (AAAI-16) in Phoenix, Arizona, a panel of experts gathered to weigh safety considerations that are critical when it comes to autonomous flight–manned and unmanned. The group included Ella Atkins (University of Michigan), Claire Tomlin (University of California, Berkeley), Amy Pritchett (Georgia Tech), and Mykel Kochenderfer (Stanford University).

When it comes to the adoption of UAS, not only must autonomous flying meet requirements– they must also be considered safe by the public. “Aviation can’t allow for single-point failures,” said Pritchett. “They need layers of defense.”

The history of flying, Atkins reminded the group, gives some interesting insight into autonomous flying. The first flight, in fact, was only possible because of unmanned aircraft. “Drones have been around in name of ‘model aircraft’ since first flight existed,” said Atkins. We take it for granted that flights today are considered safe, but we did not always feel this way.

And for some context in terms of today’s drones, there are more small unmanned aircrafts registered through the FAA than the total number of other airplanes. Yes, that’s right: more registered drones than every other airborne vehicle combined.

SEE: How a drone on a leash will transform autonomous flying

Here are the big takeaways from the experts when it comes to the current state of autonomous flying:

  • Passenger transport may come last: While the technology may be there, passengers still trust pilots more than autonomous systems (this could have important implications for autonomous driving as well).
  • Weighing costs: Even though the tech may be ready, there are times in commercial flying, for example, where airports do not use autonomy because of the cost of maintaining and verifying equipment. Most airports don’t have autoland capability, for example, because it’s cost-prohibitive–even though it’s proven more effective than a human pilot
  • Tools for safety: There are some ways to enforce the safety of unmanned flying, such as: platoons, in which UAS merge together into a line, requiring fewer safety checks; geofencing, in which a parameter is set and enforced through sensors; emergency flight management; and detect-and-avoid, or tools that help UAS sense potential collisions.
  • Human crews are still relied on: Pritchett pointed out that in 20% of flights, pilots are still required to intervene in some way. “The reality is that a lot of systems rely on crews to execute commands,” she said. “Humans capture more failures than they cause.”
  • Unknown scenarios pose a challenge: We need to understand, and be able to deal with, the anomalies. “Engineers and computer scientists haven’t anticipated even modeling the features that might be required to respond to unknown scenarios,” said Atkins. Tomlin agreed. “We don’t know all the things that could go wrong. It’s important to understand how to design and introduce the system in the safest way possible, knowing that in these systems, you can’t plan for everything.”
  • 0-500 foot airspace for UAS is a problem: “We run a risk at drawing the line from 0-500 feet,” said Atkins. “We’re taking away the margin of recovery by putting it down so low, and also potentially violating property rights. You run the risk of seeing a whole platoon of drones in front of your face.” While regulators are considering sovereign airspace of the US to go “down to the blade of grass,” that doesn’t make sense. It’s an issue that the Supreme Court will need to decide–what are the rights of the landowner versus the federal government. The nightmare scenario, she said, is that “you don’t have the enjoyment of your backyard because you happen to live in between an Amazon warehouse and a major urban area.”
  • Regulators are struggling to come up with the rules: “People haven’t figured out what drones mean yet,” said Pritchett. “There’s no single metaphor. Some people want to regulate them in the same way you’d regulate throwing a football with spikes on it. Is a local matter? A FAA matter? We don’t even know about the jurisdiction.”

SEE: UPDATED: FAA’s drone regulations: Answers to common questions

So, while the technology has definitely come a long way, it may be (as is often the case) the people that will need to catch up.

“We often have visionaries who come up with technology, people who try to keep up, and people who follow along,” said Pritchett. “Right now, we’re hearing from the innovators. To think that the rest of the world will say ‘yes, it’s economically feasible, it’s desirable, it’s meets all these other metrics,’ is less certain.”