Autonomous drones, navigated by AI, are taking flight. Drones are delivering packages, taking selfies, following forest paths, and navigating obstacle courses, offering just a glimpse of the great commercial potential for autonomous drones. The Federal Aviation Administration (FAA) recognizes this potential, but maintains restrictions on their use due to safety concerns. Although FAA waivers are helping commercial autonomous drones take flight, current rules requiring that each drone flight have a dedicated human pilot and stay within visual range pose obstacles to widespread commercial use. Recent developments, including the launch of the FAA's Unmanned Aircraft System Integration Pilot Program, may allow AI-powered drones to soar to new heights.
Sky's the Limit: The Potential of Autonomous Aerial Drones
NASA's Jet Propulsion Laboratory recently held a drone race, in which AI-powered autonomous drones raced through an obstacle course, competing against drones piloted by a professional human pilot. While the human-piloted drone was faster, the autonomous drones were more consistent in their times, and were able to navigate smoothly through the twisting course, more or less flying the same route every lap.
The drone race illustrates, among other things, that autonomous drones are well-suited for tasks where consistency is valued over speed, or where the ability to excel at repetitive tasks without boredom or exhaustion is needed. Potential use cases for autonomous drones include deployment in remote or rugged areas that are difficult for humans to reach without the need for a human pilot in the vicinity. Autonomous drones could conduct pipeline inspections in remote areas, or survey power or telecommunications lines that stretch across mountains and through wilderness areas. In agricultural areas, they could autonomously survey crops for pests or conduct imaging to measure the health and growth of vineyards. In urban areas, autonomous drones could soon navigate through our cities delivering packages, and may be trained to respond appropriately to the humans around them. Businesses are already using human-piloted drones for some of these tasks, but autonomous drones could perform these same tasks more efficiently, for longer periods, and in places less accessible to human operators.
Given advances in both AI and drone technologies, the potential benefits of this technology are evident. But current rules limit full-scale deployment of autonomous drones.
Flight Restrictions: The Current FAA Rules on Small Unmanned Aircraft Systems
Although the FAA recognizes the potential of autonomous drones to usher in a brighter future, its rules still constrain autonomous drones from achieving their full commercial potential.
Current FAA rules on drones govern the operation of unmanned aircraft weighing less than 55 pounds on takeoff (with a narrow exception allowing hobbyists to fly model airplanes). The Rule, which has been in place since 2016, was influenced by comments from the aviation industry, technology companies, agricultural concerns, academic institutions and hobbyists, among other stakeholders. These rules provide substantially greater flexibility than the previous rules – in which commercial drone use was forbidden unless the operator held a traditional pilot's license for manned aircraft and an FAA exemption was obtained. But even under the more liberal rules, each flight requires a designated FAA-certified remote pilot. And that pilot must maintain visual contact with the aircraft (either directly or by using a "visual observer") throughout the entire flight.
In addition to the FAA rules, certain state and local governments have enacted restrictions on drones. While some of these, like New Jersey's "drunk droning" law, may be inapplicable to autonomous drones, others – like a Texas statute barring drone flights over "critical facilities"– could constrain deployment of autonomous commercial drones. In the first case on the matter, Singer v. City of Newton, a federal court held that municipal restrictions on drone flights were preempted by the FAA rules. That decision is on appeal and it will likely take time to obtain clarity from the courts on the space – if any – for state and local regulation in this area.
Taking Off Now: Autonomous Flight Permissible Under Certain Circumstances
The current rules do not foreclose the possibility of autonomous flight. In enacting the rules, the FAA stated that "this rule will allow the autonomous flight of small unmanned aircraft." The FAA further recognized that "[a]utonomous operations have numerous practical applications, including agricultural operations, aerial photography, and search and rescue," and added that it "agrees … that the ability for a small unmanned aircraft to fly autonomously could add significant utility to a small UAS [unmanned aircraft system] operation and would further encourage innovation in the industry."
But the FAA went on to clarify that autonomous operation is only permitted when and where there is an FAA-certified human pilot standing by, visually monitoring the particular drone and able to take full control. In justifying this restriction, the FAA cited concerns about sense-and-avoid technologies:
[T]here is insufficient data to establish that UAS equipage is able to, at this time, detect other nearby aircraft in a manner that is sufficient to provide a substitute for the human pilot's ability to see and avoid those aircraft. Thus, a small unmanned aircraft may be unable to, without human input, yield the right of way to another user of the NAS [National Airspace System] that may enter the area of operation. Accordingly, this rule will require that the remote pilot in command have the ability to direct the small unmanned aircraft to ensure compliance with the provisions of part 107.
Thus, although the FAA rules permit some autonomous operations, they are sharply limited by the requirement to have humans within visual range, including a remote pilot that is monitoring the aircraft and ready to take control instantly. That requirement effectively precludes fully autonomous operation.
Pushing the Envelope: Waiver Procedure, Integration Pilot Program, & Future Rule Changes
Where the existing rules impede planned commercial unmanned operations, one possible solution is to seek a waiver of the visual line-of-sight rule. Over 1,600 waivers of various provisions have been granted so far, many for law enforcement, but some also for news-gathering, firefighting, photography, film, and package delivery. The majority of the waivers granted simply allow for nighttime flights, but several waivers have been issued concerning the visual line-of-sight rules.
Not all provisions of the FAA's drone rules are waivable. In particular, the requirement for an FAA-certified "remote pilot in command," is not waivable under the streamlined process. Therefore, until the FAA revisits its rules, fully autonomous flight (that is, without a human remote pilot in command for each drone) will require special authorization from the FAA outside of the regular waiver process.
The White House has encouraged the FAA to work with governments and industry in advancing the development and deployment of aerial drones. An October 25, 2017 Executive Memorandum declared that it is U.S. policy to "promote the safe operation of unmanned aircraft systems (UAS) and enable the development of UAS technologies for use in agriculture, commerce, emergency management, human transportation, and other sectors." The Memorandum does not specifically address autonomous drones or AI, but directed the Department of Transportation and FAA to establish a "UAS Integration Pilot Program … to test the further integration of UAS into the NAS in a select number of State, local, and tribal jurisdictions." The Memorandum invited proposals from state, local, and Tribal governments for the use of drones, and "as necessary, use existing authorities to grant exceptions, exemptions, authorizations, and waivers from FAA regulations," including the FAA's waiver process.
On May 9, 2018, the Department of Transportation and the FAA announced the first awardees under the UAS Integration Pilot Program, selecting ten drone programs from over 150 proposals. Several of those projects appear well-suited for autonomous drones. One project, involving Virginia Tech and several high-profile corporate partners, will involve both package delivery and power line inspections. Another, involving the City of Reno, Nevada, will use drones to deliver defibrillators to assist patients with cardiac emergencies. A project involving San Diego and various partners looks to explore the use of drones for food delivery, border surveillance, and delivering medical lab samples. Experience from these projects will both drive technological innovation and inform future rulemakings, allowing manufacturers to demonstrate advances in key technologies like sense-and-avoid, and potentially providing a basis for the FAA to modify its part 107 rules in order to allow autonomous drones to take off across the country.
Based on the FAA's prior recognition of the utility and benefits of autonomous commercial drones' potential, and the White House's push to accelerate the development of drones for commerce and other uses, the FAA might be open to making such a change in the future, particularly as sense-and-avoid technologies improve. Thus, a combination of advancements – in both technology and regulation – may allow autonomous drones to achieve their commercial promise.
Companies seeking to develop or deploy autonomous drone technologies should be aware of the current FAA limitations, and should consider whether the current waiver process may be helpful in achieving their goals, while also considering participation in future rulemakings in order to shape future FAA rules and policy. Proponents of such changes to the FAA rules should consider supporting voluntary standards or codes of conduct to ensure these autonomous systems will be safely deployed and consistent with the public interest. Doing so may be help gain support from other stakeholders and the FAA to enhance the operational autonomy of autonomous drones that can save lives, enhance food production, and keep our industrial systems safe.