Flying in the U.S., Europe and Australia is significantly safer than driving a car. Your current odds of being in a fatal accident are one in 11 million, whereas travelling by car lowers the odds to roughly one in 5000. Without taking away from the tragedy and loss of victims of crashes, the safety of flight must be preserved and improved even as we explore the new frontier of autonomous flight.

Today, some commentators have even argued that autonomous flight is likely to become a reality much earlier than autonomous driving. Unmanned eVTOLs (electric vertical take-off and landing aircrafts) are now being manufactured and are undergoing certification processes. A distinct but related issue is the extent to which artificial intelligence (AI) may be used in autonomous flight; for example, a passive sensor may eventually mimic human analysis onboard. Yet the legal regime around autonomous flight is still under development, as well as the distinct legal issues posed by AI. Having an accurate sense of how the legal regime will evolve to balance the need to innovate in aviation against protecting the public at large will help manufacturers, investors and operators better navigate this sector.

I. Route to certification and commercialisation of unmanned eVTOLS

A critical component for the safety of autonomous aircraft is the uninterrupted communications links from autonomous aircraft to (1) other aircraft, and (2) ground stations. Moreover, data received through sensors and other features on the aircraft itself, enable “detect and avoid” capabilities that are essential when a pilot is not on board. That is, an unmanned aircraft has to rely on these sensors to see and navigate around obstacles and other aircraft, including manned aviation.

For manned aircraft, detecting and avoiding other aircraft has multiple redundancies: equipment onboard the aircraft, air traffic control management, and the pilot’s visual observance. Loss of communication for one of these links does not result in catastrophic failure because communication can be reestablished through another channel. In the worst-case scenario, a pilot may be able to switch from instrument to visual flight rules (VFR) to safely ground the aircraft. VFR, of course, does not exist with autonomous eVTOLS operating without a pilot on board.

For unmanned aircraft, civil aviation authorities around the world will be concerned to ensure that autonomous flight technologies demonstrate the safety threshold for communications with ground stations and back-up capabilities that can achieve the same or higher levels of safety as piloted flights.

From our team’s experience in navigating this assessment and certification process over the years, we know that civil aviation authorities will take a detailed and technical approach to certifying autonomy in aviation, where the safety of each phase of autonomy is established before adopting additional automation.

II. Liability for accidents in a world of autonomous flight

Regulators and lawmakers around the world will also have to consider how to apportion risk and liability for accidents arising from autonomous flight

Not so long ago, flight itself was the new frontier. To attract investment capital to the nascent airline industry, the Warsaw Convention was introduced in 1929 to limit the liability of an airline in the event of personal injury or property damage. The impact of the Warsaw convention was two-fold:

  • First, the Warsaw convention creates limitations on the airline’s liability for:
    • Passenger death, bodily harm or delay
    • Baggage loss, damage or delay
    • Cargo loss, damage or delay

Unless the airline engaged in willful misconduct.

  • Second, in exchange for a limited claim, claimants would not have to prove that the airline was negligent, under the Warsaw convention, the claimant would have a claim irrespective of negligence, and instead the airline had to show freedom from negligence.

In 1929, the cap on liability for passenger death or injury was $8,300 per passenger. This cap was increased over the years, including by introduction of a concept of SDRs to address issues of inflation.

Most significantly, the Montreal Convention of 1999 imposes strict liability of up to 100,000 SDRs, and presumptive liability in an unlimited amount for injury beyond 100,000 SDRs. Under the presumptive liability regime, the airline’s liability can only be avoided if the airline can show that it was not guilty of negligence, and can be discounted by the claimant’s negligence or wrongful act. ‘

It is unclear whether Montreal Convention 1999 rules limiting liability will apply to passenger and cargo transport missions in autonomous flight. The Montreal Convention makes no distinction between autonomous and piloted flights, and its application is untested. By extrapolation from the attitude of policy makers at the inception of aircraft technology, when assessing a liability regime for autonomous flight, policy makers could include amongst their considerations, whether extending the scope of the Montreal Convention to cover autonomous flight would be supportive of more investment into developing the technology. Even without specific expansion of the Montreal Convention, policy makers around the globe may use it as a guide when considering and drafting liability laws and policies governing autonomous flight. We are keeping a close eye on how legislative bodies in the US, Europe, and beyond are debating the issues and whether an international framework is part of the discussion.

III. Financial responsibility and risk mitigants could be required by regulators

In order to envision the laws of the future, our best strategy is to look for scenarios where autonomous flight is already a reality. At present, commercial space operators are already operating unmanned space missions, with varying levels of autonomy versus remote piloting.

As part of the licensing process for unmanned space missions, the Federal Aviation Administration has required companies engaged in commercial space transportation to insure its operations. The FAA uses a process called “maximum probable loss” (MPL) determination to determine the amount of insurance needed to cover damages resulting from an accident. In the MPL process, separate values are determined for the amount of third party and government property damages. The process is further informed by separate analyses for pre-flight, flight, and re-entry activities. The MPL is fundamental because all license holders must provide a financial demonstration equal to the amount of the MPL. A licensee can meet this requirement by (1) purchasing insurance equal to the MPL amount (2) proving it has financial reserves equal to the MPL and setting aside these funds; or (3) putting the amount of MPL funds in escrow. Further details of the financial responsibilities of commercial space flight license holders are provided in 14 CFR Parts 440.

Having this knowledge of the analysis of regulators like the FAA in managing risk gives operators, investors and manufacturers insight into what could be required in order to obtain approvals for special test zones for additional autonomous flight operations, like eVTOL. It also provides a framework for eVTOL manufacturers and designers to assess the extent to which back-up technologies should be implemented to reduce the risk of accidents.