Articles that set out to consider the hidden costs of new technology often seek to catch the eye by starting with a vivid description of a recent near-miss. Nothing is as effective an antidote to entrepreneurial optimism as hard evidence of a disaster narrowly avoided.

Our chosen narrowly avoided disaster was an out of control 95kg drone (think small motorbike or large fridge) that ascended into airspace over Goodwood designated for circling passenger jets before falling into a field 40m from a house on 4 July 2019. The potential consequences of a 95kg mass being hit by a full passenger aircraft at 280mph or falling on to a fuel storage facility are obvious. Perhaps less obvious are the potential liabilities from even short periods of disruption of infrastructure such as the railways. As a result of the complex contractual arrangements between Network Rail and train operating companies what seems like relatively minor disruption (such as might result from a small drone getting entangled with overhead cables) can easily result in 6 or 7 figure liabilities – see e.g. Network Rail Infrastructure Ltd v Conarken Group Ltd [2011] EWCA Civ 644. For more detail on the liability issues arising from such events see “It’s raining drones, hallelujah; liability implications for personal injuries caused by drones” J.P.I Law 2021, 1 12 – 16.

This article examines the rise of drones from a different angle and asks the question – what will constitute a fair presentation of the risk that drones present to those that insure them? In particular the risk present by BVLOS (Beyond Visual Line of Sight) drones.

The AAIB report on the Goodwood drone accident is compulsory reading for anyone in the drone insurance market. Firstly, because it demonstrates the magnitude of potential liabilities. Secondly – and more importantly for this article – because it is revealing about the process which permitted the wayward drone so nearly to cause a disaster.

The process via which aircraft are permitted to fly is important to the aviation insurance market. In practice the certification process for conventional aircraft, and the regulatory framework in which they operate, are so rigorous and have taken so long to mature that the very fact that an aircraft can operate legally in first world States answers the question as to whether its manufacturer or operator has fairly presented to insurers the risk of selling or using it.

The commercial BVLOS drone industry does not yet have that pedigree. The risks it presents stem in part from the relatively recent (i.e. less than 20 years in aviation terms) development of Performance Based criteria for regulatory approval, particularly in relation to navigation equipment. This marks a change from the regulators stipulating what equipment an aircraft must have and how it should be constructed, to stipulating how the aircraft and its systems must perform – with greater discretion afforded to the manufacture as to build systems that do what is required and, crucially, the manufacturer providing the evidence that those systems in fact do what is required.

The UK CAA’s approach to BVLOS drones can be found in a series of pamphlets readily available on its website. It has created a “sandbox” in which “innovators” (drone manufacturers/operators) can develop safe BVLOS drones for use within unsegregated airspace. The aim is that the use of such drones will be “business as usual”. It will be for the manufactures to gather and present evidence to the CAA that their products can operate safely in unsegregated airspace.

The CAA has developed a pathway along which drone manufacturers will travel in order to develop drones with DAA (Detect and Avoid) capability allowing their commercial use without the risk of them colliding with other airspace users, buildings, obstacles or terrain. Or, presumably, birds. This sounds sensible.

It would be true to say that oversight has probably improved since Goodwood, but a careful reading of the CAA’s publications suggests that, in broad terms, the CAA sees its role in the “sandbox” as assisting innovators with regulatory challenges whilst the role of the innovators is to develop and prove the performance of the hardware. The question is whether this structure gives potential insurers a comparable level of comfort to that which they gain from the certification of commercial aircraft.

For over a decade the certification process in respect of commercial aircraft operating in the UK has been undertaken by EASA. As a result the CAA has had little requirement for, and lost much of, the expertise or resources required for, the tasks involved. It is not clear that the CAA was expecting to be given these responsibilities back as a result of Brexit. An issue for insurers is whether the CAA has sufficient expertise or resources to ensure that commercially operated drones are sufficiently safe and reliable that insurers have to look no further than CAA approval when assessing risk. The question is not whether the CAA does its job properly; by definition it will be doing so if it ensures that the applicable Regulatory requirements are met. The question is whether the meeting of the Regulatory requirements reflects the insurance market’s real appetite for risk.

Insurers may wish to start asking more searching questions than are usual in the Aviation sector about the risk they will be asked to underwrite. Public liability insurance for commercial drones will be compulsory. But there is no indication that it will follow the EU’s motor insurance model and oblige insurers to cover liabilities regardless of their contractual rights. There will be the option for an insurer to avoid a policy if the requirements of the Insurance Act 2015 in relation to fair presentation of risk are met. Apart from licensing or maintenance issues this is not really an area that aviation insurance has often had to consider. Getting a large commercial aircraft to market involves addressing just about every risk that can be thought of. Smaller General Aviation aircraft are now mostly 50+ year old designs that are tried and tested if flown correctly. By contrast an aviation underwriter presented with 1000 pages of computer code by a drone start-up business with an assurance that the computer and its associated hardware will fly a 25Kg drone safely over city streets is facing a new challenge. Is the licensing authority’s authorisation good enough? What sort of evidence has been produced by the manufacturer to achieve Regulatory authorisation? What grade of wiring has been used? What is the level of redundancy? Do the CAA know? Does anyone know? Are the manufacturers telling you and the regulator what it wants to hear? Are they big enough to care about the reputational damage of doing so?

Whilst the technology that allows conventional aircraft to fly safely in segregated and non-segregated airspace is well developed, its application to BVLOS drone operations is not. In the UK drone operators are at the stage of requesting temporary exclusive use of what would normally be unsegregated airspace in order to test their operations. This allows BVLOS drones to fly safely between two points, assured of not encountering other traffic or obstacles and primarily using GPS to navigate. It is not a particularly difficult challenge.

The step up to drone operations in unsegregated airspace and along routes that are not predetermined is significant. There is a public perception that GPS is the answer to all navigation problems – but it is not. The development of GPS primary navigation equipment in conventional aircraft has taken decades and even now radio navigation aids and laser ring gyro inertial navigation systems are required in addition. GPS signals can drop out. Built up areas can mask the satellite signals. The signals are occasionally jammed by the military. A consequence of Brexit is that the UK will no longer be able to make use of the European EGNOS facility providing precise guidance when using the Galileo GNSS.

Few conventional navigation systems can detect temporary obstacles. That is not a problem for aircraft flying at altitude except near airports where such obstacles are regulated and information about the temporary presence of tall objects such as construction cranes disseminated on a daily basis. It is a potential problem for drones flying at a low level. Given the increasing prevalence of the use of drones for various purposes on construction sites, often in crowded urban environments, the risks arising from this will be obvious.

The Detect and Avoid “Ecosystem”  proposed by the CAA will require drones to know what other drones are doing. Many aircraft achieve this by broadcasting information about their height, velocity and location using using a complex radio-based system – ADSB. However, many aircraft, particularly those flying at low level, do not use ADSB. Nor do birds.

There is no reason to suppose that these challenges cannot eventually be overcome. A great deal of investment, innovation and thought is being deployed to do so. But it is clear is that the safe use of BVLOS drones in unsegregated airspace will require navigation, detection and position broadcasting systems all of which need powering by heavy batteries before the rotors have turned. In aviation weight is everything – because it impacts on payload, which determines profitability.

Many of the drone innovators in the CAA sandbox have justified their experimental operations on the basis that their drones are carrying critical supplies for the NHS. That is certainly one use for them but there is an element of PR here. Potential drone operators are not primarily designing their products to help the NHS – they are designing them to offer a profitable way of transporting goods to businesses and individuals. That model relies on convenience i.e. point to point delivery which necessitates drone flight over and in populated areas.

There is nothing wrong with that and drone use has clear potential to be an ecologically sound and efficient part of the logistics chain. But PR should have no place in the CAA’s sandbox. It would be unfortunate if the nascent BVLOS drone industry faltered in its early years as a result of public confidence disappearing when liability insurers met the first few inevitable claims with policy avoidance on the grounds that the process by which regulatory approval was obtained hid unacceptable and declarable risks taken by the manufacturers or operators. There is, perhaps, much to be said for drone insurers having their feet in the CAA’s sandbox with the innovators now, so that they really know what risks they are being asked to underwrite when commercial activity begins.

Insurers wishing to enter the BVLOS market should also consider taking advice on how to structure the underwriting process so that proposers are asked about significantly more than whether Regulatory approval has been obtained.


Authored by Patrick Vincent QC and Max Archer