The government has identified three main benefits of driverless technology: better use of road space, reduced emissions, an estimated £50bn annual boost to the UK’s GDP, enhanced mobility for those unable to drive, and improved road safety (human error is estimated to cause 85-90% of accidents).
The latter poses a particularly interesting challenge for regulators: driverless technology is unlikely to eradicate all accidents so how comfortable will the general public be with any accident occurring over which they will have little or no control? Given the complexity of the supply chain, where does liability lie? How do you programme a vehicle to make ethical decisions when interacting with other road users? And it is not just autonomous vehicles (AV) that are giving legislators a headache; the increasing popularity of micro-mobility vehicles (e-scooters) is also creating regulatory questions.
Driverless cars – who is liable?
There are, broadly speaking, five levels of vehicle automation ranging from self-parking cars to cars requiring minimal, or no, driver intervention and it is this latter category on which regulators are focused. Inevitably legislation will lag behind technological development: you cannot legislate for something that either doesn’t exist or is still in the experimental stages of development. However, automotive technology has been in development for over 30 years and a legislative framework is evolving in parallel with assessments of driver / manufacturer / programmer liability, associated insurance and cyber-security risks, and data privacy issues.
All forms of transport operate under different regulatory regimes to reflect their individual challenges. AVs, in their specification and control systems, operate more like automatic trams than ordinary road vehicles but, unlike trams whose movements are restricted by tramlines, AV will be operating in unpredictable environments, where the software has to navigate random movements of vehicles, people and animals. In the event of an accident, who is responsible - the software or hardware manufacturer - or the driver? Inevitably, once the regulatory regime is framed it will be refined as test cases come before the courts, a process which is already underway, as seen in an ongoing case in the US after an Uber self-driving car collided with, and killed, a woman pedestrian in 2018. Investigators found that the driver had been distracted while watching a TV show on his phone. Following the accident, Uber suspended its testing programme (and will not face criminal charges) and the driver has now been charged with negligent homicide and is awaiting trial.
Building public confidence
Building public confidence in AV depends will depend on a robust regulatory framework where liability is clearly defined. Being able to establish exactly what caused an accident is also critical and this is where AV can trump conventional road vehicles. Given the amount of data AVs require to operate, their systems can be programmed to store all the information required for accident investigators to analyse how the accident occurred (although system interrogation will mean investigators dealing with such large volumes of data, it remains to be seen if they could, in reality, process it within a reasonable timeframe).
The use of AI introduces another layer of complexity. By its very nature, AI use in AVs cannot be tested meaningfully as it will keep adapting the software to the situations in which it finds itself – it would be like trying to test a moving target. AV will need a different approach and probably one dependent on a different attitude to risk. Whether the public and politicians will wear this remains to be seen.
Regulation is evolving
In the US, the National Highway Traffic Safety Administration (NHTSA) has just started to address regulatory barriers by launching a consultation process designed to improve safety and update legislation for autonomous vehicles (such as passenger protection requirements in the absence of a steering wheel). Likewise, the Chinese are also addressing the specific legal and moral complexities presented by AV through a multi-agency approach. Currently, they restrict the use of AVs and it is the person in control of the car who is liable in the event of an accident. In Europe, data protection is the principal focus with the emphasis on security and privacy issues.
In the UK, current legislation does not cover advanced trials of AVs on public roads although there is a Code of Practice for organisations trialling AVs and the Department of Transport is planning to introduce a process, via the Centre for Connected and Autonomous Vehicles (CCAV), to support such trials. In the meantime, it is a legal requirement for the vehicle to be operated by a driver, either in situ or remotely, ready to take control as necessary (as well as obvious safety considerations such as roadworthiness and insurance).
In the meantime, legislators are grappling with a more immediate problem – how to deal with the rising popularity of e-scooters (currently not allowed on either pavements or roads) which form part of a government ‘Future of Transport Regulatory Review’. In order to test the regulatory water, there are a number of 12-month trials for rental e-scooters underway around the UK for which the existing regulations have been amended to allow them to be treated like electric bikes, and to use cycle lanes and roads. Rather bizarrely, e-scooters will continue to be classified as a motor vehicle; designed and built to minimum statutory requirements, they must have adequate motor insurance, and the users must have a valid driving licence.
This is in line with French regulations: e-scooters are allowed on the road, but not pavements (and riders travelling on pavements will be fined), they must not exceed a 25kph speed limit; riders must be 12 years old or above; and they must be equipped with lights, reflectors, a braking system, and a bell.
Technology will always outpace legislation
Inevitably, technology will always outpace the law and the more technically complex the product, the more legal issues will arise. Lawyers and insurers will continue to wrestle with a range of questions including product liability, data protection requirements (including the need to balance the security and integrity of the data, data privacy and the need for accident investigators to access it) and cybersecurity issues. E-scooters may prove to be a flash in the pan, ultimately trounced by the electric bicycle, but AVs are a certainty. Too many governments around the world have staked their green credentials on the success of AV technology to allow it to fail so legislators and regulators will continue their ceaseless game of catch-up.
This article first appeared in the Dec/Jan 2021 edition of Engineering Designer, the journal of the Institute of Engineering Design.