Will Driverless Cars Need Remote Human Supervisors?

By Graham Jarvis

---

20th February 2023

Over the last few years autonomous vehicle start-ups have raised tens of billions of dollars based on their cars being truly self-driving.

Yet, international news agency Reuters reported in September 2022 that they may not in the end be totally autonomous. Industry experts and executives believe there may be a need for remote human supervisors to help the ‘robot drivers’ whenever they are in trouble.

This begs the question about why a need for autonomous vehicles is there if a human being is able to remotely drive and take control of a vehicle, when a human driver could be sitting the driving seat in full control of the car, van or even a lorry. With increasing interconnectedness, and with a third-party becoming potentially responsible for the vehicle, it also raises privacy and personal autonomy concerns that a journey could be interrupted or prevented remotely for nefarious reasons. Having a third-party in remote control of a vehicle also raises insurance liability, risk and premium questions.

In its Executive Summary, the Association of British Insurers’ response to the Law Commission’s Remote Driving Issues Paper says that it supports the development of automated driving technologies. However, it concurs that there may be a need, in certain circumstances, “to have these vehicles remotely driven for part of their journeys and we know of several companies looking to trial this technology in the UK”.

The ABI would, nevertheless, like to see more robust regulation and clearer guidance on the use of “remote driving technologies for automated vehicles for use in specific and limited operational design domains”. It believes it’s an imperative, let alone a prerequisite, for the regulations to be developed. It also wants the “responsibilities of the Entity for Remote Driving Operation to be more specifically defined to prevent the misuse of these technologies”. Among other factors, the ABI says insurers have serious concerns about the use of remote driving technologies to operate manually driven vehicles including issues related to connectivity, situational awareness and vehicle capabilities.

Raising questions

Bruno Taratufolo, marketing and product strategy, AGC Glass Europe agrees with the ABI, stating that the idea of a having human supervisor with the ability to take control of a vehicle remotely raises a few questions. This includes the potential for the abuse of power.

He explains: “If the remote supervisor overrides the autonomous vehicle, who’s responsible for any negative outcomes or damages that may result from such a decision? The second observation is a privacy concern. Inside the autonomous vehicle, there may be individuals who may not be aware of the actions or controls being taken, so this is potentially the Big Brother issue.”

He wants to see regulations put in place to ensure the ethical and responsible use of human supervisors. This is to ensure the existence of transparency, privacy, and accountability protection for the individuals using the vehicles. He adds: “There is a need for a strong framework to ensure these three critical factors. I have concerns about the privacy side but human supervision could help with safety in terms of the large-scale use of autonomous vehicles.”

Improving economics

Dr Benedikt Kloss, associate partner, McKinsey Center for Future Mobility (MCFM) stresses that he is enthusiastic about autonomous vehicles. He says at Level 4 of autonomy vehicles are able to operate at a level of 99% of all use-cases. Most of the vehicle will operate within a constrained environment. This may be a highway, or a whole city. He adds that the last 0.1-1% may require increased level of investment, making it potentially economic to have a remote human driver supervising the vehicles. The remote driver could, he suggests, supervise 1-5 vehicles at the beginning and then advance to handling 20-30 vehicles in the future, like someone in Air Traffic Control would do to ensure the safe flight of passenger jets across a particular airspace. “The question is then about what the human being in the background does: this could range from the remote human driver steering the vehicle or telling the vehicles whether it is safe to go on,” he says. This may occur when the vehicle doesn’t recognize a traffic sign or a situation, and so it would depend on the remote human driver for instructions.

Varun Krishna Murthy, industry analyst for connected and autonomous driving, Frost & Sullivan adds a potential use-case could be about ensuring the operational efficiency of fleet vehicles. He explains: “If an autonomous vehicle is stuck behind a broken-down truck, it is required to follow traffic, and it can only do what it has been taught. So, it will require a remote operator to know what to do. The teleoperator will instruct the vehicle to do make certain maneuvers and continue the journey – circumstantial actions based on road, environment and traffic conditions. Teleoperations will find an application in robo-taxis and shuttles to start with and may be used in automated parking.”

Humans handling ethics

Taratufolo believes the ethical considerations of remote human supervision can only be handled by human beings, and not by artificial intelligence and machine learning. He nevertheless concurs that remote human supervision may be necessary to enable the vehicle to cope with unforeseen situations.

Although autonomous vehicles are designed to work in a range of scenarios, there may be circumstances where human supervision is required – particularly in a smart city environment where there will also be unpredictable human traffic in the form of pedestrians and cyclists. “It would be very difficult to predict unforeseen situations that are linked to human behavior”, he says.

He adds that there is a technical limitation to consider too. He asks: “Although AVs are designed to be highly reliable, what if there is a technical failure, who has the final decision? The human supervisor could recognize them and take act to ensure vehicle safety, and of everything, and everyone around it.” Another aspect is the need to allow people, particularly in democracies where freedom of movement is an essential right, to travel where they want, when they want.

Protecting autonomy

That’s not quite the case in China. The country has developed a social credit system, which allows or prevents people from doing certain activities if the authorities deem their behavior to be acceptable or not. There are therefore concerns that the remote control of connected and autonomous vehicles could lead to similar actions being taken, perhaps, in the name of climate change. So, what is being done to counter any prospect of this kind to preserve privacy and personal autonomy?

Jack Palmer, senior consultant, european mobility and transportation team, Frost & Sullivan describes China as being a law unto itself. He explains that fingers have been burnt for not doing what the consumer wants, and if the industry is to encourage people to embrace, use and adopt autonomous vehicles, there is a need to ensure they retain their privacy and personal autonomy.

Palmer cites the consumer electronics industry’s experience, with regard to burnt fingers. However, he claims that Apple is the shining light for privacy. He adds: “Automakers have applied the Apple mantra and so automakers now feel that privacy is at the heart of it and so I would be very surprised that any of them would act nefariously to stop someone from travelling.”

Despite the rise in data aggregation platforms, where consumer and vehicle data is shared with third parties, he admits that consumers often aren’t aware of how their personal data is being used. He believes this will change over the next few years as consumers are aware that their smartphones are being used to generate data. In contrast, many people don’t still realize that their vehicles are also becoming a rich data aggregation source.

Palmer resolution is thus: “It’s pretty straightforward. Consumers have to read the T&Cs. It’s on the regulators to guide companies to use the data in the right way. You have to be consumer-centric and think about the service you are providing, rather than the money generated from it. There has got to be a motivation to do good rather than being about revenue.”

Several initiatives

Taratufolo also points out that there are several initiatives in play to protect privacy and personal autonomy. The first is about privacy regulations, about which explains: “Governments are working to establish privacy regulations that specifically address the use of CAVs, aiming to provide individuals with the right to control their personal data to ensure that it use in an ethical and responsible manner.”

He says the companies involved with developing connected and autonomous vehicles are also taking action to enable transparency and accountability in the use of data.  This includes regular reporting on data use, the emboldening and updating of security measures, and responding to privacy concerns. Much of this, he suggests, is already being done by Tesla and Volkswagen. Then there is, as highlighted by Palmer, the need for public engagement and education. People need to know more about the use of CAVs, and about the issues they could raise. Taratufolo reiterates that it’s vital to regulate the use of the AVs to ensure they are used in a responsible and ethical manner to protect the privacy and autonomy of the individuals using them.

Insurers: prepare now!

Kloss concludes that now is the time for insurers to start to prepare for an autonomous driving reality.  As such, The ABI isn’t resting on its laurels. It argues that if the remote driving of automated vehicles is legalized, and if it is established that the Entity for Remote Driving Operation (ERDO) is liable in the case of an incident with a remotely driven vehicle, “it underpins the necessity for vehicle data to be shared with all relevant parties including first responders and insurers.”

The ABI says it will be crucial to have data to show whether a vehicle was in self-driving mode, being remotely driven, or even being driven by a human driver in the driving seat of the vehicle. In the UK, the conditions of the Automated and Electric Vehicles Act 2018 may in certain circumstances apply, and it says the data set should be similarly defined to what is set out within the UNECE Regulation 157 governing the use of Automated Lane Keeping Systems (ALKS) with adaptations to remote driving.