【In-depth thinking】Talking about the ethics of autonomous driving

By Paul Golata from Mouser Electronics

New technologies are emerging one after another, driving human society forward. The driving force of technology is particularly evident in the field of transportation. Railways, airplanes and automobiles have always been the experimental fields of new technologies and innovations, pioneering many historical precedents. The inventions of predecessors such as James Watt, the Wright brothers, Gottlieb Daimler and Wilhelm Maybach have continuously pushed our imagination forward. Among the hundreds of millions of vehicles currently on the road, one of the most concerned issues involves improving occupant safety and overall vehicle operating efficiency. To help improve transportation efficiency, humans have integrated mechanical, electrical, optoelectronic and, more recently, artificial intelligence (AI) technologies into vehicles. However, in the future car, electronic and AI technologies take over too much or most of the human control of the vehicle. At this time, the transportation mode will be safer and more efficient because the vehicle is operated by an autonomous technology system (ATS), but what kind of ethical changes will this bring? Such AI requires a well-defined technology platform that allows it to operate relatively autonomously. Part of this system will include how to exercise ethical decisions to ensure human life and reduce and eliminate harm.

Currently fighter pilots are surrounded by a full complement of computers and artificial intelligence to assist them in flying at the highest level currently devised by humans. The AI ​​has the majority of control of the aircraft, with the pilot making the final decision and acting as a proxy to take over control as needed. In the future, continued advancements in electronics and AI technology foreshadow that more common items, such as cars, will have similar capabilities in terms of overall control agents for the car.

Currently, many electronic technology companies supporting the automotive industry are focusing on solving the technical needs of advanced driver assistance systems (ADAS) by developing both adaptive and predictive systems and components to ensure better and safer driving. ADAS assists automotive agents by providing warnings or taking actions to reduce risks, or by performing some partial control tasks of the vehicle to improve safety and performance. The current state of ADAS is to cooperate with the driver to a large extent: that is, the function of the human-machine interface in ADAS is part of the overall control of the vehicle, while the person still maintains the overall responsibility of the vehicle. It is expected that over time, technological advances will successfully hand over more and more agents and control to ADAS. It can be expected that ADAS may eventually develop into an autonomous driving system that will have advanced intelligence and react faster and achieve better operating results than human drivers.

Leading suppliers of all types of electronic components around the world are focusing on developing products and technologies that support ADAS and its extensions to promote the realization of future autonomous vehicles. ON Semiconductor provides powerful, AEC-certified, production part approval process (PPAP)-capable automotive products such as the NCV78763 power ballast and dual LED drivers for ADAS headlights.

Freescale Semiconductor ( now merged with NXP Semiconductors) has a professional vision of vehicles, MCUs, analog and sensor and digital networking products to help the world's most innovative ADAS solutions. Their new FXTH8715 tire pressure monitoring sensor (TPMS) solution integrates an 8-bit microcontroller (MCU), pressure sensor, XZ axis or Z axis acceleration sensor and RF transmitter to meet market-driven needs for improved safety. AVX, a global leader in passive electronic components and interconnect products, has developed the VCAS and variable gain amplifier series TransGuard® automotive varistors (MLVs), which are ideal for preventing automotive-related transients in ADAS applications. Delphi Connection Systems, a global leader in vehicle electrical and electronic distribution systems, now offers the high-performance APEX® series of wire connectors, whose rugged design and reliability make it an ideal choice for challenging automotive applications. As the importance and impact of ADAS continue to expand, electronic component manufacturers will continue to develop new products and bring them to market to ensure that vehicles can rely more on highly reliable and efficient technology systems to drive.

A car can be viewed as a tool whose primary function is to transport people from point A to point B safely and efficiently. However, there is no doubt that cars can also be operated by agents and used for sinister purposes such as killing weapons, such as using a car to run over pedestrians crossing the road.

Appropriate ethical standards must be established here. Ethical decisions are a very tricky issue because they involve at least three different factors or perspectives, including:

(a) the criteria for judging or regulating,

(b) context and environment, and (c) relationships and influences of the people involved, both internal and external. Failure to consider all of these various factors reasonably and correctly may lead to a suboptimal recovery of facts and a suboptimal ethical judgment.

If both agency and responsibility in car driving become less dependent on the individual performance of the driver and more dependent on the technology in the car such as ADAS and ultimately ATS, how will society deal with the potential change in the performance of agency as these responsibilities shift from individual actions to the vehicle itself? The logical answer is that the manufacturers of autonomous vehicles will be considered the trusted agents because they assemble and create the technology platform that controls the agents. Of course, car manufacturers do not want to accept this risk unless the market provides them with appropriate economic incentives, which it does not currently, but may in the future.

1. You are driving a vehicle along the highway at 100 km/h and an animal suddenly jumps out in front of you. You swerve to avoid hitting the animal, but in doing so you hit another car. You have a choice, and you do it. You will be liable for the accident, but you saved the animal. But what if the person you hit is significantly injured?

2. You are driving a vehicle along the highway at 100 km/h and someone suddenly appears in front of you from nowhere. You swerve to avoid hitting the person, but in doing so you hit another car. You had a choice, and you did it. You will be held liable for the accident, but you saved a life. But what if the person you hit is significantly injured?

3. You are in a self-driving car, traveling on the highway at 100 km/h, when suddenly a person appears out of nowhere in front of you. It is up to your vehicle to decide that it cannot hit the pedestrian nor can it hit another car at such a high speed diagonally, otherwise the passengers of that car and the owner of the car may be injured. The car quickly slows down and turns, and eventually hits the wall, and the vehicle is seriously damaged. You are not responsible for the accident, but the self-driving car is. The pedestrian was not knocked off the road and was saved. Other vehicles also did not get into an accident because the self-driving car is programmed not to hurt others.

   You have a spinal cord injury, serious one, and you blame it on the autopilot program.

We can see that in all three cases, the situation and ethics are crucial to the possible consequences, and all require decisions to be made within seconds. How the car is programmed in terms of liability will directly affect whether it will lead to serial lawsuits.

Some question whether the world is ready for the transfer of control of a car from the driver to an autonomous system. For many, remaining in control of the car is the best way to ensure that humans are not harmed by self-driving cars, which are not yet capable of making such complex decisions.

Perhaps the best response is to continue experimenting and proceed with caution. Technology often solves problems in unpredictable ways. When the technology is installed in the self-driving car of the future, the results will be the best performance within the specification, context and individual situation.

1In this article an automated driving technology system (ATS) is any technical system (which may include very significant and complex interactions of electronic and artificial intelligence technologies) that has been programmed by its inventors to choose between relevant available possibilities in various situations (it is able to make choices), with the final outcome of the event being determined by its response. This means that the ATS is a responsible agent because it has the relevant authority to take all actions.