Just a few years ago, the very concept of a self-driving car was the stuff of science fiction.
However, in the past year alone, technologists have made huge strides in developing autonomous vehicles, with varying levels of operational autonomy now being tested. Yet, much of the underlying technology required to build truly autonomous vehicles, such as cameras, sensors, radars and imaging systems is already available. We already let aircraft and ships make decisions and pilot themselves. What’s the red light holding up cars?
Clearly, more autonomous cars that build on a few fundamental technologies are a feature of the near future. Google has already logged more than 700,000 accident-free miles in its self-driving cars on the streets of Mountain View, California, without the assistance of a human driver. General Motors, Toyota, Mercedes-Benz, Audi, BMW and Volvo are all testing their own full or partially autonomous systems. Volvo has demonstrated its ‘autonomous valet parking cars’ in Europe, and Nissan has said it intends to launch autonomous cars by 2020.
The move towards the development of autonomous vehicles, however slow, is inevitable and is likely to progress with incremental developments such as advanced driver assistance systems, park assist systems, integrated vehicle health monitoring and autonomous systems.
The human and economic case
The human and economic case for greater autonomy is compelling. Moving from point ‘A’ to point ‘B’ involves multi-factor decision making about mode of transport, route, time constraints, safety, comfort, weather, luxury and convenience. In many cases, rules or machine learning-based decisions are likely to be objectively better than those made by a human. One of the central arguments in favour of more autonomous modes of transport is the safety value they offer. As is well documented, some 90% of road accidents stem from human error. At a trivial level, we are already safer and better off if we let a modern vehicle park itself.
More seriously, making vehicles autonomous globally could save US$300-400bn of societal and economic impact, taking into account loss of human life, social consequences and insurance claims. There is much to be gained by removing the human element. Ultimately, there is no reason that technology cannot make roads driver-free. As Google’s driverless car tests have proven, there’s a serious case that the self-driving car, with proper intelligent infrastructure, could be safer than the average driver.
Clearly, vehicles equipped with intelligent technologies that enable more efficient decisions and driving styles can provide improved fuel efficiency. There is also the potential to improve journey times and traffic flow, freeing up owners’ time. Technological cars also open the door to solving one of the great conundrums of the industry. A connection to the OEM can continue to provide updates, patches and improvements, and deliver an ongoing relationship with the consumer. The door is open to OEMs to do what has eluded them for a century: to keep adding value after the car has left the dealer forecourt.
The ecosystem: physical, legal and economic infrastructure
Apart from the primary technology required to build road-safe autonomous vehicles, huge regulatory and infrastructure challenges remain. Autonomous vehicles make sense and will offer a viable alternative to manually-controlled cars only if the infrastructure is in place to support them.
Just as electric vehicles require an ecosystem of charging points that can quickly charge a vehicle without bringing down the electricity grid, so too will autonomous vehicles require significant investment in new infrastructure. Intelligent traffic lights and smart lanes with sensors to assist automated parking are just a start.
Moreover, specific safety regulations and traffic rules within individual countries must be taken into account. Any autonomous vehicle system must be adaptable to any given regulatory regime. What happens when the cars cross national borders? Can we expect a self-driving car to drive on the left of the road onto a ferry at the UK port of Dover, and have it drive off on the right side of the road when it reaches Calais in France? Clearly, cars must be programmed to tailor their intelligent drive systems to different geographies and cultures.
Then there are less tangible factors to consider. It has, for example, been argued that autonomous vehicles, built with advanced sensing and tracking capabilities and constantly monitored, pose a threat to privacy. While this feature will improve vehicle performance, it creates new security concerns and risks commercial misuse. Whilst most of us are comfortable to receive targeted advertising from, say, a free satnav programme, will we feel differently if we’re targeted on the basis of data from our car?
Finally, there is the cultural challenge. Are consumers ready to change a fundamental part of everyday life? Can someone give up the pleasure of driving? This final hurdle could well be the most significant; but there will be new rewards, such as, for example, sitting in the driver’s seat without being discombobulated by rush hour traffic!
How will the industry change?
Culturally and ergonomically, driverless cars need not resemble the cars of today. For instance, in a completely driverless car or digital cockpit, is there even a need for a steering wheel? And if human intervention is required, is there a better way to operate the controls than a steering wheel? We can also safely assume that the role of ‘intelligence’ within such cars must be significantly greater than it is today.
These changes are reflected in an already changing business ecosystem. OEMs are great at building desirable, functional machines, but increasingly recognise that partners are needed to develop, test and implement the artificial intelligence and machine learning that must underpin the next generation of cars.
The automotive industry has anticipated this change for a few decades, growing a partner ecosystem. It started with the inclusion and integration of electronics and artificial intelligence into cars. Next-generation cars include electronics in everything from braking systems, engine and powertrain, to body controls and infotainment, enabled with advanced driver assist systems (ADAS) integrating with multiple types of sensor inputs, radar and image fusion and analytics, not forgetting the electronics and software. System architecture, integration and system testing would be more challenging than ever before. With autonomous vehicles, such requirements will only increase.
With the primary technologies already falling into place, the greatest hurdle is not technological but cultural and commercial. Bluntly, we need a more agile commercial ecosystem to create and adopt technology faster than ever before. We need to define boundaries for our personal data. The value chain – encompassing developers, content and app providers, telcos, insurers and OEMs – still has to catch up with the implications of more autonomous vehicles.