Currently, there’s a lot of posturing among the major car manufacturers as they jockey for position in the autonomous vehicle chase. Most of them are predicting that there will be some form of self-driving vehicle on the roads by the early to mid-2020s – most likely as ride-hailing services or commercial transportation (set routes, set times).
Notwithstanding the optimism, and before we all climb into robotically chauffeured cars or have our online goods delivered by people-less vans and trucks, there are still many hurdles to be overcome. Not only from a technological, but also from a business, regulatory and user point of view. Trial and error, never-ending learning, infinite software updates and our new-old friend artificial intelligence are paving the road on which autonomous vehicles will cruise.
Quite rightly, people are both excited by and fearful of the prospect of truly autonomous transportation. Positive thoughts relate to the elimination of human error: an autonomous vehicle is unlikely to be pulled over for reckless or drunk driving, accidents due to drowsiness or heart attacks. But the thought of technology, literally with a mind of its own, driving on our open roads and neighbourhood streets, is also a scary idea.
It seems that the easiest part of the equation may be in perfecting the self-driving technology hardware, despite its many complicated aspects. Companies like Ford, General Motors, Nissan, Tesla, Mercedes and others are pouring billions into their own R&D, and hundreds of smaller companies are offering improved radars, cameras, lidars, maps, data management systems, and more to the major vehicle companies.
However, according to Nidhi Kalra, a roboticist who co-directs the Rand Corporation’s Center for Decision Making Under Uncertainty, what makes autonomous vehicles extremely different from even the most advanced regular cars, is not merely the additional hardware. Rather, it is the software that is required to make these systems functional, and the amount of it.
Yes, there’s virtually no end to the intelligent systems that are required to replicate human behaviour behind a steering wheel. In addition to getting to know our human driving techniques, the complex software environment also needs to include localisation systems, high-definition map overlays, perception systems, planning systems, not to mention all the software needed to make the vehicle go forward without a foot on the fuel pedal or a hand on the steering wheel.
Other even more human aspects to driving further add to the complexity: car cultures differ from city to city, driving habits vary from courteous to aggressive and everything in between, and cities have their own style of driving that makes things flow. All this software gets updated all.the.time, so when it comes to the operating systems, self-drive cars are likely always to be a work-in-progress. It’s a comforting thought if you’re a software developer, but a bit uncomfortable if you’re a passenger in transit when the updates come through.
The safety-related updates that software in autonomous vehicles need to undergo are extremely rigorous and time consuming, and therefore also costly. This makes it difficult to gauge when the safety measures will be deemed adequate – if ever – and could also mean that the frequency of software updates is consciously reduced. Researchers are working on ways to speed up that process, to get important robotics safety updates proven, out and patched quickly.
However, the environment is in a constant state of flux or change. Even as a system is perfected according to the conditions it perceives today, so new and different conditions are unravelling all the time. Human instinct for adaption is innate, but machines need to review, assess, analyse and interpret situations and human reactions many times over in order to take appropriate actions. It is this constant machine learning that will help ceaselessly improve the software that interprets the sensor data, which is based on artificial intelligence and real-world examples to train the system.
Knowing how to build a self-driving vehicle that works is one thing: building millions of them and operating them is another entirely. Keeping vehicles on the road involves a myriad of other service providers to help keep the vehicles running: dealers, repair shops, fuel pumps, charging stations, parking garages. While this will create numerous business opportunities – some which have not even been imagined – the existing maze of interlinked companies built up over a century will need to be vastly modified to help maintain driverless vehicles.
Then, there is the conundrum around regulatory questions, which authorities around the world will still spend many years resolving.
Firstly, how do you change safety standards that have been written with human drivers in mind? How should vehicles without drivers be certified? How are insurance risks addressed?
As it is envisaged that the first self-driving vehicles in commercial operation are likely to be transit services, it may be easier to legislate for vehicles that operate within these static and limited confines on predefined routes. However, this is still a far cry from what sort of legislation will be required for truly driverless cars, and the development of these regulations will have a profound influence on the roll-out rate of autonomous vehicles.
And then, there are all the other departments that would require legislation and changes to ways of thinking and dealing with vehicle safety and road accidents or incidents. Imagine the first wrangles between autonomous vehicle manufacturer, regulator, insurance companies, lawyers and legislators in the event of an accident. We have no idea yet where this should even begin, and certainly no concept of where it could end.
Like in all healthy businesses and industry sectors, competition is inevitable: when will the price wars start between the major providers of autonomous vehicles and how will they decide to recoup their massive R&D costs? How will the terrain be shared between the big dogs? Who will dominate which section of the market? And how will these battles affect the timeframe for the roll-out of autonomous vehicles to ordinary users?
While there may be some cost saving, and certainly an increase in road safety, the modus operandi of the average person’s commute to work and completion of regular chores that involve transportation of one kind or another will change forever. As the cost of transportation decreases, and all the supporting industries essential to regular vehicle operation and maintenance have fallen away or been replaced with modified versions, the global economy will look quite different.
The pace at which these changes take place will play a monumental role in the pace at which autonomous vehicles may become a part of everyday life.
Similarly, humans may be able to change the landscape of their urban world, with different ways of mobility enabling a different community arrangement and lifestyle. With autonomous vehicles operating safely on predetermined routes, it could mean that people could travel more frequently and further, but to a smaller variety of destinations.
Will this be welcomed, or will there be resistance? Humans’ own preferences may still affect the development path of autonomous vehicles in ways that have not yet been foreseen by the vehicle manufacturers, whose eyes are naturally on the pot of gold that awaits them at the end of the self-drive rainbow.
As a positively prescient article in the New York Times* in 1908 declared, “The Horseless Carriage means trouble”. We should therefore all take a breather to wonder how our world will change when human constructs become the “drivers”, and we humans merely “the driven”. Coming sooner than we think. Or is it?
*(https://www.nytimes.com/1973/03/25/archives/the-horseless-carriage-means-troublele.html)
* Deseré Orrill is the co-Founder and Chairman of Ole! Connect, a digital marketing agency based in Cape Town. She is currently completing her MBA in Design Thinking, with a special emphasis on all things digital.