Artificial Intelligence: The History of Self-Driving Cars and Future Challenges
The self-driving car industry is now one of the very hottest trends, causing a loud buzz not only in academia and industry but also from a socio-economic perspective. What once was a dream is now a reality, holding fruitful promises for humanity, but as always at a price, with an impact that could fundamentally change our economic and social systems.
From Science Fiction to Reality
What distinguishes man from other creatures is our brain that makes us always restless to experiment, innovate and create new devices that propel us forward in our quest for comfort and maximum utilization of time and resources. We shall not dive deep into the moral issues that accompany the invention of destructive weapons, leaving that to another discussion. However, it’s worth mentioning that the same mental power that makes us yearn to create can also be put to use for unprecedented destruction. It is this reason and this reason alone that I find the most concrete foundation for our drive to build artificially intelligent systems.
One of the first documented attempts towards automation was found in Al Jazari’s “The Book of Knowledge of Ingenious Mechanical Devices” in the 14th century. The book explains in a highly illustrative way how 100 autonomous mechanical devices function, with guidelines and very fine diagrams to reconstruct them. The functions described in his book served different applications in automating time tracking, agriculture, irrigation and even the playing of music. It was also discovered that other attempts had been made earlier in ancient Egypt and in the Far East as well.
Leonardo Da Vinci drew one of the very first engineering designs for a self-propelled car, for use in theatre. Italy’s Institute and Museum of the History of Science in Florence reconstructed the model from Da Vinci’s drafts and it worked — to everybody’s amazement. It was a spring-powered car that could drive itself forward and had been programmed to steer, at certain angles. Again, this was all autonomous and without direct human intervention.
In the Google News Archives, I found an interesting story about an autonomous car driving through Broadway in the United States. The story was published in The Milwaukee Sentinel on December 8th, 1926. Even though from a technical perspective it wasn’t a completely self-driving car, it was controlled by radio signals to steer its way through the streets. To many of us, this is not a new concept, having played with RC cars in our childhoods. A more successful trial was conducted in 1977 by the Tsukuba Lab in Japan where the car followed a track made of white street markers, at a speed of approximately 30 Km/H.
Autonomous cars existed for a long time in science fiction novels and Hollywood screenplays. Men have always dreamt of cars without steering wheels, where they can sit in comfort while the car brings them to their destination, as is explicitly depicted in 1935’s The Living Machine by David H. Keller. Isaac Asimov described a similar concept in Sally in 1953. It is worth mentioning that Asimov was rather a skeptic about robots in his writings. The idea of self-driving cars came up again in 1976 in Imperial Earth by Arthur C. Clarke. I wouldn’t dive much into screenplays that include such a concept as I presume many of you will have already watched Star Wars and Transformers, among others.
The Current Landscape
Fortunately, we didn’t have to wait until the year 2050 to witness this dream coming true. Even as we speak, numerous autonomous cars are driving around on Earth as they are on Mars. Advancements within the industry are being pioneered by tech giants in Silicon Valley, Europe, and Asia.
Though the scene lacks regulations and legislation, the US Department of Transportation has outlined five levels of automation, whereby level zero constitutes human-directed transport (or what we have been doing all along) and level five entails leaving full control to systems to control navigation within different landscapes, with the same accuracy and safety measures achievable only by humans so far. In between these levels are hybrid mixtures of manual and automatic navigation. These levels control the approach to automation and each company’s policy towards achieving both short and long-term goals with research and development. Moreover, they also help to identify different insurance categories for such vehicles.
Audi has released a level three compliant production car for sale in the market. The A8 luxury sedan has a traffic jam autopilot that smoothly maneuvers its way through and allows drivers to relax from boring stop-and-go driving on congested highways. It lends itself to the autonomous system only under certain conditions and functions on standard roads with clear lane marks and exit points. Moreover, it has some nifty features like auto parking. Being the first car to benefit from the new driving laws doesn’t make it immune to legal hurdles. It still has to bypass laws in every state and every country before they come to action.
Google has conducted experiments for level four semi-autonomous driving. The results were astonishing, as reported by volunteers taking trips on the highways. Even though there were no obvious technical defects, the volunteers endangered themselves in an unexpected way. Humans tend to enjoy the luxury of autonomous driving and are all too willing to trust the system to do the driving on its own. Meanwhile, they become preoccupied with their phone or any other distraction. It takes a few seconds to gain control back from the autonomous system, but these seconds can mean life or death on the road if something happens that requires a quick response. That’s why Google has decided to switch to level five automation. A prominent application of the self-driving cars pioneered by Google is Google Street View, which could be tracking or recording you right now!
Tesla had a true setback in 2016 when a semi-autonomous car hit a large vehicle and, as a consequence, the volunteer occupant of the car died. This accident posed many questions about the reliability of self-driving cars and their readiness for the road, even though in later reports it was found that the driver had received many visual and audio warnings to take back control of the system. This incident didn’t stop Tesla from upgrading to Autopilot 2.0, with level five fully autonomous features. Tesla’s CEO, Elon Musk, plans to make self-driving the de-facto mode by 2019. The first milestone towards this goal is a demo of a vehicle driving the full length of a journey from the Atlantic coast to the Pacific.
The race between manufacturers, including BMW, Mercedes, and Nissan, is heating up. The Chinese tech firm Alibaba has joined the race by announcing a self-driving car produced based on big data made available online. Indeed, the introduction of cloud computing is a game changer in this regard. This is not a car connected to the internet, but a car fuelled by data available on the cloud. Alibaba has cooperated with Baidu and Didi, two other Chinese tech firms ranked as equivalent to Google. All new autonomous features ranging across the whole spectrum of the five levels of autonomy have been produced, with notable advancements in low-level electronics, sensors, chips and software algorithms. By the dawn of the year 2018, Nvidia announced a processor chip optimized and dedicated only to self-driving technology.
The auto industry led the Industrial Revolution in the 20th Century and now self-driving cars will play a major role in the age of Artificial Intelligence, of which we are witnessing the dawn. They bring many changes to the lifestyles of individuals and significant effects on the economy. They penetrate deep into the heart of many side industries and will also create new industries around them. Not only do they have the potential to save commuters’ time, but also the ripple effect caused by this new technology will mandate a fundamental change to the business models of many industries, especially shipping. Shipping companies will transport more goods at a lower cost and risk. Free roads will mean less transport time and higher road capacity.
The travel industry will also be impacted. People will tend to travel more, ironically as they sleep, with their cars driving them from city to city. The traditional flight and hotel booking model won’t fit anymore and the need for innovative, unorthodox business approaches will be all the greater.
Mobility-as-a-Service (MaaS) is a new term coined to represent this new trend. As many existing drivers will become passengers, the number of cars owned by a family shall decrease and more people will prefer shared transportation — this has already started happening. New businesses will form around MaaS, either as direct services to consumers (B2C) or as part of a business-to-business (B2B) model. In the United States, an economic study was conducted of the gains resulting from the introduction of self-driving cars, which predicted an increase of USD $3.8K per capita, which is USD $1.2 trillion overall. This means that the global economy will experience a major shake-up; who will be a winner and who a loser is uncertain, but it promises inclusive benefits overall.
The phenomena will result in a noticeable increase in the Vehicle Distance Travelled (VDT) for elderly people, people in rural areas and people with physical disabilities. Hence, they will become more exposed to better educational and healthcare opportunities. Roads may well become safer since 93% of car accidents resulting from human error would be eliminated or at least reduced. It is worth mentioning that traffic accidents cost the United States USD $900 billion every year. An increase in unemployment rates is inevitable and some jobs will even vanish. However, the demand for traffic police, medical care, administrative and maintenance services for roads will decrease due to the lower accident rate. These resources, once freed, could be allocated to different sectors. Due to the convenience and higher VDT however, mechanics and car maintenance centers could find their incomes increasing, as their services will often be in demand.
A new ecosystem will grow and flourish around the new industry, especially when it comes to high-end tech specializations. Governments should take pre-emptive steps towards social equality, social pensions and distributing an equal share of the gross domestic product per capita in order to counter the negative impacts of introducing this new technology too quickly, depriving people of their livelihoods. A lot of high-level questions face our governments. How should transport systems be organized? How should new cities be designed? How should old cities be rearranged? How much of rural land use should be sacrificed to build roads compatible with autonomous vehicles?
Legal and Technical Challenges
We’re still a long way from having an optimized, reliable, fully autonomous self-driving car. Learning processes and empirical data have to be improved upon, in order to build confidence in the new technology. The demand for engineers to optimize their algorithms is relatively high, so as to meet the safety criteria imposed by legislators before introducing self-driving cars to the market. Powering self-driving cars with electricity instead of fuel is desirable, to reduce global warming. Meanwhile, there is still a legal dispute over who is liable in the event of a collision involving an autonomous vehicle. Is it the driver or the manufacturer?
Due partly to the recent boom in Artificial Intelligence (AI), self-driving cars will continue to develop more sophisticated features, affecting the perception of the wider market. What was once a far-off dream is now a reality on the ground, with opportunities to seize and risks to be overcome.