Promises and nightmares around autonomous vehicle
In a world appearing always more chaotic, driverless vehicle project seems highly supponent in its desire to completely automate cars eliminating human driver, meaning implementing cars which are able to master all surrounding events to smoothly arrive to the suggested targeting place. We will get into the car without worries, and our intent will be only to relax and engage in some ways, sure to be carried wherever we asked, closed in a sort of cocoon that works both as a form of teletransportation and as immunization from the simpatethic contaminations of life events.
Warned by the continuous procrastinations regarding real product/service launches, I suspect that automotive industry is more interested in the collateral benefits of overall project instead of the final goal. Being an industry capable to move huge investments, as well as launching exaggerated challenges, we could agree with this strategy. On the other hand, what is our user experience as car owner nowdays?
For the majority part of worldwide population that dwells in the metropolitan areas is only in advertising that cars go smoothly and silently through lush and fluid landscapes. Meantime, in the same promise lands of new autonomous cars movies suggest to organize choreographies for mass dances on the same roads supposed always blocked by the traffic jams (La la land).
To be clear, these huge investments could be devoted in the public transport but, being private capitals, the liberalism ideologies running everywhere impede a wise solution.
However, this kind of technological developments reserves many interesting implications if we do not limit our observations to the news supported by mainstream marketing that scratches only the surface. To say, there will be great changes in the automotive industry in terms of production or supplay chain, and likely many jobs involved in the intermediary tasks will fade, while the power of car makers will rise due to the central control of areas as design, implementation and selling. But we could even think of the huge changes in terms of consumption and, consequently, logistics because customers could prefer to access car service based on the specific and momentary needs: instead of buying a peculiar car, a service could be transformed in a sort of car valet ready to delivery sport car, family car or off-road vehicle. Other arguments such as legal or ethical ones – what happens when a car failure provokes incidents, injuries or death – can be yet more controversial.
Nowadays we are still in the inebriate stage of novelty trying to handle every new opportunities in a positive way. But, in the wake of rising concerns derived by the notion that any globalizing challenges will bring some negative effects, observations looking at project in a more realistic and disinchanted way seem to increase.
Propositions and challenges about automation
As usual when a new industrial line is emerging, autonomous car project could be a more interesting matter if intended in terms of spillover effects, looking at the positive stimulates and actual consequences (projectual and economical ones) spreading in many and often unsolicited industry sectors.
A big shift of innovation involves an ecology of activities that do not have a well-defined borders, suggesting to prompt large preemptive actions in order to catch all opportunities. Likely, financial plan aiming to enhance safety, mobility and sustainability prepared by Obama’s Administration (4 $ billions) moves just in that direction.
Indeed, we need vision and large shoulders to face that magnitude of changes, a work of sowing that doesn’t matter to be accused to waste if some seeds could reveal themselves (but only after changes happened) unuseful.
To be clear, the idea of incremental project steps is embedded in the autonomous car evolution. For example, following the National Highway Traffic Safety Administration (NHTSA), only the last 6th stage reckons on a “full automation” vehicle capable to move “under all roadway and environmental conditions” without an human driver.
In effect, our cars are already benefiting from this evolutive process that adds devices that automate functionalities previously managed by human drivers.
The automatic gearbox is a clear example, but other less intuitive devices are installed to augment our skills as car driver using mechanisms commonly reserved only for car sporting drivers (such as EBS, ASR and so on).
However, to reach the ultimate goal autonomous car must get the most hard feature: environmental control. During movement, and instant by instant, car must be aware of its spacial collocation in reference to all surrounding elements. From this point of view, the challenge is very tricky for needed to link and relate with every surrounding elements – static and dynamic ones –, considering possible configurations as never well known because each element in the play can change suddenly.
There are many project aiming to build very innovative mapping systems capable to recreate, in real time, the reality in 3D. Really, these developments recall us videogame contexts, as well as the fact we live in times when it’s easy to pass from the science fiction to science fact!
But, as we can image, task is not easy, and at least for two reasons. First of all, the need to control our and others position in a moving landscape with absolute certainty. It requires a relentless work involving information processing feeded by both live sources (as sensors) and structured archives containing information (as static maps) useful to sketch a scenery that must be always continuesly updated on the base of the actual data, in a ricorsive process in which the same autonomous cars could update the maps for releasing them to next users.
The automotive industries and tech companies working on maps are developing solutions taking into account different computing architectures. Many companies centralize computing capacity to process a such operations in cloudified sites that, by wireless communications, exchange data and information with cars in order to update models and coordinate actions. Others prefer to distribute the computing capacity in and among the same cars so that workloads are spread and nearer to the live fields. To have a better idea about a such kind of developments you can see the interesting article by Ars technica referred in the bibliography.
Forecasting the unforeseeable
By the way, targeting a full autonomous car quality stimulates the New York Times to report five urgent and common worries: unpredictable behavior of other human drivers; indecipherability of road lines (for usury, lacks or other causes) to get correct indications; having out-of-dated maps; difficults to individuate potholes, that can be hidden under every surfaces; having to make tough decisions, for example, who to harm between pedestrians and passengers when they could be involved in an inevitable accident.
But another kind of questions is rising lately, helped probably by the huge increasing of hacking actions as well as the intentional distribution of fake news on internet. If computing and communication architectures are relevant aspects to implement autonomous car project, another important element is the risk that some devices can be prone to receive or transmit bad or deceptive information.
To highlight the problem, the Londoner artist James Bridle, famous to elaborate works inquiring about the extension of digital world toward physical world, has created a work titled “Autonomous Trap 001”. It shows an autonomous vehicle in a parking inside two circle lines indicating controversial meaning (get in/do not cross).
What you’re looking at is a salt circle, a traditional form of protection—from within or without—in magical practice. In this case it’s being used to arrest an autonomous vehicle—a self-driving car, which relies on machine vision and processing to guide it. By quickly deploying the expected form of road markings — in this case, a No Entry glyph — we can confuse the car’s vision system into believing it’s surrounded by no entry points, and entrap it” (2017, vice.com.)
James Bridle’s provocation could appear naive, but the answer recently released by systems engineering Andy Birnie to a journalist seems to reinforce, even increasing, all doubts.
By definition, a driverless car has more control units, computing power, lines of code, and wireless connections with the outside world than a regular car today, which is why it’s more vulnerable to hackers, explains Andy Birnie, systems engineering manager at NXP Semiconductors. “A hacker can potentially take control of the car, through exploitation of a weakness, and could cause the vehicle to refuse to start, or to crash, or it could exploit the privacy of the driver, and [their] data, including financial information.”
However, a shift in the approach to security is under way, he says. “There is now more focus on the basics – applying good fundamental security to the critical areas including the interfaces that connect the vehicle to the external world; gateways, which separate safety critical systems from other car and infotainment systems; and networks that provide secure communication between control units – there can be over 150 in an autonomous vehicle.” In addition, major advances to security include “over-the-air” software updates that can patch vulnerabilities seamlessly in real time [but after bugs have been recognized… Editor notes], Birnie says (2017, the guardian.)
Meet the Artist Using Ritual Magic to Trap Self-Driving Cars, 5/18/2017, Vice.com.
5 Things That Give Self-Driving Cars Headaches, 6/6/2016, The New York Times.
The most detailed maps of the world will be for cars, not humans, 3/11/2017, Ars Technica.com.
Twelve things you need to know about driverless cars, 1/15/2017, The guardian.