Share

Articles - Archive

04/09/2018
Driverless cars: five stimulating points

Self driving

 

From empathic mobility solutions to "peripheral vision" and new hi-tech developments and experiments, from deliveries to taxis, not to mention the luxury of time spent on-board totally autonomous vehicles and future smart roads just around the corner

Simone Cosimi • Wired / Audi Innovative

Detecting the exact moment of this transition is impossible and perhaps even pointless. Too many factors involved: besides, it’s not really a matter of if, but when. If, on the one hand, the European Commissioner, Slovenian Violeta Bulc, explained that “ totally autonomous vehicles will not circulate on our roads before 2050”, on the other hand, reality speaks of hi-tech solutions that are already largely available and fast developing. In other words, even if the EU considers the arrival of autonomous cars as something that will take place in decades from now, manufacturers – along with digital platforms such as Uber – are certain to be able to place such vehicles on the market as early as the next decade, using a number of these elements already on some of the most advanced vehicles currently on the market. BlackRock, for example, red circled 2025, underscoring that well before that turning point – which will raise automation to 75% within the following decade – 98% of all vehicles will be connected. This will in fact happen within three years. 

Meanwhile, Italy too has embarked in a series of road tests: the so called Smart Road decree approved in March gave the green light to such tests along with digital upgrading projects of the National road network. Motorways and other infrastructures will have to become “smart” if driverless vehicles are to become a reality. The deadlines for such a transformation, coincides with estimates by many of the industry’s experts: within 2030, V2X connectivity must become operative and allow information Exchange between vehicles and between vehicles and the infrastructures, pedestrians and so on, thus building an ideal digital ecosystem for driverless vehicles. Car makers, Universities and research centres are now already engaged in performing their tests despite a lengthy trail of legislative requirements.   

But where will this digital revolution hit first? And what elements will be affected both inside and outside vehicles? Techopedia recently considered a few of these aspects. Delivery vehicles will probably act as trailblazers alongside premium cars already equipped with level 3 automation systems, able to deal with driving in normal conditions leaving the driver to take over in critical circumstances or if the surrounding context changes. Just as with Audi’s Ai traffic jam pilot, active up to 60 km/h: the system take control of the vehicle autonomously managing acceleration, steering and braking, or other situations such as traffic jams or slow-moving traffic on motorways. 

Electric (and autonomous) commercial vehicles are already being developed and will soon hit the market managing the increasingly chaotic world of parcel deliveries from collection point destination considering elements such as traffic conditions, parking and the amount of pedestrians around. As always, a little patience is required in understanding where will all this start from, but the parcel delivery sector – as shown by consistent investments in drones and robots – will likely become the testing ground, although momentarily pushed aside in favour of autonomous cars.   

Level 5 automation, in which the system manages every situation without human assistance, is already being developed. Here too, the first application of this technology might not take place, as one would expect, on privately owned vehicles, but most likely on taxis, car sharing fleets or corporate vehicles. Watching a film, doing some work or listening to music after calling a car through an app, basically what happens already on a number of online platforms, but without anyone actually driving the car will be an epoch-making moment, no doubt about it, although at the moment still a bit futuristic.  On the other hand, growing concerns over employment and potential job losses are being fuelled by several solutions – like Nvidia’a Drive Px platform – which is making these concerns real. Soon vehicles will be redesigned so as to have more rational interiors: pedals, steering wheels and gear leavers will be a thing of the past substituted by simpler and essential commands. We cannot exclude, at this stage, that autonomous systems will find their perfect testing environment in contexts such as transport hubs, where vehicles moving on rails no longer need a driver, or towards special class of vehicles controlled and managed by public service (for elderly people unable to move independently for example).

The ability to park in full or partial autonomy, is something we are already growing accustomed to. However, park assist systems, particularly complex and crucial in large metropolitan areas and a cause of delays, productivity loss and pollution, will likely undergo further transformations. This thanks to high levels of automation – imagine a car that picks you up at home, drives you to work and then looks for a parking spot where “it” knows there are some – and huge collection of data. Places, spaces and time cross-linked according to trends and real time information will completely transform the once tedious work of finding a suitable spot, whether picking us up, driving us to destination or looking for a parking. All data that will be gathered by the vehicle’s on-board computer to be passed on to a dedicated cloud from where other vehicles could draw the needed information. A virtuous cycle able to transform congested neighbourhoods and give us back what we consider our true contemporary treasure, a real modern luxury: time. At first through a liberal and customized experience, and then cutting down idle times and stress.

The most interesting aspects, though, involve those traits we could consider as “intelligent”. The missing piece, so far, was human common sense. That infinite range of characteristics that, to say it as it is, might be responsible for a number of accidents, but the unsolved question is: how many are prevented or avoided by it? Assorted behavioural trends, sounds and noises, erratic pedestrians and so on: a human Babel with which autonomous vehicles will have to contend with. Some are already doing so, for example, the iSee startup, a spinoff of Boston’s MIT institute, is hard at work on this project: making the most of big data and neural networks so that vehicles might be able to “learn” and deal with any traffic condition. Even the worst. “ The human mind is extremely sensitive to physical and social signals – explained Yibiao Zhao, co-founder of iSee – and currently AI is still quite feeble in this sense. This is the missing piece. In other words, cars must grow to become more empathic.  

However, besides “empathy” and the ability to create a network through data, on-board devices and other vehicles, autonomous cars will be predictive. Evaluating speed and distance from pedestrians or other surrounding objects even when traditional cameras may no longer be able to, digitally glancing around even beyond the dreaded blind-spot. MIT’s Computer Science and Artificial Intelligence Laboratory (Csail) is hard at work on this. The project is called CornerCameras which uses reflected lights and shades to predict possible “surprises” hidden to the camera’s “eye”. Who knows, perhaps some of the recent cases involving autonomous vehicles could have been avoided if such a sophisticated system had been used. Who knows!

back to archive