A life-and-death tussle between rival billionaires,
outlaw hackers and desperate taxi firms is playing out on city streets around the world. And the quest to be first mover in this brave
new world of self-driving automobiles will create big winners and losers. So how close are the major players to true
automation? And what’s at stake if they get it wrong? Join us trackside as we examine who’s winning
the self-driving car race. The manner in which we zoom around our towns
and cities is ripe for revolution in a way not seen since the birth of the internal combustion
engine. Demand for autonomous vehicles is set to surge
by tens of millions in the coming years. And Goldman Sachs suggests the rise of so-called
robo-taxis will expand the global ride-hailing business from present revenues of $5 billion
to $285 billion by the end of this decade. Imagine trillions of hours of productive time
clawed back because we won’t be gawping at the road ahead. Then imagine how pleasant our city centers
might be without endless hectares hogged by greedy parking lots. To reach this utopia, self-driving vehicles
need to ascend the 1-5 scale established by the SAE, or Society of Auto Engineers. Let’s look at that scale. Level one automation is the most basic and
covers things like traffic-sensitive cruise control that uses lasers to monitor traffic. It’s been available on some Mercedes models
since the late nineties. Honda introduced a related feature called
‘lane keep assist in 2008. Level two automates a couple more factors,
including limited steering assistance. It can raise the alarm if a human driver is
tired, and can nudge along in slow-moving traffic. Many new cars have this built-in as standard. Level three can do most driving by itself
but requires human hands on the wheel at all times just in case. Level four is where it gets interesting – fully
autonomous motoring, albeit only in strictly defined areas, like cities, that are pre-mapped
to the nth degree. Level four means drivers can stop paying attention,
but they can’t go everywhere. Finally, level five means full independent
automation, to the extent drivers can work, play or doze and still get where they’re
going in one piece. Now that’s out of the way, let’s meet
some of the runners and riders in the self-driving race. Perhaps the most eye-catching, not to say
eccentric, entry in the contest is Comma AI. The comma was founded by legendary tech provocateur
and Olympic-grade scowler George Hotz. Hotz rose to fame as the first hacker to crack
an iPhone, aged only 17. He’s since been in hot water with Sony for
hacking their PS3. Hotz joined the self-drive race in unconventional
fashion after he upgraded up his own Honda Acura to be self-driving, kinda, with a glovebox
full of electronics and a shop-bought lidar sensor. Hotz’s commercial product, the Comma Two,
is roughly the size of a smartphone and plugs into assorted modern Toyotas, Hyundai’s,
and Hondas, and offers modest level 2 automation. His promise? To ‘Make driving chill’. Not too chill, however. A rear-facing camera in the Comma device,
which costs $999 and fits snugly under the rear-view mirror, will pull you up sharply
if it senses your eyes drifting off the road. Another upstart you may not have heard of
is Nuro. Based in California, Nuro is making great
strides in self-driving vehicles, although probably not in the way you’re picturing. Nuro specializes in low-speed delivery robots
optimized for residential neighborhoods. In Texas, Nuro vehicles have been busily delivering
pizza for Dominoes, or groceries on behalf of supermarket giant Kroger. Nuro’s business model is especially attractive
to skittish investors, as self-driving vehicles that only move at around 25 miles per hour
are unlikely to kill anyone. Among the more glamorous contenders for the
self-driving crown is of course Tesla. True to form, our old mate Elon Musk promised
a self-driving Tesla could complete a self-driven trip across the US in 2017. That never happened. Still, thanks to the way Tesla software updates
over the air, every model that rolls off of the production line currently has the ability
to go to level five. Musk just needs those pesky pen-pushing regulators
to give him the chance. And besides, countless drivers as we speak
are using Tesla’s lane centering, traffic-aware cruise control, self-parking, automatic lane
changes, and semi-autonomous navigation. Tesla’s vast fleet is collecting vital data
all the time to refine and improve its algorithms, running though twin onboard Tesla chips that
are said – by Musk, naturally – to be best in class. Another big dog in the fight is GM. Through its partnership with AI upstart Cruise,
the old-school Detroit maker has been running self-driving electric Chevy Bolts around San
Francisco since 2017. The thinking being that if they can navigate
the hilly, chaotic, weatherbeaten streets of San Fran, they can cope pretty much anywhere. So why don’t we all have one by now? For the answer, we need to think about just
what a fully self-driving future will look like. Many analysts think the current model of car
ownership – where families buy one or two vehicles that spend most of their time idle
on driveways or parking lots, will vanish. For most people, and most applications, some
version of ride-hailing or on-off ride sharing – perhaps akin to a Netflix subscription
– will make more sense. So those familiar lines of traffic, full of
stressed-out individuals sat in their five-seater two-ton metal boxes, should be a thing of
the past. As such, instead of trying to convince everybody
to buy pricy autonomous Chevy Bolts, GM is putting its considerable resources behind
a six-seater robot carriage named the Cruise Origin. Set for production very soon – GM’s factory
in Detroit is reportedly ‘finalizing tooling’ – the Cruise Origin promises a slicker experience
than regular public transport, without the high overheads of private car ownership. Crucially, it also means GM can install the
very latest and most expensive LIDAR sensors, which in themselves would put a conventional
family car outside most consumers’ price range. Plenty of other companies are pursuing this
‘mobility as service’ model – as opposed to the likes of Tesla, who want to sell you
an actual car. Take Zoox, the stylish upstart recently purchased
by Amazon for a cool $1.3 billion. It’s been trialing ride-hailing service
in Las Vegas, with the suggestion Zoox vehicles could easily be re-jigged to ferry packages
around for the company’s deep-pocketed new overlords. For now, though, the undisputed leader in
the US robotaxi race is Waymo. A subsidiary of Google parent company Alphabet,
Waymo is already raising billions in private equity capital, and impressively runs a fleet
of some 600 driverless robotaxis around Phoenix, Arizona. This is a solid level-four, mostly autonomous
service, by the way – not only can the passengers relax and take their hands off the wheel,
but they don’t even need to know how to drive at all. If the passenger is distressed at any time
they can punch a button to speak to a human ‘ride support agent’ and customer feedback
is sought throughout the ride. Waymo already has millions of miles of public
road driving under its belt, and the service – partly offered through ride-hailing app
Lyft – is popular. Still, critics point to the fact the comparatively
tiny region of Phoenix in which the service operates is a doddle compared to most driving
scenarios, and that the company has singularly failed to deliver on its promise of tens of
thousands of vehicles rolled out by the start of this year. The rewards for whichever company gets it
right will be huge. Analysts reckon human drivers account for
up to 80% of running costs. By removing the driver from the equation,
taxi profits can and will soar. Not to mention trucking, another area where
Waymo is leading the pack. Some 70% of goods in the US travel by truck,
so any system that brings down costs, without sacrificing safety, would be a game-changer. Still, the potential downsides are greater
– a fully-laden artic gunning along the highway at 70mph can do an awful lot of damage
if anything goes awry. For now, taxis are where the world’s attention
is focused. And as you might expect, the big taxi firms
have done their best to dominate the space. Uber has spent years working on the problem
through its ATG – advanced technologies group – subdivision. Despite its mammoth size and profile, Uber
has yet to turn a profit through its core business. As such, the move into autonomous vehicles
is no abstract academic aspiration. It’s about Uber’s commercial survival. Still, various costly setbacks – and the
tragic death of an Arizona pedestrian in 2018 – prompted Uber to recently sell its AV
operation to Aurora, which is itself closely aligned with Japanese giant Toyota. To be clear, Uber remains a stakeholder in
Aurora, hoping when – if – the technology finally matures it’ll be able to use Aurora-powered
robotaxis without absorbing all the punishing R&D outlay. Which is similar to rival Lyft’s situation. In 2017 Lyft launched its ambitiously named
‘Level 5’ division, with the promise that by 2021 the majority of Lyft rides would take
place in autonomous vehicles. Thousands of engineers were duly hired at
a shiny dedicated Palo Alto facility, with big-ticket AI acquisition Blue Vision Labs
alone setting the taxi giant back a cool 72 million bucks. That dream, alas, came to nothing. Lyft also sold its Autonomous Vehicle arm
to Toyota, saving a reported $100million a year in apparently bottomless r&d spending. The ace up Lyft’s sleeve, for now at least,
is its cozy relationship with Waymo. If Waymo can clear the necessary regulatory
hurdles, Lyft will benefit from that. But what if we told you the real self-driving
leader is none of these high-profile big beasts? Traveling under the radar, so to speak, is
Mobileye. Formerly a Tesla partner – before leaving
after a dustup regarding Elon Musk’s attitude to safety – Mobileye is an Israeli firm
bought by Intel for $15 billion in 2017. Mobileye’s plan is to license its self-driving
hardware to legacy automakers – the likes of BMW, Volkswagen, and Nissan – which neatly
solves the thorny problems of mass production and rollout at a stroke. Having a chipmaker as its sugar daddy role
is also helpful from a hardware development angle – Mobileye’s sophisticated split-beam
Lidar is industry-leading, using clever doppler shifts to calculate distance and speed with
unprecedented accuracy. Mobileye’s greatest advantage is that its
stack could be used for both private ownership – like Tesla – or public rideshare models. Delivery startup Udely has already committed
to Mobileye technology for its promised fleet of 3,000 courier bots. The dark horse in this race, as so often,
is China. A lively plethora of tech startups like AutoX,
pony.ai, and WeRide compete in the space with Uber equivalent Didi Chuxing and even search
giant Baidu’s self-drive project. The Chinese aren’t quite level 5 just yet,
but thanks to the country’s more entrepreneur-friendly regulatory framework we can expect to see
widespread commercial robotaxis at the very least in China itself, then exported to the
rest of the world. Baidu, for one, has already trialed self-driving
buses and reportedly has a fleet of more than 300 Level 4 test vehicles working daily in
23 cities. Remember iPhone hacker George Hotz? When US regulators arched a quizzical eyebrow
in the direction of his DIY self-drive plans he sniffily replied. ‘Would much rather spend my life building
amazing tech than dealing with regulators and lawyers. ‘See you in Shenzen.’ What do you think? Is China at an unfair advantage? Will you give up your own car in favor of
a ride-sharing app? Let us know in the comments and don’t forget
to subscribe for more racy tech content.
