Look Ma – no hands

What if we could have the ad­van­tages of cars with none of the down­sides? That’s the prom­ise held out by ad­vo­cates of com­puter-con­trolled, self-driv­ing ve­hi­cles. As they power ahead on solv­ing the tech­nol­ogy chal­lenges, Michael Cameron looks at the le­gal i

New Zealand Listener - - CONTENTS -

What if we could have the ad­van­tages of cars with none of the down­sides? That’s the prom­ise held out by ad­vo­cates of ­com­put­er­con­trolled, self­driv­ing ve­hi­cles. Michael Cameron looks at the le­gal im­pli­ca­tions.

It was sup­posed to be a nice day out. On Au­gust 17, 1896, Mrs Bridget Driscoll and her 16-year-old daugh­ter were tak­ing a trip to Crys­tal Palace in south-east Lon­don to at­tend a fete for the Catholic tem­per­ance or­gan­i­sa­tion League of the Cross. But the day had a tragic end. A mo­tor­ing ex­hi­bi­tion was also tak­ing place that day, and 45-year-old Driscoll was struck and killed by one of the new­fan­gled horse­less car­riages. The im­ported Roger-Benz motor car, on a demon­stra­tion ride for two pas­sen­gers, was trav­el­ling at a stately 6.5km/h, but it was too fast for the be­wil­dered woman, who had the du­bi­ous honour of be­com­ing the first pedes­trian in the UK – and pos­si­bly the world – killed by the new tech­nol­ogy that was about to sweep the planet.

Far from re­spond­ing with a reg­u­la­tory clam­p­down, law­mak­ers passed the Lo­co­mo­tives on High­ways Act less than three months after the tragedy, which ac­tu­ally re­laxed speed lim­its. The ac­ci­dent was re­ported widely, but Vic­to­ri­ans were fa­mously easy-go­ing about health and safety.

In any case, the arrival of the life-chang­ing au­to­mo­bile was keenly anticipated. In Au­gust 1899, the New Zealand Her­ald cel­e­brated the “pass­ing of the horse and the com­ing of the new era of motor cars”, which it con­fi­dently pre­dicted would “com­pletely change for the bet­ter the con­di­tions of city life” and rhap­sodised about “spin­ning along a coun­try road, up-hill and down-hill, in one of these cars, at the rate of twenty miles per hour” with “no fa­tigue, and lit­tle or no dan­ger”.

We have largely for­got­ten the so­cial ills that plagued our cities un­til the au­to­mo­bile con­signed them to his­tory. Horse ma­nure and urine caused dire san­i­ta­tion prob­lems; horses’ iron­clad hooves dam­aged street sur­faces; agri­cul­tural pro­duc­tion had to em­pha­sise fod­der crops such as oats and hay.

Ur­ban plan­ners were at a loss: re­strict­ing their use would be a cure worse than the dis­ease, be­cause horses were es­sen­tial for the func­tion­ing of the mod­ern city. But cars cleaned up the mess, even­tu­ally pro­vid­ing cheap transport to peo­ple who could not af­ford to buy and main­tain horses.

Our re­la­tion­ship to cars has many par­al­lels to the Vic­to­ri­ans’ de­pen­dence on horses: we be­moan their neg­a­tive ef­fects, such as road deaths and in­juries, air pol­lu­tion, ur­ban sprawl, oil wars, obe­sity and cli­mate change, and we try to re­duce our car use by rid­ing bi­cy­cles or pub­lic transport. But so far we have failed to re­duce the ef­fect of cars to any mean­ing­ful ex­tent. The only so­lu­tions seem to in­volve dras­tic re­stric­tions, which eco­nomic and po­lit­i­cal re­al­i­ties make very dif­fi­cult.

But what if we could have the ad­van­tages of cars with none of the down­sides? That is the prom­ise held out by ad­vo­cates of com­puter-con­trolled, self-driv­ing cars, who are ev­ery bit as en­thu­si­as­tic as their 19th-cen­tury coun­ter­parts were for the new tech­nol­ogy of their day.

Be­cause com­put­ers don’t get drunk, dis­tracted or make mis­takes, they will dras­ti­cally re­duce – even elim­i­nate – road deaths. Be­cause they are such care­ful driv­ers, more peo­ple will feel safe enough to ride bikes, and our chil­dren will walk and ride to school again. Be­cause they will fa­cil­i­tate ef­fi­cient app-based ride shar­ing, they will take ve­hi­cles off the roads and ease con­ges­tion and emis­sions. Be­cause they can be net­worked, the seething traf­fic of our cities can be ra­tio­nally co-or­di­nated, fur­ther eas­ing con­ges­tion and emis­sions. Be­cause they can drive much closer to­gether, they will in­crease the ca­pac­ity of our ex­ist­ing roads. Be­cause they can drive off and park them­selves when not needed, they will re­lease vast tracts of in­ner-city real es­tate cur­rently

More than 300 peo­ple die on our roads ev­ery year. If driver­less cars are safer than hu­man-driven ones, any de­lay will cost lives.

mo­nop­o­lised for park­ing, so car parks will be­come real parks.

Self-driv­ing cars will boost eco­nomic growth, rev­o­lu­tionise mobility for every­one, par­tic­u­larly the dis­abled, and make car own­er­ship a choice rather than the ne­ces­sity it is for many.

DAYS OF FU­TURE PAST

That’s the utopian vi­sion. But there is cause for scep­ti­cism. For starters, we have been promised self-driv­ing cars be­fore. In 1964, Gen­eral Mo­tors was pro­mot­ing the Fire­bird IV, a con­cept car that “an­tic­i­pates the day when the fam­ily will drive to the su­per­high­way, turn over the car’s con­trols to an au­to­matic, pro­grammed guid­ance sys­tem and travel in com­fort and ab­so­lute safety at more than twice the speed pos­si­ble on to­day’s ex­press­ways”. The tech­nol­ogy was in­ge­nious and im­pres­sive and it worked on test runs. But it came to noth­ing. It re­lied on ex­pen­sive buried ca­bles, which cost up to $200,000 a mile. It would never be eco­nom­i­cal to con­vert ex­ist­ing high­ways.

In 1997, the US De­part­ment of Trans­porta­tion funded a con­sor­tium of com­pa­nies and uni­ver­si­ties to de­velop and demon­strate a work­able pro­to­type of an au­to­mated high­way sys­tem. The demon­stra­tions went flaw­lessly, as fleets of Buicks drove au­tonomously in tight for­ma­tions us­ing a sys­tem of mag­nets em­bed­ded in the roads. Ar­ti­cles at the time breath­lessly re­ported that “high­ways of the fu­ture may fea­ture re­laxed driv­ers talk­ing on the phone, fax­ing doc­u­ments or read­ing a novel” and that the demon­stra­tions “il­lus­trate that the vi­sion of an au­to­mated high­way sys­tem that im­proves traf­fic safety and high­way ef­fi­ciency can be made a re­al­ity”. But again, it never be­came a re­al­ity: in­stalling the mag­nets was too ex­pen­sive.

We might be for­given for think­ing that the new hype about self-driv­ing cars is just an­other false dawn, but there is good rea­son to think it will be dif­fer­ent this time. Pre­vi­ous ini­tia­tives failed be­cause they re­quired

mas­sive in­vest­ments in pub­lic in­fra­struc­ture, but a con­flu­ence of new tech­nolo­gies, in­clud­ing GPS, cheap and ef­fec­tive sen­sors, high-def­i­ni­tion cam­eras and ar­ti­fi­cial in­tel­li­gence (AI), is mak­ing it pos­si­ble for self-driv­ing cars to func­tion with­out the need for any new in­fra­struc­ture.

Al­ready, this has re­sulted in pro­duc­tion cars with use­ful semi-au­ton­o­mous modes, such as the au­topi­lot on Tesla mod­els, which al­lows the car to drive it­self on a high­way, stay­ing in its lane, main­tain­ing a safe fol­low­ing dis­tance and chang­ing lanes safely on de­mand.

BULLISH OUT­LOOK

All these sys­tems still re­quire hu­man su­per­vi­sion and are not fully au­ton­o­mous by any stretch. But they are im­prov­ing all the time and lead­ing com­pa­nies are bullish that they will soon crack full au­ton­omy. Ford pre­dicts it will of­fer fully au­ton­o­mous ride-shar­ing ve­hi­cles on pub­lic roads by 2021. And

Tesla plans to have a car drive it­self from Los An­ge­les to New York early next year. Whether progress will be this swift re­mains to be seen. Even if it is, there are le­git­i­mate ques­tions about whether it will be so uni­formly pos­i­tive.

Re­duc­tions in con­ges­tion, emis­sions and en­ergy use may be tem­po­rary. Just as build­ing new roads is said to sim­ply re­sult in more peo­ple us­ing them, any gains made from au­ton­o­mous ride-shar­ing may quickly be eroded by peo­ple tak­ing more trips. We could end up back where we started, or worse off.

What about ur­ban com­mu­ni­ties that rely on foot traf­fic for their unique char­ac­ter? If every­one has su­per cheap point-to-point transport on de­mand, might our cities be­come fea­ture­less ur­ban deserts pop­u­lated by hy­per-ef­fi­cient peo­ple-mov­ing boxes? And what will hap­pen to the more than 40,000 New Zealan­ders who make their liv­ing as driv­ers?

The de­ci­sions we make will be guided by whether we think con­cerns about self-driv­ing cars are out­weighed by their ben­e­fits.

Of greatest alarm for most peo­ple is be­ing killed by a self-driv­ing car. We have be­come used to com­put­ers dom­i­nat­ing our lives, but so far they have largely been kept out of the phys­i­cal realm. Never mind that they are sup­posed to be safer than hu­man driv­ers: there is some­thing ter­ri­fy­ing for many about plac­ing their lives so di­rectly in the hands of ma­chines. The blue screen of death of per­sonal-com­puter in­famy may have been an­noy­ing, but at least it was only a metaphor; no one has died from one. And the death last year of Tesla driver Joshua Brown while us­ing au­topi­lot doesn’t re­ally count, be­cause he ig­nored re­peated au­to­mated warn­ings. But some­where, some­time a self-driv­ing car will make a mis­take and some­one will die.

NEW TER­RI­TORY

Our laws are silent on driver­less ve­hi­cles; whether they are even le­gal is a grey area. Some level of su­per­vised au­ton­omy seems to have al­ready been ac­cepted as le­gal, as demon­strated by the pres­ence on our roads of cars such as the Tesla Model S with its au­topi­lot func­tion. There is an ar­guable case that the fu­ture use of fully au­ton­o­mous

ve­hi­cles with­out a su­per­vis­ing driver is also le­gal un­der cur­rent laws, so it is quite pos­si­ble that we could see them caus­ing deaths on our roads be­fore too long.

So we have to make some de­ci­sions. Are we happy to mud­dle along with the sta­tus quo? Do we want to ban them? Or ex­plic­itly le­galise them? If the lat­ter, how do we en­sure safety? Should this be by way of ap­prov­ing spe­cific mod­els? Or do we want to en­cour­age faster up­take by be­ing more per­mis­sive?

The de­ci­sions we make will be guided in part by whether we think con­cerns about self-driv­ing cars are out­weighed by their ben­e­fits. If we de­cide they are ben­e­fi­cial, there is good rea­son to be in the lead­ing pack of ju­ris­dic­tions grap­pling with their arrival. They may cre­ate more lu­cra­tive jobs in the coun­tries in­volved in their de­vel­op­ment, and surely it is bet­ter to en­cour­age do­mes­tic and over­seas com­pa­nies to in­vest in the in­dus­try rather than sim­ply be a pas­sive con­sumer of its prod­ucts. Be­yond the busi­ness case, the ar­gu­ments are more com­pelling. More than 300 peo­ple die on our roads ev­ery year. If driver­less cars re­ally are safer than hu­man-driven ones, any de­lay will cost lives.

The first de­ploy­ments of au­ton­o­mous ve­hi­cles are likely to work in re­stricted cir­cum­stances – what the ex­perts call an op­er­a­tional-de­sign do­main. An­ders Lie is a spe­cial­ist with the Swedish Transport Ad­min­is­tra­tion, work­ing with Volvo on a large-scale self-driv­ing trial (he’s the man re­spon­si­ble for that an­noy­ing but ef­fec­tive sound your car makes if you don’t buckle up). He is pick­ing that the first com­mer­cial ap­pli­ca­tions will be high­way-driv­ing modes that work with­out driver su­per­vi­sion, or ride-shar­ing fleets of self-driv­ing taxis or mini-shut­tles, pos­si­bly oper­at­ing at first in re­stricted lanes such as tramways.

Other pos­si­bil­i­ties in­clude con­voys of au­ton­o­mous long-range freight trucks or small un­manned de­liv­ery pods. All of these ap­pli­ca­tions will have one thing in com­mon: “Highly au­to­mated driv­ing sys­tems will de­pend heav­ily on the prepa­ra­tion of de­tailed 3D maps, and some will be un­able to op­er­ate out­side the mapped ar­eas ex­cept on pre­scribed routes,” says Jim Sayer, direc­tor of the Univer­sity of Michi­gan Trans­porta­tion Re­search In­sti­tute. “Such sys­tems will also de­pend on sig­nif­i­cant test­ing car­ried out in the unique en­vi­ron­ments of the ar­eas in which they will be de­ployed – tak­ing into con­sid­er­a­tion lo­cal or re­gional dif­fer­ences in traf­fic-con­trol de­vices, road­way de­sign, weather con­di­tions and driv­ing norms.”

Google and Uber are among com­pa­nies that have in­vested heav­ily in map­ping and test­ing in Cal­i­for­nia, Ari­zona and Penn­syl­va­nia, which are likely to be among the first places where au­ton­o­mous ve­hi­cles are de­ployed. There has been some map­ping of New Zealand roads by satel­lite-nav­i­ga­tion providers such as TomTom, but none of the com­pa­nies de­vel­op­ing au­ton­o­mous ve­hi­cles has come here to map yet. And, with the ex­cep­tion of the ground-break­ing au­ton­o­mous shut­tle bus tri­als be­ing done by HMI Tech­nolo­gies (in­clud­ing one at Christchurch Air­port), test­ing has been lim­ited to a few demon­stra­tion runs.

THE LAW’S DE­LAYS

If we want to catch up with these

“Clar­ity in self-driv­ing law is go­ing to be in­cred­i­bly im­por­tant for de­vel­op­ers of the tech­nol­ogy.”

devel­op­ments, we need to think about how much our laws en­cour­age – or dis­cour­age – this kind of work. We need to de­cide who will be re­spon­si­ble when a self­driv­ing car causes an ac­ci­dent or whether it will be an of­fence if a driver­less car does not ex­change de­tails after an ac­ci­dent. Speed lim­its as cur­rently worded ap­ply to a “driver”, so a self-driv­ing car can­not be said to be speed­ing.

The first and big­gest is­sue is that we have not made ex­plicit le­gal pro­vi­sion au­tho­ris­ing cars to op­er­ate on pub­lic roads with­out hu­man su­per­vi­sion. Com­pa­nies might be re­luc­tant to in­vest in New Zealand if they have no cer­tainty that their busi­ness model will be le­gal.

“Our vi­sion is that self-driv­ing ve­hi­cles will be best op­er­ated within a shared-ride net­work,” says Justin Kintz, se­nior direc­tor of pol­icy and com­mu­ni­ca­tions for Uber, “and there­fore it is im­per­a­tive that any place that wants to ben­e­fit from self-driv­ing first es­tab­lishes pro­gres­sive rideshare laws, be­fore tack­ling self-driv­ing reg­u­la­tions.

“Clar­ity in self-driv­ing law is go­ing to be in­cred­i­bly im­por­tant for de­vel­op­ers of the tech­nol­ogy, be­cause the in­vest­ment re­quired for early de­ploy­ment is huge, and cer­tainty of stand­ing pro­vides con­fi­dence for busi­nesses.”

Kintz points to the ap­proach of Ari­zona, which has at­tracted a great deal of in­vest­ment and test­ing, in part be­cause of its straight­for­ward rideshare rules, but also be­cause of an ex­ec­u­tive or­der in 2015 spell­ing out the cir­cum­stances un­der which test­ing of self-driv­ing ve­hi­cles could oc­cur.

At first glance, it looks clear that the law could sim­ply re­quire cars to be sub­mit­ted for ap­proval be­fore they are al­lowed on pub­lic roads. A com­pany might de­scribe what it thinks its car can do and ex­plain why it is safe. A gov­ern­ment-ap­proval agency would then scru­ti­nise the pro­posal, test the ve­hi­cle and ap­prove or de­cline it.

This is prob­a­bly what will hap­pen un­der our cur­rent laws. For most ve­hi­cles, en­try to New Zealand is de­pen­dent on the model hav­ing ap­proval in Europe, where there is a very pre­scrip­tive process known as “type-ap­proval”. But there are com­pli­ca­tions in ap­ply­ing type-ap­proval to driver­less ve­hi­cle sys­tems.

To be­gin with, it is un­likely that govern­ments will have or be able to ac­quire the ex­per­tise to prop­erly as­sess these ve­hi­cles since it re­ally ex­ists only in­side the com­pa­nies de­vel­op­ing them and it is hard to get in­for­ma­tion out of these firms.

“Com­pa­nies are re­luc­tant to share their data with reg­u­la­tors with­out some as­sur­ance that in­for­ma­tion that they con­sider to be trade se­crets will not be­come part of the pub­lic record and dis­closed to their com­peti­tors,” says Brian Sou­blet, deputy direc­tor and chief coun­sel of the Cal­i­for­nia De­part­ment of Motor Ve­hi­cles. The im­pli­ca­tion is that any ju­ris­dic­tion in­sist­ing on such dis­clo­sure risks be­ing ig­nored as an in­vest­ment des­ti­na­tion.

It is sim­plis­tic to say that reg­u­la­tion al­ways ham­pers in­no­va­tion; some­times it can stim­u­late it. When ozone-de­plet­ing chlo­roflu­o­ro­car­bons (CFCs) were banned, re­place­ments were quickly found. But there is ev­i­dence that heav­ily pre­scrip­tive type-ap­proval reg­u­la­tion can sti­fle cre­ativ­ity. Re­quir­ing man­u­fac­tur­ers to use cer­tain ma­te­ri­als, for ex­am­ple, can dis­cour­age the

search for bet­ter ones. But in the ab­sence of reg­u­la­tion, we must rely on as­sur­ances from man­u­fac­tur­ers, and this raises many ques­tions, not least how the safety of driver­less ve­hi­cles can be mea­sured and demon­strated.

“Cur­rent known meth­ods for as­sess­ing the safety of sys­tems are not work­able for au­to­mated driv­ing sys­tems be­cause of their ex­treme com­plex­ity,” says Steven Sh­ladover, a re­search en­gi­neer and man­ager at the Part­ners for Ad­vanced Trans­porta­tion Tech­nol­ogy Pro­gramme at the Univer­sity of Cal­i­for­nia, Berke­ley, and a vet­eran of the US na­tional au­to­mated high­way sys­tem demon­stra­tion of the 90s. “Ve­hi­cles would need to drive billions of miles to achieve a sta­tis­ti­cally mean­ing­ful demon­stra­tion that they are sig­nif­i­cantly safer than av­er­age driv­ers.”

And whose safety are we talk­ing about? Is it ac­cept­able to pri­ori­tise the safety of peo­ple in­side the car over oth­ers? Does that mat­ter if the car is still safer over­all? If it is 10 times safer for oc­cu­pants but only five times safer for oth­ers, do we in­sist on equal­ity, or are we just grate­ful for any im­prove­ment? If we do in­sist upon equal­ity, will we be dis­cour­ag­ing the up­take of the ve­hi­cles with their lifesaving ad­van­tages?

A sur­vey pub­lished in Sci­ence last year showed that al­though most peo­ple ap­proved of au­ton­o­mous cars be­ing pro­grammed for equal­ity, they wouldn’t per­son­ally want one that didn’t pri­ori­tise their safety. And is there ac­tu­ally any­thing new about these dilem­mas? Or have man­u­fac­tur­ers al­ways had to bal­ance the safety of dif­fer­ent groups? As Bart Simp­son re­torted when his sis­ter ex­pressed con­cern about the propen­sity of SUVs to be in­volved in fa­tal ac­ci­dents, “Fa­tal to the peo­ple in the other car, let’s roll!”

THE TEACH­ABLE CAR

De­mands by the pub­lic for man­u­fac­tur­ers to dis­close ex­actly what a car’s ethics are and what it will do in var­i­ous hy­po­thet­i­cal sit­u­a­tions (sw­erve into the pole or plough straight into the pedes­trian?) can be dif­fi­cult for man­u­fac­tur­ers to an­swer. This is not nec­es­sar­ily be­cause they are re­luc­tant to tell you; they may not even know them­selves. The rea­son these ve­hi­cles are now get­ting so good is that many are, to a large ex­tent, no longer for­mally pro­grammed.

The prob­lem of writ­ing code to spec­ify dif­fer­ent cour­ses of ac­tion for all of the prac­ti­cally in­fi­nite sce­nar­ios a car could en­counter once seemed in­sur­mount­able but it is now be­ing con­quered through the use of ma­chine learn­ing. In­stead of de­tailed pro­gram­ming, the self-driv­ing pro­grams spend thou­sands of hours prac­tis­ing to learn what works best. De­sign­ers can’t nec­es­sar­ily pre­dict what their car will do, but they can say that, what­ever it does, it will be safer than what a hu­man would have done.

Many will find the no­tion that these cars are not pre­dictable more alarm­ing than any­thing else. Not know­ing why com­put­ers are com­ing to the con­clu­sions they reach can be dis­turb­ing. There have been cases in the US where ma­chine-learn­ing pro­grams used to as­sess the risk of crim­i­nal re­of­fend- ing turned out, after an investigation last year by the ProPublica web­site, to be racially bi­ased. And with the Volk­swa­gen emis­sions scan­dal fresh in peo­ple’s mem­o­ries, trust in man­u­fac­tur­ers’ as­sur­ances about their sys­tems can­not be taken for granted.

If these as­sur­ances are not enough for us, we will need to find other safe­guards. Putting ve­hi­cles through driv­ing tests or demon­stra­tions is an ap­proach favoured by Dave Verma, direc­tor of au­ton­o­mous ve­hi­cles for Auck­land-based HMI Tech­nolo­gies. Verma says “the tests would need to be care­fully de­signed to in­volve a level of un­pre­dictabil­ity to pre­vent man­u­fac­tur­ers from teach­ing to the test”. Some man­u­fac­tur­ers would ar­gue that such tests are su­per­flu­ous. What can a few hours with a test­ing of­fi­cer tell us that hun­dreds of thou­sands of hours of real-world test­ing and sim­u­la­tions can­not? “Not much,” says Verma, “but they have the great ad­van­tage of pro­vid­ing us with level-play­ing-field re­sults that any­one can un­der­stand and that seem less re­liant on hav­ing to trust the as­sur­ances of man­u­fac­tur­ers. There has to be a bench­mark that the pub­lic can trust as the min­i­mum ex­pected per­for­mance.”

If con­cerns arise, man­u­fac­tur­ers risk ex­pen­sive prod­uct re­calls and the US Fed­eral Gov­ern­ment is al­ready well equipped for those. “The Na­tional High­way Traf­fic Safety Ad­min­is­tra­tion has the au­thor­ity to iden­tify safety de­fects, al­low­ing it to re­call ve­hi­cles that pose an un­rea­son­able risk to safety even when there is no ap­pli­ca­ble fed­eral standard,” says the ad­min­is­tra­tion’s fed­eral au­to­mated ve­hi­cles pol­icy. There may be po­ten­tial for New Zealand to make greater use of these kinds of mech­a­nisms.

Self-driv­ing cars have many en­thu­si­asts, kin­dred spirits of that writer in the New Zealand Her­ald more than a cen­tury ago. The ben­e­fits can vastly ex­ceed the costs if they are man­aged well, and if we can avoid coun­ter­pro­duc­tive reg­u­la­tion. But sooner or later self-driv­ing cars will face their Mrs Driscoll mo­ment: a self-driv­ing car some­where in the world will make a mis­take and kill an in­no­cent per­son.

When this hap­pens, the re­ac­tion is likely to be a lot more vig­or­ous than that of the Vic­to­ri­ans to the first au­to­mo­bile death. The aim should be to avoid over­re­act­ing and not lose sight of the big­ger pic­ture. Ev­ery death is a tragedy. But at the mo­ment, our small coun­try is suf­fer­ing tragedies of death and in­jury ev­ery day that don’t even make the papers. The larger tragedy would be if a knee-jerk re­ac­tion jeop­ar­dised our op­por­tu­nity to seize the fu­ture.

Michael Cameron, the Law Foun­da­tion’s 2016 in­ter­na­tional re­search fel­low, talked to ex­perts from Sil­i­con Val­ley to Sin­ga­pore about the po­ten­tial of driver­less ve­hi­cles in New Zealand. His re­port, to be pub­lished in Novem­ber, will make rec­om­men­da­tions for law re­form.

“Ve­hi­cles would need to drive billions of miles to achieve a sta­tis­ti­cally mean­ing­ful demon­stra­tion that they are sig­nif­i­cantly safer than av­er­age driv­ers.”

Clock­wise from top, An­ders Lie, Jim Sayer, Brian Sou­blet and Justin Kintz.

Driver­less elec­tric shut­tle-buses tak­ing to the streets of Paris in July.

Newspapers in English

Newspapers from New Zealand

© PressReader. All rights reserved.