Look, Ma, no hands Melissa Ce­fkin on self-driv­ing cars


You have just ar­rived at a four-way stop, and you no­tice that the car to your left has no driver. Sud­denly, it ac­cel­er­ates through the in­ter­sec­tion and af­ter a jolt, you re­al­ize that, yes, it was that car’s turn. As it turns out, the ve­hi­cle “knew” ex­actly when it should go. The chal­lenge of the four-way stop is an ex­am­ple cited by Melissa Ce­fkin, who speaks about driver­less or “au­tonomous” cars for the School for Ad­vanced Re­search on Thurs­day, Sept. 14. “I come to a stop, and I’m look­ing to the left to see if the car that got there ahead of me is start­ing into the in­ter­sec­tion,” she said. “At that point, with­out even think­ing about it, I might let my foot off the brake, and that slow rolling of the car would be a sig­nal to peo­ple that my car is about to go. An au­tonomous ve­hi­cle would have to de­cide or learn to take the ac­tion, or be told to re­move the brake.” That sit­u­a­tion is one of many that an­thro­pol­o­gist Se­fkin and Nis­san Mo­tor Cor­po­ra­tion en­gi­neers are try­ing to an­tic­i­pate in de­vel­op­ing soft­ware for such ve­hi­cles. These cars will ac­tu­ally ar­rive at the in­ter­sec­tion with an ad­van­tage — they will be equipped with sys­tems that will have cal­cu­lated, to a very pre­cise de­gree, when ev­ery­one ar­rived at the stop.

But that doesn’t mean the car should just plow through, re­gard­less of what the other driv­ers are do­ing. The au­tonomous ve­hi­cle needs to know when to go and turn and stop and not run into any­thing, or over any­thing — al­to­gether, a fairly in­tim­i­dat­ing propo­si­tion. “It’s pretty awe-in­spir­ing. There are the en­gi­neers who are very good at un­der­stand­ing and work­ing with things like road­ways, lines on the roads, and the vis­ual images of reg­u­lar kinds of things on the street. But the group I lead re­ally fo­cuses on the as­pects of mov­ing around on the road that have a so­cial di­men­sion to them, and they’re very sub­tle.”

She gave another ex­am­ple, one from a so­ci­ol­o­gist col­league. “If you’re driv­ing along and there are cars parked on the right, and you see some­body on the side­walk step­ping out toward the street, an au­tonomous ve­hi­cle would see a mov­ing ob­ject en­ter­ing the road and pos­si­bly cross­ing its path. And that’s good be­cause then maybe it wants to slow down or be care­ful. But if you saw some­one stand­ing and wait­ing on the other side of the car, you’d think, no, the per­son step­ping be­tween cars is go­ing around to the driver’s side, which in the United States is on the left side of the car, so they’re prob­a­bly just get­ting in their car.”

A hu­man would make in­stant, al­most un­con­scious judg­ments about that sit­u­a­tion and, while ex­er­cis­ing cau­tion, would pro­ceed with no need for dras­tic sw­erv­ing or brak­ing. Can ma­chines han­dle such cir­cum­stances? That’s one of the tasks fac­ing the Nis­san Re­search Cen­ter in Sun­ny­vale, Cal­i­for­nia, where Ce­fkin leads the hu­man-cen­tered sys­tems prac­tice. A Ful­bright award grantee, she has a Ph.D. in cul­tural an­thro­pol­ogy from Rice Uni­ver­sity. Nis­san hired her a lit­tle over two years ago. What an idea: a car man­u­fac­turer em­ploy­ing an an­thro­pol­o­gist. “The direc­tor of the lab here, whose area of ex­per­tise is ar­ti­fi­cial in­tel­li­gence, works side by side with so­cial sci­en­tists and strongly em­braced the idea that au­tonomous ve­hi­cles need to re­ally be de­vel­oped through the lens of a very user-cen­tered, so­cially-cen­tered point of view,” she ex­plained.

“The main thing that goes on here is the de­vel­op­ment of the soft­ware for driv­ing the ve­hi­cle: how it will move, how it will op­er­ate, how it will make de­ci­sions. The au­tonomous-ve­hi­cle soft­ware is mak­ing de­ci­sions: It knows where it is, it knows where it should be, in this lane or that lane, it knows what it sees around it, and it then can make de­ci­sions and can com­mand the car.”

The tim­ing of Ce­fkin’s talk, “Walk, Don’t Walk: Ev­ery­day In­ter­ac­tions With Self-Driv­ing Cars,” is

in­ter­est­ing. Domino’s Pizza planned to be­gin test­ing pizza de­liv­er­ies in Ann Ar­bor, Michigan, in the last week of Au­gust, us­ing a driver­less Ford Fu­sion sedan be­decked with sen­sors, cam­eras, radar, and li­dar. Ac­cord­ing to an Aug. 29 story in The New York Times, spot­ting a driver­less ve­hi­cle is not un­com­mon in that city, where the Uni­ver­sity of Michigan has a pi­lot project in “con­nected-car tech­nolo­gies.” The Domino’s car will have hu­mans aboard to mon­i­tor not only the au­to­mo­bile but pizza cus­tomers’ re­ac­tions. It’s an ex­ploratory foray into the realm of prod­uct de­liv­ery; even­tu­ally a re­tailer like Home De­pot might use an au­tonomous truck to trans­port building ma­te­ri­als directly to a job site.

Even af­ter the de­vel­op­ment and test­ing phases, self-driv­ing cars us­ing Nis­san’s Seam­less Au­tonomous Mo­bil­ity (SAM) tech­nol­ogy may op­er­ate with some kind of re­mote mon­i­tor­ing, and per­haps emer­gency con­trol. “There are likely to be times when the ve­hi­cle would need help or would be able to do more if it had a re­mote mon­i­tor,” Ce­fkin said. “We worked with NASA, us­ing the same tech­nol­ogy as what’s used with in­ter­plan­e­tary ro­bots when they have to as­sist a robot that’s some­where else to get over a boul­der or what­ever. The idea is an as­sisted tool. It’s not what we con­sider a dras­tic op­tion, be­cause no­body else ever fully con­trols the ve­hi­cle. It’s al­ways op­er­at­ing au­tonomously, but what the SAM oper­a­tor can do is pro­vide it with more in­for­ma­tion and give it some new in­struc­tion to say, ‘This is what we think you should do in this case.’ ”

In the ex­am­ple of a pizza-de­liv­ery fleet, mon­i­tor­ing could func­tion sim­i­larly to air-traf­fic con­trol keep­ing track of air­planes. The ve­hi­cle soft­ware would in­cor­po­rate real-time learn­ing. “All the ve­hi­cles would be send­ing in­for­ma­tion to the Cloud and the SAM ca­pa­bil­ity would get in­for­ma­tion from the Cloud, up­dated on an on­go­ing ba­sis,” she said.

Ques­tions of own­er­ship is­sues arise — for ex­am­ple, whether au­tonomous cars will be shared in­stead of hav­ing one in your garage. “As an an­thro­pol­o­gist, that’s the kind of ques­tion that re­ally drew me in. One thing I re­ally want to fo­cus on at SAR is the way in which, as peo­ple move about, we move about through iden­ti­fi­ca­tion of no­tions of place. One of the things I’ll be ad­dress­ing is place-mak­ing and ex­pe­ri­ences and the creation of mean­ing, how our ex­pe­ri­ence of place is very ex­pe­ri­en­tially sig­nif­i­cant to our hu­man­ity and our way of be­ing in the world. So I’m go­ing to ex­plore how au­tonomous ve­hi­cles might or might not sort of play into that, and dis­rupt or par­tic­i­pate in the mean­ing of place or places.”

How long will it be be­fore there are driver­less ve­hi­cles all around us? “Not right around the cor­ner,” Ce­fkin said, but she added that in some spec­i­fied, de­fined ar­eas and un­der cer­tain con­di­tions, it will be sooner. “One ex­am­ple would be what is re­ferred to as the ‘last mile’ is­sue. Stud­ies have shown that one of the prob­lems for mass transit use is not that peo­ple don’t want to use it, but that they first have to get to the point of de­par­ture, or get back home from the point of de­par­ture, from the bus stop, for ex­am­ple. If you had a more flex­i­ble, vi­able means to trans­port peo­ple to the points of de­par­ture, that could be a great ap­pli­ca­tion.”


Melissa Ce­fkin

Newspapers in English

Newspapers from USA

© PressReader. All rights reserved.