Faculty Forum Online, Alumni Edition: Autonomous Vehicles

Faculty Forum Online, Alumni Edition: Autonomous Vehicles

Show Video

Hi, I'm Whitney. The MIT Alumni Association and, I hope you enjoy this digital production. For alumni and friends like you. Hi, good. Morning welcome. To the MIT, faculty, forum. I'm Aviva, Hoch Rudkin and I'll, be today's moderator. I'm, a master's alum from, 2013, in science writing and I'm. Currently the math and data editor at the conversation, us the. Conversation, is an independent, source for informed analysis, that's written by the academic, community and edited. By journalists. So. Thank. You so much for joining today as a reminder we, welcome, all your questions during this chat if, you're one of our alumni who's joining us via zoom you, can use the Q&A feature, found, on your toolbar, and. For those viewing on YouTube you can add your questions to the comments field right next to the stream and. We also encourage, you to tweet using, hashtag. MIT, better, world and we'll, try to get to as many questions as, we can today. Today. We'll, be discussing several. Areas of research around autonomous. Vehicles, and advances, made in their technology, deployment. In cities and safety our. Guest is Don McKenzie who, earned his master's, and PhD degrees, in MIT, in tech policy and engineering, systems, Don. Is an assistant, professor of civil and environmental engineering at, the University, of Washington, he, leads a sustainable, transportation lab, which develops, and evaluates, technical, and policy solutions, for making our transportation, system more economically, viable and, environmentally. Benign, Don. Is a member of the Transportation Research, Board Standing. Committee on transportation. Energy and chairs Subcommittee. On energy and demand implications, the connected, and autonomous vehicles. Automated. Vehicles done. Thank, you for joining us we're. Looking forward to hearing some of your latest insights into autonomous. Vehicle deployment, and practices, here in the US. My. Pleasure. All. Right so, Joe. And Aviva I trust, that you will let me know if you. Have any trouble if you cannot see my, slides. But I'm going to jump right in so it's a real pleasure. To be, here today and and. To be talking, to you all and. Appreciate. The, the. Registration. Numbers and I think it speaks to what an exciting, and and an interesting. And new topic, area this. Is so I, appreciate. The generous introduction Aviva I will, just. Say briefly that, I lead, a group called the sustainable transportation lab, here at the U dub in Seattle, we, have a great set of students from around the country and around the world working. On these. Questions. In the areas of automated, vehicles electrification, and, new. Mobility services. And and. Just I'll say a little bit more by way of introduction, we, look, at sustainable, transportation. We. Think that is not just an environmental issue, but, as. A system, that include that sort of supports, a dynamic, vibrant economy, and provides, equitable, access to, opportunities. Goods and services, for all and we view these outcomes, as emerging, not just, from new technologies and. Old technologies, but from how that technology interacts. With, individual, choices and behavior, as mediated. By public, policies, so that really guides our work on both abs and on. These, other areas where we work electrification, and, mobility services. Anyway. I thought I'd start by saying a little bit offering. My perspective. On when. Automation. Is coming and that's specified here, when it's safe automation. Coming because really. I think safety, is the number, one. Sort. Of bottleneck, right now on getting automated. Vehicles, onto the, road the.

Reason I say that is that you know you can put vehicles, out there today, that can drive themselves the, question is can they do it safely. Now. When is automation, coming well in one sense it's already, here if we talk about partial, automation. What SAE, would, call level 2 automation. Where you have you, know several. Features. Working together in a coordinated fashion level, 1 or level 2 automation this is already here. So. We have one or a few features, working together to, support the drive with user driver assistance, systems these are already here and they're already in mainstream parts, today. But. When we look at full. Automation, getting, towards what SAE calls level, 4 which is a car, that, can drive itself, all. The time without a human backup, driver and. Level, 5 which is a car that can do that in any conceivable, driving. Situation. It. Gets a little more challenging I think this was underscored, by the. The, accident, this spring in Tempe, Arizona that. Ended in the unfortunate, death. Of a, pedestrian and, so. I think one way to kind, of look at this and try and interpret this is you. Know we, have here a fatal, crash and, human, drivers, in the u.s. despite, the fact that we. Have you know many many 30,000. Plus highway, deaths a year in the u.s. that. Really, amounts to about one fatal crash for every 93, million miles, of driving, at. The time that. This crash in Arizona, happened, uber. Had accumulated about 3, million autonomous. Miles in their vehicles, in their automated vehicle, fleets and so, if you if you want to kind, of take, one perspective on this it would be that if. Boobers. Automated. Vehicles, really were as, safe. As human. Drivers, there's, only a three percent chance that. We. Would have seen a crash occur, in the first three million miles. Another. Way to look at this you might say hey you're being unfair and singling, out uber there's been plenty of other companies. That have been. That. Have been accumulating. Autonomous, miles at the same time and we'd be having this same conversation regardless. Of, which. Company was involved in this accident so we could say that there, was a probably about 9 to 10 million autonomous. Miles that have been accumulated, at. That same point across, the whole industry. And, that. The probe, he still is less than 10% if, those cars were as safe as human drivers are so less than a 10 percent probability that we'd see the first fatal crash in that first nine or ten million miles so, it. Seems, this that rates and the challenge, there's there's some some work by r and that has dug into this it's. Really hard to know for sure if, these cars are safer, same, safety, less. Safe than human drivers, until, we have hundreds, of millions or hundreds of billions, of miles, long so it's a little hard to know but this does kind of point in the direction that these vehicles are are not, quite ready for primetime. We. Can also take the perspective of. Looking at so-called disengagement, so, in the on-road, testing, in California. Companies. That are testing, their automated vehicles have. To. Report. So-called, dis engagements, these are cases where the human, backup. Driver has, to take over for the system, and. So if you look at the rate of safety, dis engagements. By way mo which is pretty much the undisputed market, leader this is formerly, Google, right. You. Can see that their, progress. In their their disengagement, rate kind, of leveled off in in, 2017. After making considerable progress in, the past couple of years now we, can't really say that that's a trend and there are many reasons, that this, could have happened right, this is not necessarily, that. It's that they're reaching a point where they can't make further improvements it could be that they're getting to the point where, they're putting those cars into more and more demanding, and challenging situations. But. The fact remains, that about. This is about two, orders, of magnitude above. The. Human driver crash, rate so. We can see that here - whoops. We. Can see that here as well that about. That. There's in 2017. Wham-o. When about 5,600. Miles between, safety-related, dis, engagements. Previous. Analysis. Looking back to 2015, suggests that about one in five of those same disengagement, would, have resulted in a, crash so. We can estimate that way, Moe would have had a crash about once every twenty eight thousand, miles in 2017. Is that exactly, right, no but, it's order, of magnitude, probably. About right and you can see that that is in contrast. To one, human. Driven crash, every four hundred ninety, thousand, miles so we're a long ways away from. This technology, matching. Up to human, drivers. Now. Once. This once, the technology is ready for market introduction, it's going to take some time before, it completely, takes over the new car market so, this is work by, Steven zoth who, is, another, MIT, alum, actually three degrees from MIT, and.

Now Directs the Center for Automotive Research at, Stanford, but when he was doing his grad work at MIT he. Did this work with John, Heywood and, what. They looked at was basically, the amount. Of time, that elapsed. Between. When a new automotive, feature was introduced, to the market and when, it hit the, maximum. Rate of growth in in, its growth of market share and what they found is that this great this delay this lag has been falling, over time but, it's still on the order of five to ten years so you're looking at five to ten years between initial, introduction, and what a feature really, takes off and starts and hits its maximum rate of growth. Moreover. That, maximum, rate of growth itself, the maximum. The steepest, part of the adoption, curve, is usually. No more than 10% per, year this. Is from the same work bides Dauphin Haywood and. So when you put those together you can develop a, plausible. Adoption. Curve from initial, market introduction, to. When, that technology is pretty much available on all new vehicles and it's, it's more or less something that plays out over a couple of decades to get to basically all new cars. Moreover. We, have an, installed, base a fleet, of vehicles already, in circulation. 60%, of cars last. More than teen years the average car in the road today is 11 years old and so, there's this additional, about, 15, or 20 year turnover, period, to, get rid of the remaining manually, driven cars and so the bottom line is that we are talking, 40. Or 50 years probably, before, all vehicles. On the road are automated. Now. Much. Of my work is. Focused on the energy and environmental, implications of. Abs. And so, I'd like to say a little bit about that and the. Key point here I think is that energy. And environment, are not the. Driving forces. Behind why. We're looking at automation, in vehicles, vehicle, automation, is about. Safety, it's about, providing, access to, underserved. Travelers. It's about roadway, capacity, it's about making, more productive, use of time in, the vehicle, but, energy and environment are not really, why people, are doing this for the most part why companies are pursuing this that, said, they, do have. Potentially. Significant. Very significant, energy and environmental, benefits for for. Automation. May have very significant. Energy and environmental benefits and if, we think of vehicle. Emissions, their total emissions, as being, a product, of this identity so, people. Miles of travel how many miles people, travel multiplied. By the ratio of vehicle, miles per person model multiplied. By the energy consumption, of our vehicles, multiplied, by the emissions, intensity, of our fuel. We. Can imagine. Plausibly. That. Automation. Could, have. Beneficial. Effects on almost all of these factors. And. I've listed them here I'm not going to to. Read. Through all these but, what I will say is that people, have been talking about this for a while and it's probably five years ago, now that people really started paying attention to the vehicle automation, and a, lot of folks at that time we're, presenting, ABS, as this panacea, for energy, use and emissions from vehicles and, you.

Know Identifying, many. These reasons, it's kind of the explanation, for that so, what. We wanted to do in, our work on this topic was to try and put some quantitative, bounds. On how big these, impacts, might be and I don't know that we got all of them but we know a lot of them and we also anticipated. Some potential, negative. Impacts, so ways that. Ways. That automation might actually increase. Energy use and we've listed some of those here, in, red. Anyway. Um to. Kind of skip, to the bottom line, the, biggest. Risk, that automation. Of vehicles, poses, from an energy and emission standpoint, is. Inducing. More demand for travel and the reason, it does that is that, automation. Makes vehicle, travel safer, cheaper, and more convenient. I think. This testimonial. From a Tesla, driver talking, about his autopilot that's a level 2 automated. Automation, system really. Sums this up nicely right it was the same distance, but the commute felt like it took half the time so what keeps saying here is that, there's a reduced burden. There's a reduced implicit, cost, to his time spent, driving and. When. We look at the generalized, cost of travel, the single, biggest component. Of that cost on average is the, value of the drivers, time so. We're. Talking about taking, the single biggest component. Of cost of operating, a vehicle and reducing. It by making that travel, safer cheaper and more convenient, and, I have. Thought about this and I can't think of any good or service that has ever become safer, cheaper and more convenient, and seeing, people consume, less of it so I think this really points in a direction of more travel, in the future so. By our estimates. We, figured somewhere, in the range of five to sixty, percent increase, in, vehicle, travel, could result from this, reduced burden, of driving. Now the challenges, we. Don't really, know and there's very little empirical, evidence, on. Which to judge how. Much that perceived. Cost of time will, decline. I. Will. Say there have been some other research, the Atlanta, Regional Commission the Puget Sound Regional, Council, run sort, of regional, travel, simulations. To assess the impacts of automation, they found something like a twenty to twenty five percent increase, in, vehicle miles traveled. That's accounting, for all of these feedbacks, between, roadway. Capacity, additional. Travel that. Additional, travel leading, to more congestion. So. They're capturing sort of the full feedbacks, but. The bottom line is they get about a twenty to twenty five percent increase, I. Think. By far the. Most innovative, creative. And gutsy. Investigation. Of this is as. This work here, that was done by John Walker at Berkeley and Pat. Victorian, at Georgia Tech I really, wish that I, had had this idea because, it's awesome they said what. Is the, travel, mode today that. Most resembles. A personally. On the automated, vehicle, and is it well it's pretty much like having a chauffeur, and so, they said how would we assess how people might respond, to having that well, we'll, give people a chauffeur so they ran an experiment, where they gave people a chauffeur, for a week to drive their existing, vehicle, and what they found was that during that week their, travel, increased, by eighty percent, their vehicle, miles of travel increased by eighty percent, when they have that chauffeur, now. I. Can. Say more about this but I I don't want to get bogged down it's I think it's just a really cool, interesting. Way, of approaching the problem now. So. That's the risk right the risk is more vehicle, travel, the, major, opportunities. To. Cut energy and emissions, really. Hinge on automation. Acting, as an enabler of mobility. Services, so the idea, is that if automation. Can lead more people to use mobility services. Than, mobility, services, unlock. All of these other. Mechanisms. That can help to reduce energy, use. And emissions, the. Challenges, that these relationships, are all uncertain. We have very little empirical data to back, up just. How strong these different relationships, are. But. I will talk about a couple these interns so the first is paper. Trick and so the thinking, goes like this this is work out of Switzerland, where. They've looked at the total. Cost of travel, in Swiss francs per kilometer, so, the thinking goes like this if you've got a private, car, it is. More expensive, than, using, mobility services.

So The right three bars here are mobility. Services. And so, the thinking is that because a mobility, service, can be cheaper, here 0.39. France. Versus, 0.51. France per kilometer, that, people will opt to use the service, but. Personally. Owned vehicles, are mostly. Fixed, costs, and only, a little bit of variable, cost so, the thinking is people switch to using the service because it's cheaper overall but. Once, they do they will travel west because, the variable, cost is higher, than when you own your own car, that's, the theory I'm unskipable. About, it because, what, we already see with mobility services, today is the emergence, of subscription-based. Business. Models, and what that does for, instance lyft offers, a plan where you pay a monthly fee and you get a certain number of rides and what that means is, that at the margin, you're not really, paying for, each of those rides we, see this with, people's use of phone plans so most people have, a plan that gives in certain number of minutes certain number of gigabytes, of data every month on their phone so, that they're not paying by the megabyte, paying by the minute, to use their phone because people, don't like that people don't like feeling, like they're paying at the margin, and paying by the minute and getting. Squeezed every second that they're using a service and, we see that the subscription, model is. Proving, to be popular. So. Paper, trip may, be challenging, because because. Companies. Will innovate and off of these subscription, services, that undercut, that that, paper trip, incentive, to reduce travel, so. Ride-sharing maybe, we can enable ride-sharing getting. People, to, share, rides with one another, because. We. Have automated. Dispatch, automated. Matching it, can be done very fast and efficiently well, the challenge, there is that. Automation. Makes. Mobility. Services, cheaper so today. If. You were to look at a manually, driven vehicle this is Switzerland again, the cost per kilometer, of a private, taxi is about, 2.6. Franck's for a kilometre versus, one point for France per kilometer for. Shared. Taxi, so your savings is 0.8, per kilometer, that's that's the bars in blue when. You automate, those, costs, gets so small that, the gap between them, also shrinks, so now instead of saving point 8 you're saving, point 1 to France per kilometer, under, an automated system that's because you've taken away the cost of the driver and you've made this so cheap so the idea here is if it's so cheap to get a personalised, vehicle, who's, gonna share a ride with. A stranger. Without even, as a driver there as a third party, who's going to get in a car with a stranger just to save point one to France per kilometer, or in the u.s. context, you might be saving 25, cents, a mile. So. I'm a little skeptical about that and this is work by David. Keith, who's, a faculty, at MIT Sloan. And, what. He's what that is used he's a stated, choice experiment. To, see how this, might. Evolve and so he gave people a hypothetical choice, between a personalized, ride and a shared ride and, what. He found was that if we assume that over time that. Price differential. Falls then. We see the market, shift much more to, the private. Vehicle. Rather, than the pooled ride so, again we're, a little a little skeptical, about this now, it's. Not all lost I think the big opportunity. Is in, right sizing, so what we mean by right sizing, is. Matching. The capabilities. Of the vehicle, to the needs of a specific, trip so today most. People who buy a vehicle buy the vehicle, that's going to meet not just their average, needs but their peak needs you want a car that works for you all the time so what that can mean in an extreme, case is you, have an SUV or a minivan that, can hold seven or eight people and you use that when you're driving along the, idea of right sizing, is simple if you're not tethered to that specific.

Vehicle, A company, can dispatch, a different vehicle to you based on your needs then, they can send a small car when you're traveling alone or with one other person they can send bigger cards when that's needed that's the idea of right sizing, we, see this in emerging, of course in mobility services, today all right with uber ex uber XL lift and lift plus we see it in the car to go car share fleet in the reach now car. Share fleet people. Want to have that choice to choose, the. Vehicle that meets their needs not, pay more than they need to when. They're just traveling alone and a new example, of this that I don't have on the slides of course is the. Booming. Popularity. Of scooter. Sharing, and bike sharing and electric bike sharing in markets, where those services are being introduced, so what you see there's people, not, everyone but many people are happy to take those sort of ultralight, vehicle, modes. When, they're traveling alone. So. There's a big question though about how much people will actually embrace, this and, their willingness to accept this this is an open question but, we. Figured that the, potential the. Potential, is for. About a, up. To a 50% reduction in energy. Use per mile. If. You were to if you were just sort of right. Sides very aggressively. The. Final, area where. Automation. May. Effect these. Emissions, and environmental, impacts energy and environmental impacts is through changes in land use and residential, location, choices so this is some new work that we've done with support from Toyota, looking. At different. Models for, the introduction, of automated, vehicles, so, on the left here, we have a scenario where private. Vehicles, become, automated, on the, right we, have a, scenario. Where we introduce. A new shared, automated. Service, and what, we've mapped here is the change in accessibility. What we mean by accessibility. Is it's a quantitative, measure, of how. Easy, it is to, get from, where you are to, where you want, to me and in this case we focus specifically, on commute, accessibility. So how easy it is for people to get from home to, their workplace, and then, what you see here, is when we automate. Private. Vehicles, then, we see the biggest increases. In. In. Accessibility. Following. In these kind of fringe areas, this is the Puget Sound region around Seattle, so Seattle, here's sort. Seattle, and Bellevue are the major, sort. Of urban centers along, with Tacoma, down here in the south sound and then what you see are these fringe areas, out around that the air at the edge those. Are the areas where private, automated, vehicles provide. The biggest boost to accessibility. In contrast. When. We introduce, a shared mobility service, it is, much, more, even. Throughout the region, although, less so in these very fringe, areas, so it's kind of the reverse and, then, we can translate, that into changes, in people's choice and preference about where to live and, what you see is that, again. When we introduce private, automated. Vehicles we see more increase, for demand, for. Residences. Out in these outlying, areas, and when, we. Introduce. A shared mobility service the, bigger increases, in demand tend to be in this more existing. Urbanized, area, in the center. And. So. I think this is my last slide, and. The point that I would just leave you with here is that some, of the impacts, that we've talked about some of the impacts of automation, they. May depend, on the. Level of automation and, so we can reasonably, expect in, particular, that, many, of these travel, costs, reduction. Impacts. This, reduced, value, of travel time conducing, more demand for travel could, very well occur, with.

A Level 3 automation. That is basically. What outtie has introduced, today, Tesla, is getting close to level 3 this is a system, that. Can, allow. The driver to disengage, and will alert the driver when. The driver needs to resume, control and so with a level 3 system, that works under a, significant. Range of conditions you, can get much, of this induced, demand effect, but. That. Is not sufficient, to run a mobility, service, you really need level 4 or level 5, automation. In. Order to support. An on-demand. Will automated, mobility, service because that vehicle, needs to be able to deadhead, it needs to be able to drop one passenger, and go to the next without a backup, driver in the car and that means level 4 or 5 automation. So, a big, risk is if we get stuck, here at level 3 if, we do we can see many of these downsides, and not yet. The the benefits, that come from unlocking. Mobility, services, so that's, that's, pretty much what I had and I'm happy to take, questions about, this or you can throw some curveballs, at me, thank. You all, ok. Thanks, Don, that. Was great. We've, got some. Great questions coming in already I should, remind people before we jump in that if, you have questions you want to ask you. Can put. Them in the Q&A, feature on zoom or you can add them in that comments, panel on YouTube live I. Have a few questions of my own but I actually want to start with this question from Tom that I thought was interesting, Tom. Wants to know do. You believe that an autonomous. Vehicle should. Always have the capability, to be driven manually, if the driver chooses, and. Does, the industry say. The same thing do they agree with you - I. Think. You're on mute. Good. Catch ok so, I don't know that I have a position on that kind, of normatively. In the sense of like should, they is that good public, policy I. Tend. To think that, for. The for. The near to medium term that. Will probably, be a business, necessity. For. A couple of reasons one. People. Many. People don't. Want to, to. Have. An automated vehicle, drive them all the time or, any, of the time and so, if you want a vehicle that, can appeal to, everyone and that's generally, the the, model in the in in, trance can the. Auto sector is that you try and make, cars that appeal to as many people as possible. That's. Probably a necessity, for that reason, I think it's also probably a necessity because if you don't have those, human. Controls, that you really need a level-five vehicle. In order, for that to in. Order for that to work so I tend to think that we will probably certainly. For the personal. Vehicle market continue. To see. Continue. To see manual. Controls, in those vehicles for a long time to come, I also. Think, that. One. Of the possibilities, that I haven't heard a lot of people talk about, is. With these automated. On-demand mobility, services, a lot of people tend to think that that will work like, uber. Essentially. So the car will come, it will drive itself to you you will get in and it will drive you to your destination, again, I think for a lot of people that's an appealing option but, I think you could appeal to a lot more people, if you, had a car that could drive itself to you but, then you can, drive it to the destination if you so choose and you kind of get the best of both worlds in that way if, someone wants, to let the car drive they can but people who prefer to drive themselves can, also do that, ok. I think, that speaks to something to that I wondered about talking to people like. You talked about trucking for example, which I. Think some people really excited about, does. It make sense to say when, we think about here's the potential, costs, here's potential benefits, maybe we should, restrict. Autonomous. Vehicles to certain kinds of domains, you. Know someone asked you know are these, insights, the same if we focus on moving. Goods rather than moving people around so. I. Think one of the challenges with. With trucking of course is that. Drivers. Are very well organized, so the trade associations. For drivers, are very well organized, and you're. We're, certainly, can. Anticipate a conflict. Between sort. Of shippers and sort, of major fleet. Operators. And independent. Operators.

And Drivers, whose. Livelihood depends. On them driving the trucks, you. Can certainly think of many. Sort. Of special, cases, in the, near term sort, of truck maintenance issues, chaining, up in the mountains, these sorts of things that it. Would be hard to get a robot to do in the near term but you can also easily. Envision, that you don't need a driver in every truck to do that that. You could come up with other solutions there's, certainly a large, financial. Incentive, to to. Automate, trucking. Okay. Let. Me ask another, question since, this is MIT focus, we have a question from someone in Boston, who. Notes that you, know in Boston, there haven't been that many traditional. Public hearings about tests. On deploying ABS, are. There other cities, where this is really contentious, where the public's really involved you know what kind of places are you watching, when you think about all. These issues you. Know that's interesting, I haven't, heard. Much about a. Lot, of public, input on testing. And using, cities as test beds. That's. Not to say it hasn't happened a. Lot, of that kind of. Certification. And, and. Regulation. Of these that's not the main area that I focus on so it's entirely possible that I've missed it but, I have not heard. About a. Lot of pushback on, this okay. What. If I have interested. One thing I've been looking, for and. We may collect some data on this but I haven't really seen good data yet is, I suspect that people feel differently about. Sharing. The roads with automated vehicles, versus. Actually. Being driven in one themselves. And. I think that would be an interesting question that. We. Can and should be looking, into or the, any gaps in perception, they're in. Perception, of their. Perception of safety in the context, of other, vehicles, on the road versus. You, know driving. You I mean in general, we know that people, you. Know people, okay. It's a little bit of an exaggeration but people, we think that they're great drivers, and everyone else. Sorry. Sending that logic, sure it would be a good idea to replace all the other drivers with robots, but not me. And so. I. Tend. To think we would probably see, some differences, there and, I think you're liable to get a lot less push back on the idea of having a B's on the road. Versus. You. Know something, which by the way no one is talking about basically, requiring everyone to travel in a navy I don't. People. Would not like that I would not recommend that I don't think anyone would try and advance that policy, at least not in the near term okay. So. There's another related. Question maybe this is also a little out of your ear usual zone but David, Wang wanted to know if there were also interesting.

Differences Either in the impact, or. In the adoption of automated. Vehicles between different countries you know particularly. You know we think about markets, that aren't is motorised as here in the United States or there develop places. So. Have we looked, at, no. So the work that we've done looking, at energy and environmental impacts has been primarily it's. Been in the u.s. context. I suspect. That that impact, would be similar. In other you. Know highly motorized, developed, countries, and, I and. I would say I think that the impact would be similar. Any. Pretty. It would be okay similar, enough anywhere. As long as you're counterfactual. Is automated. Vehicles, versus, widespread. Privately, owned manually, driven vehicles, now, that, counterfactual. May not hold right, so in some countries, it might, be that, introducing. Automated, mobility, services, into a transit. A region that's you, know that uses, a lot of transit. Has. A, bigger, energy impact, because it basically moves people into. Into. Cars rather, than buses, or trains. So, directionally. That would be the impact right so it kind of depends on what's the what's, the comparison you're making against but if you're comparing against an area where. There is less, motorisation. Or, less. You know personal, automobile, 'ti then, you would see bigger. Bigger. Adverse, consequences. From an energy and environmental standpoint, okay. I, want. To ask a question we've gotten a few people asking, about this including, them I think the first person to ask was Jordan, Luke class of 76. Autonomous. Vehicle, you know could it go goat could. It go from place to place and be, used for nefarious purposes kind. Of like a car. Bomb or a terrorist tool, is. That something that you're worried about what do you think about safety. So. No I know laughs, because. It's not because it's a funny question because yes but, like I don't have a solution to it what. I know is, there, are, definitely. Folks, and there is serious, money being invested in, some, of the kind of cybersecurity. Topics. That. Come up with abs and this could be I mean you can imagine all kinds of, nefarious deeds right sort of you. Know you could use it as a you, know a bomb, delivery, vehicle, you, could use it to mow. Down pedestrians, with, these you, know sort of vehicle assaults. That you've seen in you, know many countries, right you, could also imagine using, it for you. Know for as basically, to attack or kidnap, the. The person who's travelling in the vehicle so there's all kinds of. Bad. Things that could, happen and I. What. I can't say that oh that's totally in hand or oh. That's a, issue. What, I what I can say is that there is money being spent, by. Various. Agencies to, you. Know to research, this and try and come up with better, better. Sort of cybersecurity and, ways. To mitigate those risks but. I don't I don't know if that's a problem that we can solve and I don't know how big a problem that would really be. Yeah. I mean I think it speaks to to that as humans, anything, new maybe, we're a little more skeptical, of it we want it to be you know thinking. About the possible bad situations, I mean there's a question here from Paul Clarke who.

Says You know do you society will require lower lower, accident, rates for autonomous vehicles. You. Know because they're perceived as machines for, which the risks are more controllable, for as humans which no, one can control I guess you know are. We gonna expect a lot more and is that going to affect how these are adopted, so. I think again. My, sense is that it the public probably, does expect, more and the challenge we face is that the public expects, more and today we, have companies, who are working really hard on this and they're nowhere near as good as human, drivers. So. That's, the kind of tension that we have and the economic. Incentive, for. Companies is to get these things out as, soon as they can and, that may not align. With what the public is willing to accept, on. The, roads. I was, a terribie, a couple of years ago so this is a poorly sourced, anecdote, but, I was a terribie, a couple of years ago and some guys stood, up in a session and said that he had worked at, FAA back, when they were introducing. Automate. Automate anding. Systems, for a passenger aircraft, and. What he said was that the. This or the rule of thumb that they used was that it had to be a 10x, improvement, over. Human pilots. To. Be acceptable, and, because, accidents, are still going to happen incidents, are still going to happen but, they had better happen, at just 1/10 the rate, of. What. Human pilots, experience. And. So I don't, know I suspect. That it's we're looking at something similar, to that that, that's the level we'll need to get to for people to really embrace this as an alloy good from a safety standpoint, and. Certainly, we know that people perceive, risks. Differently, when, it's a machine versus. A human, and. I. I. Can, I can speculate, on this a bit but I think in part it's that you, know it's very easy for humans, to go back and reconstruct, a story and a narrative about. What, happened, in those seconds, and it was just you know and even if it comes out and there was just not enough time to react we say ok but if it's a pewter we expect, perfection, right we expect flawless, performance, and so. We'll, be very disappointed, if we if, we don't get that and again part, of the challenges, it's hard to know really. How safe these things are until. We have you. Know hundreds of millions or hundreds of billions of miles traveled. And. That's more than you can. Cost-effectively. Accumulate. In just, test leads you really need, something you need some, kind of staged commercial, introduction to gather that data and, that's what this this Rand report that I would highly recommend that's. Sort of the take away from it I. Think. You know going back to this idea with with the commercial, air travel system I think one, of the potential, benefits that we have from.

Handing. More control over to the computers, and taking this away from individual. You know from hundreds of millions of individual, drivers, is that, we can really have a much more safety, first oriented. System, the way we have with. The FAA, NTSB. Today. Working. You know hand-in-hand, with, the industry. Very collaboratively. To. Identify root. Causes and every, incident is treated, as a case that has to be investigated, and closed that's, not really, something that we have today with. With. Automobiles. But it's something we couldn't get to but a big challenge to that is a lot of the companies that are developing this technology. Are. Reluctant. To share data. And proprietary. Algorithms. With, regulators, or with the public, for. You, know for because it's their trade secrets and so there are sort. Of incentives. That strongly, push them towards, less. Transparency. And I don't know, and if we don't, figure out a way to kind of to buck, that trend I don't know that we can get the benefits, but we can realize the same kind of benefits that have been realized in air transport, system, mm-hmm. Okay. Well I think that leads to chuse we've gotten a few people asking about how. The insurance industry. Is, adapting, to this and what on earth that's gonna look like. And, you. Know as we move forward we eventually have more people in these vehicles. Getting. Into all kinds of scrapes. Yeah. So I think that, I think, I, think, the the, first, thing that a lot of people think is that the insurance industry won't, like this because, they make their money selling insurance, policies, and if, you reduce the need, for insurance, you reduce the frequency and/or, severity, of crashes. You. Reduce the need for insurance. But. That's. Not my, experience of, how this is going so the insurance wasn't a big meeting. For this that is the, automated vehicles impose a tree July. And. I've. Been going to that, you. Know and helping to plan that every year since, they started it and. We see in sure the insurance industry is there and they are highly engaged and, I think that in general they. Want to see, improvements. In technology, that that save, lives and, partly. I think you can understand, that for competitive, reasons it may well be that this is something that ultimately, shrinks, their, their market, but, as individual. Companies, in competition. With other companies. It's. Who's, them, they. Can make more money in, a, competitive market, by you, know getting this technology, into the hands of their customers, and appropriately, pricing, it right so they can gain market share by.

Selling, Policies, to people who are using technology that, improves safety so that's so. So it's it's. Something that may ultimately shrink, that market, but the companies are pursuing. It even though it's kind of self cannibalizing. Yeah. I mean the data you show to at the end where you look how, people, are gonna make these decisions on what, kind of vehicles to ride in was really interesting, you. Know Howard asked, you know if you've looked at what, the impact of automated, vehicles will be on public transportation subways. Buses that, kind of thing so. Yeah. I don't have a slide on this so the one. Point, to make is that the, the. Technical, case for. Public, transit does not go, away so, technical, case for transit, being that in high, demand corridors, you, can move a heck, of a lot more people, through a given amount or ride away per hour, with. With, rail or with buses than you can with cars and even you, know, you. Can look geometrically. You take three prius. C's and park. Them bumper-to-bumper and you get 15 people that's gonna be pretty cozy in a Prius see right get, 15 people in the same amount, of Lane space, you. Can get a 40-foot, bus with. Comfortably. 40 people and 80 and a pinch, so. You can move far more people through a given amount of right away with, with. With rail or bus transit. So. I think that the challenge. That we create is that, by reducing the. Burden, of in vehicle, time it, makes, so one of the big advantages from an individual. Standpoint it's an individual, travel, or choosing, to use transit one, of the one of the few advantages that, it offers, over. Over. A car is the, ability to multitask and. Removing. The need to pay for parking, and. Automated. Vehicles, can give you both of those you can multitask in, a self-driving, car and, then if it's a shared vehicle, it just goes off and serves another passenger, if it's your own vehicle it can drive itself off in park so you might still have to pay but, you don't have to search for parking, so. Those.

Two Big advantages, of transit, get blown away by automated, vehicles so the question is we still have a need for transit, to, move people through congested. Corridors, but what's the individual, incentive, for, each, traveler. To choose to use transit right so it's kind of a it's a it's a tragedy of the Commons, here where, everyone, has this incentive, to travel by car, but. Everyone traveling by car will make everyone's travel, experience, worse and abies may exacerbate, that. Even. Though they do provide some incremental increases, in capacity. So. That's that's that's a challenge, that I haven't. Really heard a good, answer to other. Than, the one that's kind of the third rail which is pricing that, we need to be pricing roads, and have dynamic, or at least time to use pricing. On. Our, roadways so that people, have an incentive to use transit or to use spare. Advance or carpooling, or whatever. The. Other thing now looking away from the highly. Congested corridors. And just looking at the others are the other function, of transit, which is to provide a baseline. Level of access, to everyone. I see. Even. Today if you, in, the Seattle, area where it costs $150, an hour to, run a bus. If. You, could take bus routes with an average occupancy, of same. From memory I didn't have these like nine or less average occupancy of nine or less you. Could more. Cost-effectively. Provide that service with. Shared. Ride Uber's. And, as. You, automate, those vehicles, it makes running a bus cheaper, but, it makes running a fleet of cars a, lot cheaper if you're getting rid of a lot more drivers, and so there's, your opportunity to provide a vastly. Improved. Quality. Of service. Shorter. Weights less walking faster. Travel time to, people to UM particularly. To underserved, communities. With. These on-demand. Automated. Services, relative, to what they get today from. From. Municipal, transit, which is generally, a pretty. Poor product. Okay. Well, there's a ton of more good questions but I just realized I've last couple, minutes over. So. I think I'm gonna close. It out here. On. Behalf of the Alumni Association I, want to thank everyone for, tuning, in to this faculty forum online I think we had closed it to 50 people and, thanks, again to Don for. Sharing your expertise, with us today for. Everyone who did have questions that didn't get answered I'm. Told, alumni office staff are going to forward all those questions to Don that weren't addressed. And. Again, if you want to tweet about today's, chat you can use the hashtag MIT, better worlds, if, you have more questions or, feedback you, can send that to alumni. Learn, at, MIT edu. And. We'll post an archive video of this on YouTube likely. Within the next week so thanks, again for watching. Thanks. For joining us and for more information on how to connect with the MIT Alumni Association, please. Visit our website. You.

2018-08-11 22:57

Show Video


Great information and analysis

Terrific, informative presentation!

Other news