In response to a dramatic increase in speeding drivers beginning with the pandemic stay-at-home advisories in 2020 and continuing now, traffic safety experts and law enforcement officials are working to understand the behavior.
On this week’s edition of the Talking Michigan Transportation podcast, Carol Flannagan, director of the Center for Management of Information for Safe and Sustainable Transportation (CMISST) at the University of Michigan (UM) and research professor at UM's Transportation Research Institute (UMTRI), talks about her research and theories about the epidemic of speeding and other risky behavior.
This comes as the Michigan Office of Highway Safety Planning (OHSP) and the National Highway Traffic Safety Administration (NHTSA) are once again partnering on a regional traffic safety campaign. As noted in their news release: "As part of the 'Great Lakes, High Stakes' campaign, dozens of municipal, county and Michigan State Police (MSP) law enforcement agencies across Michigan will focus on speeding drivers between June 18 and 26."
Among other topics, Flannagan talks about the challenging decisions for auto manufacturers in rolling out automated vehicle (AV) technology. She acknowledges that it's much easier for AVs to communicate and predict what other vehicles will do rather than what humans will do.
She also discusses:
Also discussed, the resistance and challenges to acceptance of Advance Driver Assistance Systems (ADAS) and projections for broad adoption. Just this week, the NHTSA released data on the topic, but there are concerns about context.
And a reference to a supercut video of Elon Musk predicting the timeframe for broad deployment of AV technology.
Podcast photo: Carol A. Flannagan, Ph.D., Research Professor, University of Michigan Transportation Research Institute (UMTRI).
Jeff Cranson: Hello! Welcome to the Talking Michigan Transportation podcast. I'm your host, Jeff Cranson.
Cranson: As you know if you've been listening to this podcast regularly or keeping up with the news the past couple years, the pandemic launched what some would call an epidemic of speeding across the roads in Michigan and many other states. I want to examine that more closely. We've talked to some national experts in the past, and we've talked to some people with Michigan State Police about it, and what they're seeing. Today, we have some new data thanks to the University of Michigan Transportation Research Institute. And I’m fortunate to have with me Carol Flannagan, who is a research professor at UMTRI, and director of the data center. She looks at a lot of information related to safe and sustainable transportation. So, Dr. Flannagan, thank you for taking time to be here.
Carol Flannagan: Sure, no problem.
Cranson: So, let's start high level with what you think has been going on the past few years, and how it relates to higher crash rates, not necessarily early on—the crash rates themselves were lower, but the crashes that did occur were more severe and more people died. What does your data show you?
Flannagan: Right, yeah so, it was actually really interesting when we started getting data from the pandemic, from 2020, and started looking at it because, when you took it from a 10,000 foot view, there were sort of these two messages. So, people of course stopped driving. I mean, almost completely stopped driving in March/April and then were, on the year, the total miles were substantially lower. And yet, the total fatalities was actually up a little bit. When you initially look at the long-term trend, the total fatalities for 2020 actually looked like it was in line with what had been about a 10-year increasing trend. So, to try to make a word picture, starting from about the end of the great recession, when people started to have more money and started to get back on the road, fatalities have actually slowly been creeping up on the whole and within lots of categories. So if you look at pedestrians and pedestrian fatalities or alcohol fatalities, these things have been creeping up. In 2020, the totals looked like they were in line with that trend if you just look at that graph. But then when you look at driving miles, there's this huge drop in 2020. So then, if you think about dividing one of those by the other one, you get a rate. And suddenly the fatality rate exploded in 2020. And so, when we started digging into that, what became clear was that the sort of risky behaviors like drunk driving, speeding, which were focusing on drugged driving, unbelted, fatalities—all of those things really spiked in the summer of 2020 after the lockdowns eased up. And so, whereas driving in general was still down, and even crashes in general were down by quite a bit, suggesting that people were either doing extra risky things or anybody who was inclined to do something risky didn't stop that. The things that stopped for the pandemic were maybe more commuting driving, which is pretty ordinary. So, hopefully that made sense.
Cranson: No it did, and I saw it anecdotally, you probably did, too, when I was on the road. And even on the local streets, not just on the freeways, that people were just more aggressive. And I never understood like, we're in a pandemic, why are we in such a hurry?
Cranson: And I think maybe it's just stress. I mean, I don't know how much of your research overlaps or delves into behavior, but what are some of the causes you think? How does it relate, I mean, clearly just not as many cars being on the road makes people drive faster, but…
Flannagan: Yeah, that's a good question. And of course, the data that we have access to, they are police reported crash data. And so, they don't exactly ask people, “What were you thinking?”
Flannagan: [laughing] It’s an interesting question. And to be honest, I’m unaware of other research. Although, I’m sure that people have sort of asked that question. But if you think about, in a sense, the crashes speaking for themselves, it was clearly the risky behaviors that not only remained. So I have several theories about it. And one of those theories is that people who, let's say, were commuting stopped commuting. So if commuting driving occurs at two times a day, it's really quite safe on a per mile basis. People are generally not drunk. They mostly can't speed because there's congestion. And so generally what it results in is a whole bunch of rear end, fender benders that don't result in fatalities. That's sort of typical. So if you take away all of that kind of driving, and what we saw in the crash data was that the amount of rear-end crashes was down almost 40 percent in 2020. So you take away that kind of driving, and you leave—so this is one theory, is you leave behind all the other stuff. You can think about risky choices. So in the pandemic, going out was a risky choice, going out without a mask. And we don't know, obviously, if people are wearing masks. But generally, it is plausible that people who are comfortable going out in the pandemic might also be sort of generally more risk comfortable. But so, that's one idea is that. I call that the “only the good miles disappeared” kind of thing. And then you would get this basic sort of retention of the bad outcomes. But, when you look closely at the month-to-month changes, the summertime, all of those risky behaviors were way above normal levels. And anybody who was going to drive drunk still drove drunk—there was something beyond that. So another idea that I have is that the open roads or frustration over the pandemic itself, which was certainly frustrating, it was frustrating to be locked down for anybody. It was stressful. And so, the combination of those things potentially led people to speed more than they would have otherwise.
Cranson: Yeah, it's almost like grinding your teeth, right? The things that you do subconsciously out of stress. And maybe hitting the accelerator is just like that. You're wound up, and you're angry, you're stressed, and so, instead of an instinct to protect oneself, that kind of goes away.
Flannagan: Yeah, and again, I don’t have data that says that that's what people were doing. But it is intuitively plausible. And I think anybody listening to this knows how they felt at the time. I can tell you that I was pretty stressed out and frustrated when we first were sent home. And so, like I said, the sort of risky behaviors component, the crashes, individual crashes, were just more deadly at that time on average, and those behaviors were higher than normal. Yeah, so this is consistent with the idea that people were possibly blowing off steam in some way.
Cranson: In terms of your own stress, I mean, anybody in academia, whether they're teaching at the secondary level or at the college level, what the students and professors were going through trying to figure it all out and still be productive. I mean, I can imagine how difficult that was. And still is to an extent, because people are still trying to come out of this and sort out the new reality. So going forward, one thing that Office of Highway Safety Planning is talking about, with conjunction with their umbrella agency, the Michigan State Police, is some stepped up enforcement over the last couple weeks of June. And do you feel like, based on your research in the past, can that, I mean, it could make a difference during that period, especially the more that they get the word out through media, let people know that this is happening. Can that have some long-term effect?
Flannagan: Yeah, I mean, I certainly hope so. The evidence on high visibility enforcement is that it works. It's the combination of the enforcement being present but also making sure that everybody knows that that's what's going on. So that when you're out, that if you are a person who is inclined to speed, you might think, I don't want to pay the fine, and I don't want to get caught, and so on, that you might make different choices. And I think that what seemed to be going on prior to the pandemic was that we were sort of hitting whatever the normal level was at that time with respect to speeding. And I was actually going to take a quick look at the graph, but I don't think, speeding wasn't much on the rise, at least in terms of fatalities and speeding. Yeah, it was actually quite—sorry, I know you can't see it, but I’m looking at a graph that runs from 2011 to 2020, and speeding related fatalities are basically flat through that whole period.
Cranson: But then they jumped from 2019, right?
Flannagan: Well yes and no, so what really happened—it's hard to, you have to build a word picture, but essentially what happened in 2020 is that the total number of speeding crashes was in that same flat range. It wasn't higher; what changed was that there were so many fewer miles. So when you look at the rate of speeding fatalities, it jumps. When you look at the sheer number, it continues to be flat. So if you think about that, you say, well that's kind of a weird coincidence that it landed at about the same level. But what's really going on is, from 2011 to 2019, there was this kind of flat level of speeding in fatalities, in speed to fast crashes. And so speeding was occurring at kind of whatever its level was. People were speeding, sometimes that resulted in deadly crashes. It's not a good idea, but you know it was kind of at a level. And then suddenly in 2020, there's way fewer people on the road, there's way less driving, so way fewer miles. And so then speeding, and you can see it in the month to month, speeding related fatalities actually spiked in July. The total was offset by having hardly anybody on the road in March/April. But now what appeared to happen in 2021 was that people notched up their level. So it's like, you can imagine this kind of jump in speeding, and then suddenly you reach this sort of new normal. And so enforcement, and especially high visibility enforcement, and returning to the messaging that occurred prior to the pandemic is one reasonable tool for trying to put downward pressure. It's sort of too bad to say, well, could we please get back to 2019 levels? But that's the current reality. And so if we can push down on people's inclination to speed, if we can relearn reasonable driving behavior, and then obviously what we want is to reduce it further going forward, if that made any sense.
Cranson: No, it makes sense, and I think you're absolutely right about the enforcement and the threat of enforcement. And often it's the threat of the financial pain. We know from studies that young people, especially who feel invincible, they're not so scared from the messaging that says you might die. So you focus on how much it's going to cost you if you get a speeding ticket or if you get a drunk driving arrest. Which, OHSP had a campaign that focused on that, and you saw the young male being put in handcuffs and thinking about how much it was going to cost him. The Governor's Highway Safety Association expert has told me that they found, in some studies, that the fear of disfigurement affects teens more. They're more scared of suffering an injury that hurt their face then they are dying. So, it's all about trying to find the things that resonate with the people, the most vulnerable, the most likely to do this kind of behavior. So, tell me your theory on this, if we know that speeds increased and that we saw leveling off, and then, as you say, back to 2011 coming out of the recession, we started seeing more crashes, just more risky behavior. Our cars are safer than ever. We have all kinds of technology. UMTRI is a big researcher in this field and works with OEMS on coming up with these kinds of things that are working, testing them. And whether it's, as we move toward automated vehicles, the incremental changes with adaptive cruise control and lane control and all those things. So, I mean, where are we going ultimately if we keep making our cars safer, yet we have more crashes? Are we creating a mindset of, I’m in a bubble and nothing can happen to me, or what do you think is going on there?
Flannagan: Yeah, that's a super interesting question. And I know if you've been reading the news about NHTSA’s investigation into Tesla crashes and other automated vehicles, or ADAS, we call it advanced driver assistance systems, they are definitely concerned that there's a tendency on the part of the driver to become kind of complacent, to over rely on systems as they are now, which is quite an interesting question. So the basic bottom line on vehicle technologies; so I use the term ADAS, which is what a lot of people have in their cars, they're not just Teslas. You mentioned adaptive cruise control, there's lane centering. There are also sort of more emergency style safety systems like automatic emergency braking, blind spot warning, lane change warning, and so on. So one thing that we know is that, especially the frontal automatic emergency braking, works really well. Our estimates are that it reduces frontal crashes by like 50 percent. And we also know that electronic stability control, which was mandated back in 2011, works really well. And you can see in the data that it reduces loss of control and rollover crashes, and you can clearly see it in the data that because they mandated it, there was this sudden spike to 100% fleet penetration in new vehicles. And you see this big drop in rollovers for those vehicles, and similarly, AEB isn't mandated, but the auto manufacturers have generally committed to having it in all their new vehicles in a short time frame, and so it's ramping up. And automated vehicles, like robot cars, really show a lot of promise in, well, for one thing they won't speed, they definitely don't drink, some of these.
Cranson: Don't get distracted, right?
Flannagan: Right, exactly. So there are other challenges with those vehicles, obviously. People's brains are very flexible and can sort of manage new situations in ways that are challenging for AVs. But in principle, the promise over the next five or ten years is certainly huge. One of the trends that was happening, actually a while ago, was a shift pretty quickly from warning systems, which are meant to engage the driver, “hey, you seem to be coming up too close to that vehicle in front of you”. Those were sort of the primary systems a number of years ago, and the manufacturers shifted pretty quickly to automated systems. So this is with forward crashes, which again, are the ones that are managed the best by automation. So you have like the warning. You hope that the driver hits the brakes, but then you have the automatic emergency braking, which is going to be like, ah, we're way too close, I’m just going to hit the brakes for you. And it turns out that AEB, the automation version, is much more effective than forward collision warning, the version that tries to get the driver to do something.
Cranson: So I think that it's incremental, and we have to adapt, and we have to accept it. There are so many people—you know this better than anybody—that say that they're opposed to automated vehicles, let alone the concept of fully autonomous. I think Tesla made a huge mistake. I mean, I understand from a marketing standpoint why they used the term autopilot. But it was a huge mistake, I think, and because people think of a pilot in a cockpit getting up and being able to walk away. It's just not a mindset that you want to create right now. But I also think that over time we're going to accept that, while we all think, I'm a good driver but those other people aren't, that the computer is a better driver. And if we didn't think the computer was better than us, then we would never use a calculator, right.
Flannagan: [laughing] Right.
Cranson: Stay with us. We'll have more on the other side of this important message.
Narrator: Did you know that most work zone crashes are caused by inattentive motorists? It only takes a split second of distraction to dramatically change lives forever. The Michigan Department of Transportation reminds you to slow down, follow all signs, and pay attention when driving through work zones, because all employees deserve a safe place to work. Work zone safety—we're all in this together.
Cranson: Tell me how long, I mean, this is crystal ball stuff, and nobody's going to hold you to it. But how long do you think, based on your colleagues at UMTRI, and what they're seeing, and what you hear from industry, until we really are like the cars in sci-fi movies and futuristic looks like Minority Report where we are fully automated?
Flannagan: Mm-hmm yeah. That's a very funny question. So I think, I’m trying to remember when it was; I think in something like 2015, I was on a panel somewhere and got asked, and there were other panelists, one of whom was actually working in development of automated systems, and they asked us the same question. And I think I said 2025 at that time, and now we're at 2022, and it's clearly harder than it looks. And I also watched a pretty interesting piece on Tesla and Elon Musk the other day where they showed a sequence of clips of him in like 2017 saying we're one year out, and then in 2018 he says we're one year out, and it goes on like that.
Cranson: Huh. I haven't seen that.
Flannagan: Yeah, it's very good actually. So, from talking to OEMs and talking to AV companies and stuff, what's clear is that a huge proportion of the problem of driving and normal driving is really pretty well tackled by AVs. And the problem is that, now we're into the…diminishing returns isn't quite the right word, but each incremental change is harder and harder. And so sorting out the nuances of ways that people think about it, the nuances of cues that human drivers get from other drivers about what they're about to do that the AVs don't necessarily do. And so, and I’ll get to your actual question in a minute. But the other thing that I kind of want to comment that I find…I mean it makes sense; it's fascinating, but the thing is, when we have automation all over the place, the problem is so much simpler. Because those vehicles can all tell each other what they're planning. They can co-plan together, and they'll be like okay, you go first, and then I’ll go after you. So, the coordinated version where all the vehicles are talking to each other and coordinated is so much easier to solve than the version that the companies are forced to solve right now, which is to have a little lonely, automated vehicle tooling around in traffic with a whole bunch of humans whose state of mind and plans are unknown. And so, I think that's one of the ironies that I find is that the problem that they have to solve now is actually substantially harder than the problem that will need to be solved once it gets rolling.
Cranson: Very interesting way to think of it.
Flannagan: Yeah, so my best guess, in 2015 I said 10 years out, and now we're sitting at 2022, and I’m sort of inclined to say 10 years out. But that seems overly pessimistic, so I’ll adjust forward five years. I mean, there is a taxi service in Phoenix, Cruise is on the verge of deploying theirs for business purposes and things. So we're getting there. But like I said, it's just it's harder than it looks, and people's tolerance for error on the part of a robot is wildly lower than their tolerance for error on the part of fellow humans.
Cranson: Absolutely. The acceptance is so difficult because the reporting is so often out of context. I mean, it seems like every time there was a crash that involved an uber with automated technology or a Tesla, early on, that was the focus, that was the headline. And there was very little follow-up later when you found out that it was the other driver the human driver that caused the crash.
Flannagan: Right, yeah.
Cranson: That makes it difficult. Well, thank you, Dr. Flannagan, for taking time to talk about this. I really appreciate the work that you and UMTRI do and the support it provides to MDOT and OHSP and to NHTSA and others across the country. I think it's really valuable.
Flannagan: Sure, actually, this is fun. Thanks so much for including me.
Cranson: Thank you again for listening to this week's edition of the Talking Michigan Transportation podcast. I would like to thank Randy Debler and Corey Petee for engineering this week's podcast. To subscribe, to show notes, and more, go to Apple podcasts and search for Talking Michigan Transportation.