Sarah Dods: the last episode we explored how real-time data is changing how transport networks are planned and operated. Today we're going to look at a different layer of that story. What happens when we start designing ⁓ around how people actually behave on our roads? To explore that shift I'm joined by Penry, one of GHD's most experienced licensed
Tash Wilson: You've been listening to Transform by GHD. Big ideas, real impact.
Sarah Dods: Welcome back to Transform. My name is Sarah Dodds and I am the APAC lead for Advanced Analytics and AI for GHD.
Tash Wilson: Welcome to Transform, tomorrow's thinking today. Real conversations on the innovation shaping businesses and our society.
Sarah Dods: traffic engineers, bringing over 30 years of understanding from US transport systems. Frank, great to have you with us. Thank you, Sarah. Great to be here. Great. Let's get into it. So Frank, you've been on tour across New Zealand and the Australian Eastern Seaboard for the last few weeks now. Part of that has been talking about your work as a transport engineer. And part of that has been observing the way people drive differently in different cultures. What are some of the observations you've got from that really side by side of observations? Yeah, I'd say that with time here ⁓ in Australia, the stark difference between ⁓ the observation what here in Australia would be a upper limit to speed zones ⁓ and what very clearly in these states, a lower limit. Driving upon the roads in the US, we would functionally drive at a ⁓ well above the posted limit. Here in Australia, I found particularly ⁓ interesting is the strict adherence to posted limits ⁓ the changes in those limits. So when, ⁓ we drive through construction zones or other ⁓ potential hazards, that adherence to lowering or reducing speeds when necessary are required ⁓ by the environment, everybody moves that speed down. Right, and we complain like crazy about it being there and go, why should we have to do this? But we do actually slow down. And I'll just remember the first time you told me about this difference being horrified. What do you mean? you drive the speed limit. ⁓ How can work? But it's an example of how it's a really deep cultural difference that you were so surprised that Australians are compliant. The compliance aspect of it was fascinating from my perspective to see how you have almost universally across the areas in which I've been and to some extent a good portion of the major cities is that strict adherence with some margin ⁓ associated with variations. when those speed limits flashed and the speeds went down, you saw the universal ⁓ collective of all the drivers all kind of moving together in that lower speed. That's something you don't see in the US. To be honest, ⁓ of a lower limit, if you were to ask anybody, the speed limit as posted is a lower limit. If you're not going that, move as far over to our right, your left as possible, and don't get in their way. ⁓ It's ⁓ culturally a very distinct difference. ⁓ The development of the speed limit is absolutely the same. It's the adherence to it ⁓ that is the difference. ⁓ We do the same thing. We look at our 85th percentile speeds and we set that limit accordance with what good driving behavior would look like behavior what everybody should be doing within that margin and it's just not what happens interesting about how you change that ⁓ I still from ⁓ when I learning to drive being told that the safest speed to drive at is the speed of the other traffic around you and It's the diversity of speeds in the same section of road that is probably one of the bigger risks on our road Absolutely differential and speed is probably the largest contributor to collisions on American roads Because we have such high know people do conform to the speed limit and in some cases ⁓ Adherence to the limit or slower speeds is usually due to an attention ⁓ or distraction raises some interesting questions about what's going to happen as more autonomous vehicles hit our roads because they will be programmed to travel at the speed limit. And if that's not the platoon speed, then you're actually introducing risk. That's a great point. would there be some ⁓ within the AI to drive an autonomous vehicle at a higher speed based upon the presence of ⁓ higher vehicles that are on the And that's a whole lot of really interesting AI ethics to unpack about do you do the safer thing or do you do the legally compliant thing when they are not aligned with each other. So Sarah, ⁓ this year while we were at TRB together, you ⁓ kind of significantly said that this was the year driver behaviour became a ⁓ What shifted in your mind? What I found unexpected about TRB ⁓ there were a number of papers there that were explicitly about understanding driver behaviour, measuring driver behaviour and how you can influence driver behaviour. We had people coming up to the booth we were at going, hey, do you guys do ⁓ driver behaviour? And then talking to them about what we're doing with Compass data and with connected vehicle data. which really is a direct observation in the wild of what drivers are doing on the road in an anonymised fashion. But it is exactly that scientific observable that we can use to understand the behaviour that we want to change to keep drivers safe and make their driving patterns safe for the other road users around them. I thought that was astonishing and ⁓ delightful and really empowering. ⁓ I'm a data nerd. ⁓ As far I'm concerned, anything that want to do, you have to be able to measure. What gets measured gets managed. And up until now, we've been measuring pieces of road ⁓ and what happens on pieces of road. the things that we care about on roads, the economics and productivity of freight movements and drivers, the people not dying on our roads is about drivers and humans we want to keep safe. So it makes perfect sense to me that we want to understand how humans are behaving and then how are we going to be able to effectively influence that behaviour towards a safer outcome. Our earlier conversation about Australia and the US have very, very different starting points in terms of the culture around how both populations behave on the roads. The outcome both countries are looking for is the same. How can we encourage drivers to behave in a way on our roads that minimise the potential that they're not going to survive the journey or somebody isn't going to survive the journey? I think what's going to work for those two different populations is really, really different. But the outcome and where the connected vehicle data comes into it is we can try things. We can understand what works. We can measure the effectiveness of how it works. We've done a whole bunch of before and after studies in Australia about here's an intervention, here is the behaviour pattern before the intervention, here is the behaviour after intervention. Did it change? Was it sustainable? Or did the behaviour gradually revert to where it was as drivers got used to it being there? Yeah, yeah, I don't need to pay attention to that. really is. It's a big difference. We've been able to observe ⁓ for years. Probe data, tube data, whatever needs to occur, we see it at a very high level. What is happening on the road with regard to speed? This observable that you mentioned ⁓ very How ⁓ do think we can leverage that? so first of all, I'm a scientist. I believe in the scientific method. I believe that we can have a hypothesis out there that we think this is going to work and that we poke holes in it. And you need data to understand whether you are making ⁓ both significant change and a statistically ⁓ relevant And this data can help us actually measure both of those and ⁓ report back in terms of... bodies being able to go, okay, here's how much it cost us to do that and this is the outcome from it. And whether it's an evaluation outcome in terms of we have reduced speed by ⁓ kilometres an hour along a road, along a much longer section of road across a neighbourhood, we can do those studies ⁓ this data. ⁓ that longevity, that where the cars are coming from and where they're going becomes a part of that. It's not just that isolation of where a collision, that bit of information, that probe, that location where something is occurring, but to see more globally about what is happening. When we're looking at what happens out on roads from a safety perspective, realised risk, where people have unfortunately passed away from a road accident or where their lives have been changed from a serious injury on the road, one of the things we struggled with for a very long time is understanding whether that was a random one-off combination of factors that isn't going to be seen again or whether there is a repeating pattern of risk. at that location that says an accident is inevitable. Unless we do something to change that section of road, there is going to be an accident and there are probably going to be multiple accidents. When we start combining the accident history with high g-force events, braking and swerving events, with speed data and potentially looking at the region around the road, we can start layering up a pyramid about saying, yes, there was a fatal accident here. but the speed is compliant, we've got no swerving, no braking events, there's nothing else in that space that seems to be an issue versus we've had an accident here. We've also got little red dots that represent the near misses and there's 15 and 16 of those. And then we've got some information about if the speed is compliant, there's clearly something causing those hygiene events that may be an inappropriate speed limit, that may be. non-compliance with the speed limit but there is something that we can do at that location to help drivers understand what a safe speed is to drive at. ⁓ does not replace the super important work that is done around investigating accidents. Where somebody has died on our roads we need to understand why and that is necessary but I think it's no longer sufficient to be able to get the safety outcomes that we're looking for to really understand how we can influence towards zero, which is what everybody is shooting for. It becomes that observable ⁓ ⁓ like this that helps us understand and correlate that to the greater ⁓ significant injuries, right? I mean, it's a pyramid. You mentioned that pyramid. We need those lower layers to really understand where ⁓ collision that results in, well, an unfortunate fatality or even significant injury differs where collisions may not always look different at face value. by the type of the collision itself or the speed associated, what is the correlating factors that go into that? And we need those observables along the way. And that helps ⁓ engineers and the people who design our roads to go and have a look at those locations that may be a hotspot because there's been a single accident ⁓ has been a hotspot because it looks like a really high risk area and then be able to ⁓ do some work to prioritise there isn't sufficient funding, which of those should have the high priority for funding. You guys have ⁓ a funny term with that identification of those hot spots, right? Do you call them black spots? ⁓ to me, ⁓ we don't that term. when I of a black spot, I of something a little different, black ice, a little dark spot. ⁓ in things Yeah and you know this brings me to the other opportunity here is not just being able to understand the data and the scientific observations what we really to understand is how to get drivers to I won't say behave better But how do we influence drivers' behaviour to behave less in unsafe manners? So one of the things that we do in Australia around that black spot locations is we put up a sign saying, this is a black spot accident area. Be aware, people have died in this space. And sometimes even that signage and the awareness that this is a potentially hazardous environment can actually influence drivers to behave more safely. been some fun analysis done ⁓ Compass themselves, I think, ⁓ one of the state governments in Australia on ⁓ around an event that's happening further up ⁓ freeway. You call them motorways, we call them freeways. Telling drivers to slow down doesn't work. Telling drivers that there is an accident ahead might be more effective. information what message is going to be effective, what message is not going to get through, and that is cultural, that is not scientific. Well, ⁓ it's behavioural as opposed to engineering and road science. ⁓ And the to bring in nudge and behavioural economics to how do we help drivers behave more safely, I think that's a massive opportunity globally. ⁓ The carrot the stick, I mean, is really what it comes down to. I think we've... ⁓ all tried enforcement. say enforcement seems to be a bit more driving conformance here in Australia than it does in States. ⁓ understanding you're required to or leading you to a certain ⁓ outcome, I would become more appropriate. And I'd love to be able to test that ⁓ through the data ⁓ in the States. Understanding ⁓ you know, that sign alongside the road that says there's been so many collisions on this stretch and there's a reason why you need to have your headlights on and why you need to travel more safely becomes very apparent and testing whether or not that is an appropriate measure is really going to be fascinating. Yeah, I'm laughing because one of the signs we see along the road in Australia is speed cameras enforcing in this area. So it's coming back to the actual hazard we don't care about. potential of getting the demerit points and in Australia quite a large fine associated with being caught doing not very much over the speed limit is actually a fairly effective deterrent and in the US it kind of works differently. Unfortunately it works very differently but hopefully we can also learn from Australian colleagues and start to bring some aspects ⁓ ⁓ from behavior to ⁓ to the mix so we can really understand what is working and what is not working. Because all we see is ultimately at a probe level data, whether or not speeds are going down, but we don't always know why speeds are going down. Congestion drives speeds down, but if congestion isn't there and enforcement isn't there, left to our own demise, ultimately we tend to keep creeping up and that ultimately is leading to those higher fatal significant rates that we see across the country. Right, and we can also have a look at how drivers are responding and are we getting a broader range of speeds on the same stretch of road, which is not safer necessarily, or are we actually getting a uniform drop of drivers moving to a lower speed, which is actually the safer outcome that we're looking for. What does that transition look like and how safe or unsafe is the transition and how do you get to that safer outcome? I think that's just an awesome conversation for us to have in a lot of countries around the world. Towards zero is a challenge for almost every global government. And in most countries, it's currently going the wrong way. ⁓ So maybe this can help us by shifting ⁓ the understanding from ⁓ what's happening on a piece of road to ⁓ this is what we think safer would look like. and we have our own cultural pathway to get there, but this is the data that might help us understand for our particular set of circumstances what is effective and what's not. Frank, thank you so much for joining us for that conversation. I've learned things I didn't know about the US. Great to chat, safe travels home, and thank you everybody for listening.
Tash Wilson: You've been listening to Transform by GHD. Big ideas, real impact.
Sarah Dods: Welcome back to Transform. My name is Sarah Dodds and I am the APAC lead for Advanced Analytics and AI for GHD.
Tash Wilson: Welcome to Transform, tomorrow's thinking today. Real conversations on the innovation shaping businesses and our society.
Sarah Dods: traffic engineers, bringing over 30 years of understanding from US transport systems. Frank, great to have you with us. Thank you, Sarah. Great to be here. Great. Let's get into it. So Frank, you've been on tour across New Zealand and the Australian Eastern Seaboard for the last few weeks now. Part of that has been talking about your work as a transport engineer. And part of that has been observing the way people drive differently in different cultures. What are some of the observations you've got from that really side by side of observations? Yeah, I'd say that with time here ⁓ in Australia, the stark difference between ⁓ the observation what here in Australia would be a upper limit to speed zones ⁓ and what very clearly in these states, a lower limit. Driving upon the roads in the US, we would functionally drive at a ⁓ well above the posted limit. Here in Australia, I found particularly ⁓ interesting is the strict adherence to posted limits ⁓ the changes in those limits. So when, ⁓ we drive through construction zones or other ⁓ potential hazards, that adherence to lowering or reducing speeds when necessary are required ⁓ by the environment, everybody moves that speed down. Right, and we complain like crazy about it being there and go, why should we have to do this? But we do actually slow down. And I'll just remember the first time you told me about this difference being horrified. What do you mean? you drive the speed limit. ⁓ How can work? But it's an example of how it's a really deep cultural difference that you were so surprised that Australians are compliant. The compliance aspect of it was fascinating from my perspective to see how you have almost universally across the areas in which I've been and to some extent a good portion of the major cities is that strict adherence with some margin ⁓ associated with variations. when those speed limits flashed and the speeds went down, you saw the universal ⁓ collective of all the drivers all kind of moving together in that lower speed. That's something you don't see in the US. To be honest, ⁓ of a lower limit, if you were to ask anybody, the speed limit as posted is a lower limit. If you're not going that, move as far over to our right, your left as possible, and don't get in their way. ⁓ It's ⁓ culturally a very distinct difference. ⁓ The development of the speed limit is absolutely the same. It's the adherence to it ⁓ that is the difference. ⁓ We do the same thing. We look at our 85th percentile speeds and we set that limit accordance with what good driving behavior would look like behavior what everybody should be doing within that margin and it's just not what happens interesting about how you change that ⁓ I still from ⁓ when I learning to drive being told that the safest speed to drive at is the speed of the other traffic around you and It's the diversity of speeds in the same section of road that is probably one of the bigger risks on our road Absolutely differential and speed is probably the largest contributor to collisions on American roads Because we have such high know people do conform to the speed limit and in some cases ⁓ Adherence to the limit or slower speeds is usually due to an attention ⁓ or distraction raises some interesting questions about what's going to happen as more autonomous vehicles hit our roads because they will be programmed to travel at the speed limit. And if that's not the platoon speed, then you're actually introducing risk. That's a great point. would there be some ⁓ within the AI to drive an autonomous vehicle at a higher speed based upon the presence of ⁓ higher vehicles that are on the And that's a whole lot of really interesting AI ethics to unpack about do you do the safer thing or do you do the legally compliant thing when they are not aligned with each other. So Sarah, ⁓ this year while we were at TRB together, you ⁓ kind of significantly said that this was the year driver behaviour became a ⁓ What shifted in your mind? What I found unexpected about TRB ⁓ there were a number of papers there that were explicitly about understanding driver behaviour, measuring driver behaviour and how you can influence driver behaviour. We had people coming up to the booth we were at going, hey, do you guys do ⁓ driver behaviour? And then talking to them about what we're doing with Compass data and with connected vehicle data. which really is a direct observation in the wild of what drivers are doing on the road in an anonymised fashion. But it is exactly that scientific observable that we can use to understand the behaviour that we want to change to keep drivers safe and make their driving patterns safe for the other road users around them. I thought that was astonishing and ⁓ delightful and really empowering. ⁓ I'm a data nerd. ⁓ As far I'm concerned, anything that want to do, you have to be able to measure. What gets measured gets managed. And up until now, we've been measuring pieces of road ⁓ and what happens on pieces of road. the things that we care about on roads, the economics and productivity of freight movements and drivers, the people not dying on our roads is about drivers and humans we want to keep safe. So it makes perfect sense to me that we want to understand how humans are behaving and then how are we going to be able to effectively influence that behaviour towards a safer outcome. Our earlier conversation about Australia and the US have very, very different starting points in terms of the culture around how both populations behave on the roads. The outcome both countries are looking for is the same. How can we encourage drivers to behave in a way on our roads that minimise the potential that they're not going to survive the journey or somebody isn't going to survive the journey? I think what's going to work for those two different populations is really, really different. But the outcome and where the connected vehicle data comes into it is we can try things. We can understand what works. We can measure the effectiveness of how it works. We've done a whole bunch of before and after studies in Australia about here's an intervention, here is the behaviour pattern before the intervention, here is the behaviour after intervention. Did it change? Was it sustainable? Or did the behaviour gradually revert to where it was as drivers got used to it being there? Yeah, yeah, I don't need to pay attention to that. really is. It's a big difference. We've been able to observe ⁓ for years. Probe data, tube data, whatever needs to occur, we see it at a very high level. What is happening on the road with regard to speed? This observable that you mentioned ⁓ very How ⁓ do think we can leverage that? so first of all, I'm a scientist. I believe in the scientific method. I believe that we can have a hypothesis out there that we think this is going to work and that we poke holes in it. And you need data to understand whether you are making ⁓ both significant change and a statistically ⁓ relevant And this data can help us actually measure both of those and ⁓ report back in terms of... bodies being able to go, okay, here's how much it cost us to do that and this is the outcome from it. And whether it's an evaluation outcome in terms of we have reduced speed by ⁓ kilometres an hour along a road, along a much longer section of road across a neighbourhood, we can do those studies ⁓ this data. ⁓ that longevity, that where the cars are coming from and where they're going becomes a part of that. It's not just that isolation of where a collision, that bit of information, that probe, that location where something is occurring, but to see more globally about what is happening. When we're looking at what happens out on roads from a safety perspective, realised risk, where people have unfortunately passed away from a road accident or where their lives have been changed from a serious injury on the road, one of the things we struggled with for a very long time is understanding whether that was a random one-off combination of factors that isn't going to be seen again or whether there is a repeating pattern of risk. at that location that says an accident is inevitable. Unless we do something to change that section of road, there is going to be an accident and there are probably going to be multiple accidents. When we start combining the accident history with high g-force events, braking and swerving events, with speed data and potentially looking at the region around the road, we can start layering up a pyramid about saying, yes, there was a fatal accident here. but the speed is compliant, we've got no swerving, no braking events, there's nothing else in that space that seems to be an issue versus we've had an accident here. We've also got little red dots that represent the near misses and there's 15 and 16 of those. And then we've got some information about if the speed is compliant, there's clearly something causing those hygiene events that may be an inappropriate speed limit, that may be. non-compliance with the speed limit but there is something that we can do at that location to help drivers understand what a safe speed is to drive at. ⁓ does not replace the super important work that is done around investigating accidents. Where somebody has died on our roads we need to understand why and that is necessary but I think it's no longer sufficient to be able to get the safety outcomes that we're looking for to really understand how we can influence towards zero, which is what everybody is shooting for. It becomes that observable ⁓ ⁓ like this that helps us understand and correlate that to the greater ⁓ significant injuries, right? I mean, it's a pyramid. You mentioned that pyramid. We need those lower layers to really understand where ⁓ collision that results in, well, an unfortunate fatality or even significant injury differs where collisions may not always look different at face value. by the type of the collision itself or the speed associated, what is the correlating factors that go into that? And we need those observables along the way. And that helps ⁓ engineers and the people who design our roads to go and have a look at those locations that may be a hotspot because there's been a single accident ⁓ has been a hotspot because it looks like a really high risk area and then be able to ⁓ do some work to prioritise there isn't sufficient funding, which of those should have the high priority for funding. You guys have ⁓ a funny term with that identification of those hot spots, right? Do you call them black spots? ⁓ to me, ⁓ we don't that term. when I of a black spot, I of something a little different, black ice, a little dark spot. ⁓ in things Yeah and you know this brings me to the other opportunity here is not just being able to understand the data and the scientific observations what we really to understand is how to get drivers to I won't say behave better But how do we influence drivers' behaviour to behave less in unsafe manners? So one of the things that we do in Australia around that black spot locations is we put up a sign saying, this is a black spot accident area. Be aware, people have died in this space. And sometimes even that signage and the awareness that this is a potentially hazardous environment can actually influence drivers to behave more safely. been some fun analysis done ⁓ Compass themselves, I think, ⁓ one of the state governments in Australia on ⁓ around an event that's happening further up ⁓ freeway. You call them motorways, we call them freeways. Telling drivers to slow down doesn't work. Telling drivers that there is an accident ahead might be more effective. information what message is going to be effective, what message is not going to get through, and that is cultural, that is not scientific. Well, ⁓ it's behavioural as opposed to engineering and road science. ⁓ And the to bring in nudge and behavioural economics to how do we help drivers behave more safely, I think that's a massive opportunity globally. ⁓ The carrot the stick, I mean, is really what it comes down to. I think we've... ⁓ all tried enforcement. say enforcement seems to be a bit more driving conformance here in Australia than it does in States. ⁓ understanding you're required to or leading you to a certain ⁓ outcome, I would become more appropriate. And I'd love to be able to test that ⁓ through the data ⁓ in the States. Understanding ⁓ you know, that sign alongside the road that says there's been so many collisions on this stretch and there's a reason why you need to have your headlights on and why you need to travel more safely becomes very apparent and testing whether or not that is an appropriate measure is really going to be fascinating. Yeah, I'm laughing because one of the signs we see along the road in Australia is speed cameras enforcing in this area. So it's coming back to the actual hazard we don't care about. potential of getting the demerit points and in Australia quite a large fine associated with being caught doing not very much over the speed limit is actually a fairly effective deterrent and in the US it kind of works differently. Unfortunately it works very differently but hopefully we can also learn from Australian colleagues and start to bring some aspects ⁓ ⁓ from behavior to ⁓ to the mix so we can really understand what is working and what is not working. Because all we see is ultimately at a probe level data, whether or not speeds are going down, but we don't always know why speeds are going down. Congestion drives speeds down, but if congestion isn't there and enforcement isn't there, left to our own demise, ultimately we tend to keep creeping up and that ultimately is leading to those higher fatal significant rates that we see across the country. Right, and we can also have a look at how drivers are responding and are we getting a broader range of speeds on the same stretch of road, which is not safer necessarily, or are we actually getting a uniform drop of drivers moving to a lower speed, which is actually the safer outcome that we're looking for. What does that transition look like and how safe or unsafe is the transition and how do you get to that safer outcome? I think that's just an awesome conversation for us to have in a lot of countries around the world. Towards zero is a challenge for almost every global government. And in most countries, it's currently going the wrong way. ⁓ So maybe this can help us by shifting ⁓ the understanding from ⁓ what's happening on a piece of road to ⁓ this is what we think safer would look like. and we have our own cultural pathway to get there, but this is the data that might help us understand for our particular set of circumstances what is effective and what's not. Frank, thank you so much for joining us for that conversation. I've learned things I didn't know about the US. Great to chat, safe travels home, and thank you everybody for listening.
