What you need to know about driverless vehicles

0:00Hi, it's Brooke DeVard, host of Naked Beauty, and I want to talk about how I upgrade my hosting experience. There's a moment just before guests arrive when my home stops feeling like a series of rooms and it becomes an atmosphere. The light settles, the air shifts, and then the beautiful scent of jasmine, rose, and fig fill the air. It feels warm and considered. That's what Pura does. It turns fragrance into environment. A collection of beautiful, clean fragrances, all controlled from your phone. Choose your sense of place at Pura.com.

0:42Hello,

0:46welcome to the Naked Scientist podcast, the programme that brings you the biggest breakthroughs and talks to the major movers and shakers in the worlds of science, technology and medicine. I'm Chris Smith and today, in partnership with the Royal Institute of Navigation, I'm climbing aboard Cambridge's first autonomous bus, which is undergoing trials around one part of the city. It's a first for me. Come aboard and hear how I get on.

1:21so today we're exploring driverless vehicles also known as autonomous vehicles or avs these are now set to become a reality in london after transport for london granted private hire licenses to a fleet of autonomous taxis albeit on the condition that a human driver is present there for now but the pace of change overall is incredible in this space and as we'll hear it's not going to be long before this is all very much de rigueur on roads everywhere so to see for myself what all the fuss is about and to sample what boarding the clapamonny bus of the 21st century feels like i've been off to maddingly where they're putting one of these new buses through its paces my fellow passengers on our test run include the royal institute of navigation director ramsey farragher and our Naked Scientist intern, Laura Ferdina, as well as some experts from the industry.

2:18Morning. Hello, who are you? My name's Nick and I'm the autonomous bus safety driver. But you're sitting behind a steering wheel. I am. I'm here just for safety. Tell me what I am getting aboard. This is Cambridge's autonomous bus service, self-driving. on this route we're going around the west side of cambridge and hopefully we'll be driving itself well i've just got aboard i mean this looks to all intents and purposes like a normal bus you look like a normal bus driver i presume that's what your job is normally yes it is yes i'm a normal bus driver the bus looks normal but obviously there are bits and pieces on it that make it special that make it self-drive and what you just press a button and off we go That's correct, yeah.

3:02I can engage the autonomous mode while we're going along, and it takes over. When do you take over? I would take over if there's any obstructions, parked vehicles or anything like that, or maybe there might be some roadworks, and then I will take over. I'm really excited because I've never done anything like this. I've never been in an autonomous vehicle. I mean, for you, having driven one, how long have you been doing this? I've been driving these for now about a year, and it's great. you're not scared it's going to put you out of a job no no well take us away

3:49so we're off on our way nick you're actually driving at this stage though yes i have to be um the autonomous part comes in when i have to be in the right position doing the right sort of speed and then I can put it in autonomous mode. So we're leaving the park and ride at the moment. We'll then go out onto the main road. Does the bus know when to stop, as in when a bus stop comes? Would it know there's a bus stop coming and pull in or do you have to do that? No, it will do. The bus stops are actually programmed into the bus. And does it know about traffic lights? Because we're just stopping at some traffic lights.

4:26you haven't engaged it automatically yet, but were we to be just coming along this road, it would know that's a red traffic light and I have to stop. Yes, it does, yeah. It sees them from quite a good distance away and registers them, and it will stop at a red light. Right, well, I'm going to go and sit down, because I gather that you're not allowed to press the button to make this take control until I'm safely sitting down. But I'm going to go and do that, and then I'm going to enjoy full automation. Enjoy the ride. Right, I'm sitting down, safely in my seat. Are we now being driven? We are. The fasten your seatbelt light on an aeroplane on this bus says autonomous.

5:12It says, big red lights on, it says we're being driven. It's a very smooth ride. I'm going to go and talk to some other passengers, because Laura's with me. Your thoughts? I can't tell that it's autonomous. It seems the same as any other bus. There's a bit of harsh braking going on, but we're now coming up to a junction. This is a T-junction. It's got white lines across to where you would normally stop your car, and then you would look both ways and then turn left. It has done that. Are you impressed, Ramsey? Yeah, especially given the weather, because famously autonomous cars often don't like very difficult weather conditions, and we've got very, very heavy rain, we've got big puddles, There'll be strange reflections on the grounds that it's having to deal with.

5:56The worst situation is if the road is so wet that it can't see the white lines. And it'll be very interesting to chat with the guys today about how they have built the sensors to be smart enough to deal with Cambridge weather, because this is pretty normal as we get into the autumn. Just working my way down the bus, this is a very important passenger who I'm now sitting next to, because I guess you're sort of the guy with the checkbook and the decision making. Tell us who you are. So I'm Dan Clark. I'm the Head of Innovation and Technology for the Greater Cambridge Partnership. We're a collection of councils and we're really interested in how autonomy could support the public transport system.

6:34Tell me where we are today in terms of how this has all come together and why we're all here. So we have some government funding to trial autonomous buses. We're really interested in full-size buses, so not autonomous taxis, which you read about a lot in the newspaper, but really how buses could be part of the public transport system. So we work with a technology provider, Fusion Processing. We have a bus manufacturer, Alexander Dennis, and an operator in Whippet. We brought those people together to be able to run a service. What's really important for us is to be able to understand the technology, how it works, how it operates, and potentially where it could be in parts of the public transport system, but also really interested in the passenger experience.

7:23So how do people get on and off the bus? How do we deal with things like antisocial behaviour and ticketing? And then also the business case that sits behind it. So we, as an area, are likely to franchise our bus network, and those plans are already in place. So we'll be buying buses in the future, and so we need to understand how do we specify and buy this technology and where do we deploy it. What do you see the attractions of this as being? Because we've still got a driver, because presumably the biggest cost to an authority of operating any kind of vehicle like this is the person sitting up front.

8:00That's right, the driver is the biggest cost. So if we can take the driver out, we have the potential to make things like rural bus routes, which at the moment aren't commercially viable, which may see a huge expansion in the future of the bus network. also we think there are opportunities to operate outside the hours that buses normally operate at the moment but it'll also be a spectrum so we think there may be a bus captain or conductor in the vehicle that can help people get on and off talk to people improves the quality of the service and may increase the number of people using it obviously you don't generate the savings that you do if you take the driver out completely but there's still some advantage to be gained And then there will be areas where we can completely take the driver out.

8:45And obviously that really does lower the operating costs and makes running public transport a much cheaper proposition. Does it really translate into a cost, though? Because if you're just substituting a person who would steer the bus for someone who wanders around policing unruly behaviour and making sure people actually buy a ticket, it's still a person, it's still costing you money to employ them. And as a passenger, I kind of like having that human contact, knowing I'm a bit safer, because if someone did get on and start saying things to me I didn't like, I've got an extra person to back me up, etc., etc.

9:19That's right. I mean, and it is really the biggest challenge is how do we make people feel safe and, you know, how do we get people on and off the bus? But I think particularly in certain parts of the country, there are a real shortage of drivers. there's real competition for driving jobs you know now obviously with home deliveries and uber and and other kind of taxi driving jobs so it can be quite difficult to recruit a driver but it may be much easier to recruit a kind of a customer person who would be in the bus and also it increases the the kind of quality of the service because you know whereas the bus driver is normally just driving having somebody in the body of the bus can answer questions particularly a tourist place like Cambridge could maybe sell you a ticket to go punting or you know there are other opportunities I think that open up once you've got somebody in the bus you really thought about this haven't you but in in terms of um the other jobs that and you talk about more specialist clientele the other jobs that drives you they help people with disabilities for example so do you see that as an added attraction that you you've got a bus that can operate autonomously but also have somebody who is freed up from just being behind a steering wheel because they can improve the customer service?

10:34Yeah, potentially. I mean, you could also have somebody who is roaming, so who serves a number of buses, and if you have a disability, you could tell them that you were going to get on a certain bus and they could arrive and help you on the bus and help you off and improve your experience as somebody with a protected characteristic of using public transport. So again, that could be a real benefit, I think, to the travelling public. one of the things that we have done is we've been working with a company called Anthrometric to really understand how people with disabilities could get used you know autonomous vehicles so we built the buses in virtual reality so that people can experience them and give us feedback on the virtual reality and so how does that work then so you put on a virtual reality headset the the bus is rendered in there and then you can experience getting on and kind of using the bus as somebody with a disability.

11:22That sort of optimises the service side of it. So you've got kind of working out how to make it work practically and you've got how to make the experience work both virtually before you then have to implement that. Yes, yeah, that's right. And then obviously the big piece of the digs or puzzle is kind of what's the business case behind it? You know, what are the cost benefits? And yeah, how do we procure when really it's a new technology and the market is just emerging as well? This has been an incredibly interesting experience for me and it almost looks like it's ready to go now. But how far away are you looking before you could say, as a local authority, Cambridge could have a fleet of these running and you'd feel comfortable and safe that they were going to do what you hoped they would do?

12:08So we've just got another chunk of government funding which will take us on for another two years and our plan is towards the end of that to be commercialising services. so we've got a number of organizations that are interested in deploying automated vehicles we will do the business case we will hopefully have you know solved some of the tricky issues around how do we get people on and off how do we make them feel safe and we will have developed a commercial model which will allow people to have the confidence to deploy the technology on the roads and for people to use it. Our overall aim is eventually to make this technology boring.

12:49So, you know, you don't think about it. You get on, you get off, and you don't think there's no driver here. Well, boring it definitely isn't. That was Dan Clark. And also aboard next to me is Jamie Wilson. He's head of concepts and advanced engineering at Alexander Dennis, and they're the company that actually build these buses. It's a development path that they've been following for over a decade, and he regards the process of automation of these vehicles as part of an improvement continuum, adding safety and improved commercial viability, which will culminate ultimately becoming fully autonomous and then deployable into a range of different applications worldwide.

13:26One of our key technology themes and one of the things that we try to solve for our customers is safety on vehicles. We've been a massive part of supporting Transport for London with their bus safety strategy since like 2018 when they launched the bus safety strategy. As progressively technology is advanced, we've introduced things like advanced emergency braking and various degrees of automation and automated controlling to vehicles. I just see this as another step on that roadmap to developing more capability in the vehicles and allowing us to release more safety on the public highway. There's no secret that bus drivers, you know, sometimes they do have off days, sometimes they do get tired.

14:04Fatigue is quite common, especially if you're doing multiple jobs around the clock. and I guess this aims to enhance the driver's capability on routes where they're still in the hot seat driving the vehicle but also it allows us to look at other routes. So there's about 45 % of the cost of running vehicles is attributable to the drivers. If we can make marginally commercial routes more commercial and support operators and delivering services which are cost effective and supportable, that is a good thing for us and it allows us to sell more buses ultimately. I mean, on certain routes, you have to take more than one driver because a person can only be behind the wheel for a certain amount of time.

14:44Does this help to solve part of that equation? Because if you've got less driving time, can you cut down your staff time? Absolutely, absolutely. We see that drivers can only drive for about four and a half hours before they have a break. They do that twice a day as part of an operating licence. So this offers us the opportunity to do a more comprehensive coverage and be able to use the asset. These are expensive vehicles, they're expensive assets, and they need to be used efficiently. So the more you can use them, the more you can get value of them and deliver services for customers, everybody wins.

15:17So buses are part of the critical transport infrastructure. They make their key to unlocking economic activity with moving people to work. They're key to getting people to places to see doctors and access health services, things like that. And as simple as getting your local shopping at the shop. So they're vital to it and the more we can make these services economic, the more we can contribute to solving the transport poverty problem, which is about 5 % of the population of the UK. Doesn't your business make other kinds of vehicles as well? I'm sure my bin lorry is part of your business, isn't it?

15:53Once upon a time, the Dennis part of Alexander Dennis did make bin lorries, but we also made tricycles back in the day and lawnmowers. Before we divested, we now focus down on buses and coaches specifically. We're part of a larger group, New Flyer Industries, which is a global organisation. It's one of the largest independent bus and coach-focused companies globally. And it's great to see that we're getting lots of support from the wider group in terms of pushing ourselves forward and preparing ourselves for this next stage in automating vehicles. So I might not get my bin emptied automatically, but your positioning globally means you're in a really strong position then, potentially, to see this technology extend as a British company beyond the shores of Britain.

16:39Absolutely. In the UK, most of the bus services are run privately by private operators, and therefore it's a very lean-focused industry, let's say. So there's been a motivation to try and solve the problem of affordability in the UK and provide transport services. Wider in the world, you've got things like airside and landside application near airports. You've got car parks and more simpler use cases. And we see demand globally through the world with this. On bus, dedicated busways, and the role in franchising as well in the UK, is now starting to look at the context of connected services. and if you start looking less at a bus service more as a transport system this is where autonomous vehicles and autonomous buses specifically play into a large part of it.

17:30They're much less expensive than trains. They're easy election winners because you just deliver them within very simple election cycles and another thing is that they have much more penetration in terms of being able to service the sort of minority groups in society that can access public transport otherwise. Jamie Wilson at Alexander Dennis.

18:18in our control. O-Pill is a daily birth control that's FDA approved, full prescription strength, and estrogen free, and 98 % effective when used as directed. Grab it online or at most major retailers, no prescription or doctor's appointment needed. So if you're thinking about birth control, check out O-Pill to see if it's right for you. Use code Giggly for 25 % off your first month of O-Pill at O-P-I-L-L.com. That's code Giggly at O-P-I-L-L.com. Birth control in your control. We love to see it.

19:11This is The Naked Scientist podcast with me, Chris Smith. Now today I'm taking my first ride aboard an autonomous vehicle, in this case a bus, which is being road tested in West Cambridge to explore how this technology can revolutionise public transport in cities and rural communities. We're in partnership with the Royal Institute of Navigation. Now let's head a bit further down the bus and we'll find out who is also on board. So my name's Jim Hutchinson, I'm CEO of Fusion Processing and we provide the technology for this vehicle. So we provide automated drive systems which can be fitted to any vehicle and allow it to drive itself.

19:48Does that go right through from the cameras that are watching what's going on around this vehicle right through to translating that into where the wheels go? Yeah, that's right. We provide a full system. So it's all of the sensors, all of the processing and any communications modules as well. And so essentially our system is taking all of that data. It's working out what's going on around the vehicle. It's working out what the people are doing, what the vulnerable road users are doing, what the other traffic's doing, what the traffic signals are doing. And then with the knowledge of its destination, it's working out how to get the vehicle safely to where it needs to be.

20:24When I climbed aboard, I said, well, this just looks like any normal bus. Is your system retrofitted to a normal bus or has this bus been designed around your system? So our system can be retrofitted. actually what we're working towards together with Alexander Dennis is a factory fit. Let's start working through the technology then. What's actually going on under the hood to make all this happen? So we've got a range of sensors on the vehicle. So this includes optical sensors like cameras and lidars, and there's also radars as well, and they're positioned around the vehicle. We tend to take a fairly sort of subtle approach to that.

21:00I know, you know, there's different approaches and some like to make a feature of it. We're trying to make the vehicle look not at all scary, fairly ordinary to people. So they're installed to give a full 360 vision around the vehicle. And then they will pass their data to the processing units, which essentially have various levels of software and AI built into them. And their job is really to make sense of all of that data. And together with the knowledge of where it's got to get to and where all the stops are, then it will calculate a safe path. And it breaks that down into lower and lower levels to the point where it's actually sending the messages to the vehicle actuators.

21:41So the steering, the brakes and the throttle to actually, you know, make that route happen. How do you train it in the first place? Well, this is something which is proprietary and we all do things a little bit different. We've been going since 2012, so we've captured a lot of data in that time. So we're using a lot of historical data plus a lot of other sort of learning sources which we have in our armoury in order to make the system function. So you've got historical data of vehicles like this working their way down roads and looking at the various road markings, other vehicles, signs and so on.

22:19Is it taking all those data points into account? It can read road signs, see traffic lights and you've sort of fed that in and told it this is what these features are. so it knows how to recognise them. Is that what's going on? Yeah, so essentially it will understand the infrastructure, it will see the road markings and it will understand where there are pedestrians, where there are cyclists, etc. and it will take all of that into account in its path planning. A lot of vehicle manufacturers regard their vehicles as sensors on wheels these days and they're just hoovering up enormous amounts of data about the road surface, the driving conditions and they use that as a learning experience to inform not just what that vehicle does but what their entire fleet of vehicles do.

23:01So do you have that built into your system so that what this vehicle encounters every minute of every journey is informing how it will react and how the rest of the fleet will react in the future? Yeah, that's right. So we are collecting a lot of data and that's really quite important as the systems evolve and getting better and better and more and more capable. So, yeah, the data is an important part of that. I notice in my car, which is not an autonomous car, but it does have autonomous features, if I'm, for instance, approaching a bend and the road curves round to the right 90 degrees and there's a car parked on the corner, it keeps trying to slam the brakes on, thinking I'm about to go up the back of the car that is just down the track, which I wouldn't.

23:45I'll be turning round the bend. Does this system fall for the same problem? And is this a risk in the future or is this surmountable? So the sorts of systems that are on most cars at the moment are what we would class as a level two system. So they have some useful capability and they can improve safety in a lot of scenarios. But they are quite a long way from being a level four autonomous system, which means it's safe in all scenarios without any driver intervention. So what we have is a level four system. And so, yes, it's a lot more capable than what the average car would have. Is there an insurance slash economic model built around this now?

24:24Because people often say that it's difficult. How do you decide who's at fault when something like this goes wrong? Is the industry mature enough to cope with this yet? Has that been solved? The sort of insurance model for this so that when something goes wrong, because one day it will, how that's dealt with? Yes I think some insurers have been very proactive in this and we've been working with them for quite a few years now. It's not something which is prevalent yet but yes there are solutions there for that now and the legal frameworks and liabilities agreements and that sort of thing are all fairly mature at this point now.

25:02I'm just going to make my way to the back, your back row boy Ramsey. I bet you always sat on the back seat of the bus didn't you? It's been a very long time since I was being naughty on the backseat of a bus, Chris. Are you taken with this? What do you think? Well, I mean, it's fantastic to see autonomous vehicles arriving. It's been a long time coming. I first worked on autonomous vehicles about 20 years ago in the defence sector and had all of the challenges of how to make the LIDAR, the radar, the GPS, the cameras and the inertial sensors all make sense and talk to each other. But we're in a completely different age now with AI.

25:36it's a lot easier to do a very hard task of replacing a human being and how clever we are. In terms of cleverness and the computing power, what is it taking to realise what we're experiencing? Well, driving a car is actually a very difficult challenge if you think about it. So you learn to drive the car and you pass your driving test and you're given a licence. but you only do that aged about 17 or 18 because you fully understand how the world works so you drive down a road and there's a petrol tanker on fire next to the road you don't drive past that petrol tanker on fire but the road's completely empty to a lidar to a radar to a camera to the GPS it's safe to drive forwards but you and I know how dangerous a massive oil tanker on fire just to our left is.

26:29Similarly, you're sat at red lights one night and someone starts coming towards the car with a knife or a gun. You drive through those red lights. You don't mind breaking the law for that moment to get yourself out of that dangerous situation. And we're currently making sure that autonomous vehicles can obey the road rules. But there's this bigger picture of how how complicated the world is. And autonomous vehicles, back 20 years ago, when you wrote if-else statements in code in order to make them drive, there was no way you were ever going to code up a petrol tanker on fire might explode. But with modern AI, we're literally programming intelligence via simulated brains.

27:13And for the first time, we will be able to have autonomous vehicles that are not only trained with the difference between a red, a yellow and a green light, but what it means if they see a man with a knife, what it means if they see fire, what it means if they see a big crack across the road that wasn't there last time. And the computation required is quite serious. There will be big computers in very clever cars in the near future. But potentially when you're using technology at the bleeding edge like that, there are vulnerabilities. Yes, the more complicated the system is, the more ways there are to try to mess with it in many ways.