Latest / Autonomy Insiders / The UK's Autonomous Vehicle Permit Process Explained
Transcript
- Daniel Abreu Marques: Welcome to Autonomy Insiders, the show where global industry leaders in autonomous driving unpack their real-world insights. Tym: to have a supervised driverless vehicle on the road, you can just do that. There is no statutory requirement behind that. they are taxi drivers, Uber drivers over at Wayve. there's no legal difference between the current service available on the Uber app to take on a supervised autonomous vehicle ride and a standard Uber vehicle ride. politically they can't stop it this forever, but actually legally they can actually delay this for a very long time. and ultimately for London forever, because they are the consenting authority. Daniel Abreu Marques: Welcome to Autonomy Insiders. I'm your host, Daniel, and on September third, London got its first Robotaxi ride-hailing service open on the Uber app. And you might get a Ford Mustang MACH-E driven by Wayves AI, but there's still a licensed driver behind the wheel. And removing that driver is a completely different legal process as of now, as no company had received that permit. So how does it actually work? Who approves what and why is it taking longer than many expect? My guest today is Tim Syrytczyk Tim co-founded the Institute for driverless Transport, an independent UK research body that studies what autonomous vehicles mean for Britain. And he also writes self-driving insights. And his latest piece breaks down exactly how you get permission to run a robotaxi service in London. So, Tym welcome to the show. Tym: Great to be here. Thanks for the invite. Daniel Abreu Marques: Tym you started self driving insights and then co founded IFDT. So what gap in the UK conversation made you think a newsletter was not enough? Tym: Yeah, so great question. The UK has had incredible research and interest in autonomous vehicles over the past twenty years, you could say even longer. And there's great work being done right now in the government. right now we're processing through the legislation as to how we can admit any form of autonomous vehicle service in the UK. And there's also great academic work. But outside of academic work and the government, there was Not enough industry work in our view. There were some academic and industry partnerships, but not a lot of considerations as to how to make this a technology integrated in the economy. Which areas of our economy, which cities would benefit the most from autonomous vehicles, and what would be the long-term implications of that? There was quite a lot of conversations on business cases and safety cases in the mobility sector, but we wanted to think: well, how does the world look different 20 years from now? A sort of analogy that we like to describe is it would have been great if we had an institute for AI in 2015. If we could actually think, what's gonna happen in 2030, 2026 about AI? And then maybe some of us would have been able to do research about labor transitions or adoption in different economic sectors. So that was the motivation behind IFDT. We had a launch event in April, attended by about two hundred, and ever since then we've been more and more busy with questions and answers and research that we've been very happy to take on here. Daniel Abreu Marques: Sounds good. And who is IFDT then really for? Is it the government, is it the industry or the general public? Tym: I would say the public, right? Ultimately we want to benefit the UK and benefit economic outcomes, social outcomes, also make sure that the geopolitical conversation is made more salient, and also we are really interested in the job displacement question. Those are our four areas. All of them are areas that benefit the government, they benefit the general public, they also benefit the industry to successfully roll out in a way that minimizes disruption and brings out economic benefits. and we're not focused entirely in the UK either, right? A lot of our interest has also been the public and transport operators from outside the UK. So we're open to that conversation. We're open to producing research that hopefully is a general purpose good. And just as I wrote about the permitting licenses here in the UK, I might in the future write about permitting licenses in other countries if I think there will be benefits to do so. What motivated me to write this particular piece was because I personally thought there wasn't a very good explainer of the government legislation. There's a lot of gov.uk writing, which is an excellent website and all the details that you need will be there. But for something that like an everyday person or someone really interested in the industry but not very familiar with all the legal jargon or the legal agencies involved, this was a great way to move forward. And one question I want to preempt. as well is who are we and who funds us? This is a question that we get quite often. We are an independent research organization. We're not part of any academic organization and we're currently philanthropically funded. So we've previously received a grant from the Mercatus Center, known as Emergent Ventures. This is a one-off grant, no strings attached, just to help us get off the ground. And now we're planning to write commissioned research but also apply for other grant processes that hopefully will allow this small organization to continue in the near future. So yeah, hopefully that addresses that. Daniel Abreu Marques: And you mentioned your your piece about the permitting process already. So most AV coverage is about the technology. So why should someone who builds AV software care about the UK permit process just as much as about the technology? Tym: Well, if you care about the technology being used, if you care about outcomes, then the permitting technology and the licensing agreements are just as important as the software. We can have the software, we can have a great operation going in test cases, in private land and private roads, but we want passengers on the road and we need more than just software to make that work. We need a lot of safety assurances. We need A lot of understanding of who is responsible in any given moment. We need integration with other forms of transport and other forms of action that happens on the road. So for instance, we need to know what to do when there is an emergency vehicle driving past the software, not just from the software side or driving past the robotaxi, for instance. Not just on the software side, but also from a operational side. So those are the motivations. Not every software engineer has to be interested, but if you're interested in the outcomes, this is my recommendation. Daniel Abreu Marques: Alright, so let's clear up the basics first. So everyone listening has also the same vocabulary. So in the UK law, what is the difference between a supervised autonomous rides and a driverless one? Tym: Great, so a supervised autonomous ride is a ride where the user is still in control. That means that there is still a driver being responsible for the behavior of the vehicle. This was defined in the Automated Vehicles Act, and even if the vehicle is driving by itself all of the time, because it is legally defined as a supervised ride. The person, the safety driver, will be responsible. A driverless vehicle doesn't have that. Currently, there are no driverless vehicles legally in the UK operating as driverless vehicles. They have to be placed as supervised vehicles because they don't have a permit to operate as driverless vehicles. So that's the distinction, and the conversation we'll have about the permitting scheme will define how we can switch from a supervised into a driverless vehicle. Daniel Abreu Marques: Yeah, this make sense. And this legal difference comes also from from two different laws, which probably confuses a lot of people. So there are two key laws. It's the Automated Electric Vehicles Act from twenty eighteen and the automated vehicle act from twenty twenty four. So what does each one do Tym: What does eight? Daniel Abreu Marques: and why does the twenty eighteen act still matter in twenty twenty six? Tym: Yeah, great question. So the Automated and Electric Vehicles Act of twenty eighteen provided the first sort of insurance and definitions of what an autonomous vehicle is and when would responsibility be set if there would ever be a payout. This is actually really important even for any trials or tests, so that even if there are no public uses of autonomous vehicles, we can conduct trials and if something goes wrong All the money gets passed around legally as clearly defined. In practice, I think, in the vast majority of cases, we've been always in a user-in-control environment. But it also set up a very important foundation for the Automated and Vehicles Act in 2024 and for these trials that we see today. The Automated Vehicles Act of 2024 wants to define autonomous vehicles permanently in the UK for all use cases. So that will explain who will be responsible for licensing, who will be responsible for operating. It will define any sort of incident investigation, everything that you can consider the state or the government being responsible or interested in when there's an autonomous vehicle on the road. This is where it's getting defined. The law has passed in 2024, but the way British law has worked for a while, and the way that even European Union law works, is that even when the act gets passed, there's a lot of definitions and a lot of secondary legislation or clarifications that have to be published. and that happens often in conjunction with industry. And right now, this regulatory environment will be completed or scheduled to be completed by twenty twenty seven. So by the end of next year will we have everything legally defined in autonomous vehicles? Right now we're in this middle stage. Daniel Abreu Marques: And the laws are the one thing but the people who apply them are another. So can you walk us through the cast, so the Department of Transport, the VCA, the DVSA and TfL so who does what? Tym: Yes, there's quite a lot of acronyms here. as always you can reference gov.uk or search visa and DVSA transport or the like. So let's start from the national government. The national government sits in London, Westminster, and has several departments. The one that we're interested in today is the Department for Transport. The person responsible for the Department for Transport is a politician. We're currently held by Heidi Alexander, the role of Secretary of State. So whenever you see SOS, it actually doesn't mean that anyone's in trouble in transport. It means that we're referring to the Secretary of State, the most important politician, who officially will be the person to grant the license. In practice, the Department for Transport has different agencies, has different organizations that help do the decision making. There's the Driver, Vehicle and Standards Authority, which does the vast majority of the permitting process. They will accept the application from an operator to operate a private hire like vehicle service or a robo taxi service. And they will process it, they will request a safety case, and they will be the ones to give approval on behalf of the Secretary of State. At times, another agency will get involved. This is the Vehicle Certifications Agency, VCA. Both the DVSA and the VCA are part of the Department for Transport. So ultimately they'll be working very closely together. They'll be, in fact, working parallel to each other in different applications. And those are the major players from the government side and the central government. Then there is the local transport operators and the local consenting authorities, the organizations that have the local voice in mind whenever a permit gets licensed. In the case of London, that will be Transport for London. The operator that operates the London Underground, the tube network, as it's known, and also the bus network and light rail services. In other places in the UK, it might be other transport operators, or transport authorities, to be precise. Transport authorities don't always include operators like TFL, but instead could also include local government. So the transport authority in the city of Leeds will be the City of Leeds Council. so as these licenses get applied in other places in the UK, there might be different local authorities or local voices or local consenting authorities that will be in play. Daniel Abreu Marques: And the government also has fast Tym: And then you don't want to. Daniel Abreu Marques: tracked the the pilot scheme from late twenty twenty seven to spring twenty twenty six and what did fast tracking actually change and what did it maybe also leave the same? Tym: Fast tracking allowed for the automated passenger service permitting scheme to be opened. The automated passenger service permitting scheme is the legal pathway through which any company that wants to operate an autonomous vehicle in a bus-like service or a taxi-like service can get the legal permission to do so, including fully driverless services. So you'll be legally allowed to take out a driver. Have a remote operator, or not even have a remote operator, if you can make the safety case for it, and supervisor, and have these on public roads. Currently, this is the only legal pathway to do so. It is a introduction of a part of the regulatory environment earlier, so that. Full regulatory considerations for instance, nationwide permitting haven't been considered yet. We've only started considering these local services for very particular use cases. So for instance, there isn't a legal pathway for semi-public or private, local or government-owned private autonomous vehicles. For instance, so companies can apply for the automated passenger service. But it is less clear as to for instance, if you didn't want to do a bus like service and you want to use this for commercial operations with some humans involved, then there might be other pathways to consider. Daniel Abreu Marques: And what happens to such an pilot permit when the permanent framework arrives? Tym: The current framework of automated passenger service permits are by nature temporary. So they have to be renewed. And when the when the full regulatory environment arrives, they'll keep that same structure, essentially. there'll be maybe more clarification as to how quickly you can get a license, but the same process will be at play. So the same actors, the same paperwork will be important. Daniel Abreu Marques: let's follow a company through the process in order. So it starts long before any passenger gets in. So Waymo or Wayve and also Apollo Go, they are all testing in London. Yet there is no trial permit. So how does that work? Tym: So to have a supervised driverless vehicle on the road, you can just do that. There is no statutory requirement behind that. If I personally buy myself a Tesla and I have the driver assist features, whatever they may be, the lower features of a full self driving supervised, which isn't yet legal in the UK, I can do that because I am ultimately responsible as the driver for this. So in these trials, the companies are collecting data across London with however many vehicles they want, with however many supervisors or safety drivers. And ultimately, there's only a guidance and a recommendation that the government published to support this, to make these trials as conducive with government preferences as possible. known as the code of practice. As mentioned, it's not statutory, but it allows these companies to gather evidence and conduct these rides in a way that's legally safe and also might help with any future applications for autonomous passenger services when they want to take out the driver eventually. Daniel Abreu Marques: And you mentioned the code of practice is non statutory, but what happens if a company then ignores Tym: company. Daniel Abreu Marques: it? Tym: Well, most ways that they can ignore it would involve criminal activity. It that has been long defined, for instance, passing a red light. So the standard practices of fining would occur. Points on a license, eventually there might be consider it would fare badly as well on any applications for autonomous. passenger service permits, but ultimately we haven't yet got into that situation. Later on, as we talk about the permitting scheme, we notice there are elements of judgment. There's almost elements of trust, both from the DVSA and from TFL, towards the operators. So here the relationship would get damaged, even though there might not be a legal mechanism to directly punish the company to the scale that maybe some people would consider important if this code would be violated. Daniel Abreu Marques: Yeah, and I think one part of the code is Tym: Yeah, then I think one part Daniel Abreu Marques: also a safety case that is expected and you mentioned it several times already. So in plain terms what goes into that or into a good safety case and also who actually reads that safety case. Tym: So the safety case is a form of evidence which explains why a particular autonomous vehicle is allowed to operate on the roads or should be allowed to operate on the roads and why it's safe to do so. It includes not just the definition of the software and what can the software do in which environment, but also explanations as to what is the company doing to ensure safety from an operational standpoint, cybersecurity considerations, vetting and screening of the safety drivers. Considerations of interactions with road authorities. and all of this can be updated, changed at any point, because the operators should provide the most recent evidence and these trials can help in upgrading the safety case. Any UK authority, in this case, the DVSA and also the VCA would both be readers of the safety safety cases. The VCA The vehicle certification agency would be interested most in the software and in the vehicle design itself. So let's say there is a vehicle design that's submitted for a permit that doesn't have a steering wheel. Then there would have to be specific considerations of okay, how can we be sure that remote operations are as reliable as we want them to be? We assume reliability of control by a human with a steering wheel, but there will be additional evidence required for this particular case on control of the vehicle. The DVSA will be more interested in the operational side, but both would want evidence. And evidence from the UK in particular will be excellent for these organizations because they want to know safety on British roads, safety around edge cases on Roundabouts. That might be harder evidence to gather in the United States, for instance. Daniel Abreu Marques: Makes sense and now you mentioned several times the VCA and one thing that VCA does is also the vehicle special order. So can you ex explain what is the vehicle special order and also when does a company need one? Tym: Yeah. the vehicle special order is a license that makes sure that the vehicle is safe. obviously, in the case of autonomous vehicles, it will consider things like do we actually need rear view mirrors for safety here in this case? So the VSO will be a new application for any new design of a vehicle. So, in my view, the Ford Mustangs probably did not require a VSO, but I'm not entirely sure. but if we were to you know bring out a Zoox vehicle onto British Roads, a VSO would be required before you're even allowed to do the trials. This is different to the code described earlier. This is a legal requirement, that has always been the case. Daniel Abreu Marques: And do you also have an estimate how long such an VSO usually takes? Tym: From what understand, this is a matter of weeks for set blueprints. I think the more complex the design, this could be a matter of months, but at first it could be less than a month, especially when it's a small modification. Daniel Abreu Marques: And looking at the company, so how does Tym: That's cool. Daniel Abreu Marques: a company know it has tested enough in a city to apply for the driverless permit? Is does is there a specific threshold in in the laws or does the company decide? Tym: The company decide. The company decides when to make a application because being granted the automated passenger service permit doesn't mean you have to take out the driver immediately. You can be given a permit on the conditions that you are expecting to take out the driver and that you'll reapply for a permit without the drivers later. I think from a bureaucratic standpoint, most firms do not want to apply for the same permit twice just to take out the driver. but they are very well able to do so. from a government standpoint, they would prefer this application to be reconducted. So there is no baseline, there is no ceiling and there is no floor, and this will be an active conversation. It's very scale dependent because if you make a application to provide a fixed route bus service, the scale of evidence and miles of reliability will be a lot smaller than if you want to put in a very heavy vehicle, many of these vehicles on the most busy roads in Britain. so it's case proportionate. Daniel Abreu Marques: Makes sense Wayve and Uber found a way to carry passengers Tym: Way too very present. Daniel Abreu Marques: before any driverless permits exist. So Wayve and Uber did not use the new APS permit which we talked about earlier. So which route did they take and also why did it? Tym: Wayve and Uber collectively applied for a standard private hire vehicle license. So they are taxi drivers, Uber drivers over at Wayve. there there's no legal difference between the current service available on the Uber app to take on a supervised autonomous vehicle ride and a standard Uber vehicle ride. For any operator or any person who wants to start again the license and start driving in London, you need three different licenses. You need a license for the driver themselves. This is different to the driving license. You need a private hire vehicle driving license. So usually this requires a criminal background check, right? You can be a criminal and drive around, but you probably shouldn't be a criminal and drive strangers around. There's the operator license, which is making sure that there is fine operations for any sort of private hire vehicle service. So for instance, if you want to have a mini cab service, you have to make sure that you have a clear booking page that communications and business processes are friendly to customers. And then you have the vehicle license, which is the TFL's assessment of whether the vehicle is suitable. For passengers. And here, Wayve would have had to apply to have their modified Ford Mustangs to get a vehicle license. Each of the 15 cars available today would have their own separate vehicle license. So this would be an advanced MOT, you could describe it as done by TFL instead. So Uber already had the operator license and they have a very good process to get driver licenses. And Wayve or Uber would have applied for the vehicle license. Once they got all free, it's legally perfectly fine for those humans to drive or not drive, but supervise the driving. But legally they are still liable. Daniel Abreu Marques: Okay, so but there would be no Tym: Okay. Daniel Abreu Marques: different license for the driver because like it's an AI driver more or less that is just supervised, so it it's still the driver in the vehicle that needs to have to have a license. Tym: Yes. Daniel Abreu Marques: And do you think this is more like an shortcut route or is it this a natural first step for everyone that comes after? Tym: Step for everyone that comes after. it's an opening. It's a path that is available but doesn't have to be taken. And it could make things a little bit quicker and easier. It could not necessarily. It doesn't have to be a shortcut. Any autonomous vehicle passenger service could be opened entirely without considering this legal pathway of private hire vehicle licensing from TFL. After all, if you want to start a service in Manchester, you shouldn't. need to have a shortcut through London and TFL to do this. however, evidence of operating in the UK, evidence of operating in London, can help in building future safety cases. Hey, between you know, July 2026 and December 2026, there were 15 Wayve vehicles operating this many times and there were this many incidents. That's very good evidence for a future permit application. That's how it can be helpful. On the other hand, if an operator might not feel mature enough or might make certain mistakes, it could also create negative evidence for consideration as well. so hence why it's not immediately a shortcut, but in most cases a responsible deployment will help in providing a full driverless license. Daniel Abreu Marques: And in this setup as far as I know, Uber owns and operates the cars and Wayve only supplies the driving software. So why does that split matter for the permits? Tym: So in the automated passenger service permit, there will be only one operator, one company, providing the application and the evidence. And it is their responsibility to manage all third party responsibilities and relationships. This means that both Wayve or Uber could make that application. And how does it matter that currently Uber has the operating license? Not much necessarily. Wayve could very much make that same application. it depends who wants to submit all the evidence required. Uber will have better evidence in some details, especially on the operational side right now, saying, hey, we will have these vehicles driving in this location and we'll have these standards of safety around the depots, for instance, so that there is no malicious intent placed onto those cars. but From a software standpoint, well, Wayve will have way better evidence. Here are our edge cases, here's our safety statistics for millions of miles of testing. and there isn't a direct relationship because it's a separate legal process. What happens in the private hire vehicle licensing doesn't mean that the same organizations have to make the same applications or use the same evidence even. It's just probably helpful for them to do that. Daniel Abreu Marques: now let's go to the main event. So what it takes to get a human out of the driver's seat So we talked now a lot about the the automated passenger service permit. So the APS permit. Can you explain what it is in plain terms and also maybe why does DVSA issue it and not TFL? Tym: So the APS permit is this fast-tracked service for the government's providing or structure for any company to provide an automated service for passengers publicly for profit, for commercial reasons, and it may be a private hire-like style service, or it may be a bus-like style service. So This is the only legal pathway to do so. It's defined as part of Automated Vehicles Act 2024. It will be the permanent way as to how these things get processed. And as mentioned, it is processed by the DVSA. Why is it DVSA and not TFL? Well, the DVSA is a national agency, so they'll be able to take in applications that are considered by any place in the UK that apply to any location in the UK. But why aren't local authorities the major sort of consenting and applicant bodies rather than just consenting bodies? Well, in part it might be that not every local authority has the same bureaucracy, the same capability of assessing this evidence. the national agency will have many experts on transport, many legal experts on hand. The Leeds council may not have this many experts on hand. So it makes for more institutional sense and more institutional knowledge to be built on the hand of the central government. Also the central government wants to have control here. They don't want to be surprised by a local authority admitting loads and loads and loads of autonomous vehicles that they themselves might have not predicted. So control is very important here. And TFL is maybe slightly frustrated that ultimately they don't have the final control. But they have had some control placed in this structure as well. Daniel Abreu Marques: And every applicant must then answer basically two questions. So can the vehicle drive itself safely and also can the business run as a safe passenger service? So what's in your opinion the harder to prove? Tym: I believe it will be the vehicle driving safely, is the harder to prove. often we'll see, an announcement that businesses that are already providing safe passenger services are applying. Uber is going to apply, most likely Bolt or Lyft may apply in the future. Local transport authorities, so Stagecoach, might apply if they ever get an autonomous bus. These are companies with a very long historical track record. of legally operating in the UK, meaning that for them that should be the easy Daniel Abreu Marques: maybe now let's switch to to the vehicle side. So the VCA assesses whether the vehicle can be listed as self driving and your article says the car cannot rely on a human watching and stepping in immediately. So what does that mean for the remote assistant? Tym: So for a vehicle to be fully defined as self-driving, it has to provide full commercial routes or a very, very high standard or maybe entire standard of driving without any direct remote control. Waymo has a structure in which remote operators can provide driving suggestions and the vehicle itself makes a decision as to which route it controls, it takes. This is probably what the secondary legislation, this permitting scheme had in mind. They do not want the idea of a driving wheel being set away from the vehicle and yet controlling the vehicle. So in any circumstance a vehicle is not fully legally driverless unless it provides evidence that it can drive and it never needs human control. To complete a route. How many routes, all routes, does it have to go up Mount Everest completely autonomously? Probably not. That is then probably later set in judgment and it has to be a very high bar. Daniel Abreu Marques: And will companies have to disclose how often a remote staff helped the car? So like this in the Waymo model, this is purely on assistance and giving some suggestions but not driving, but do they need to disclose that? Tym: So there's nothing explicitly in the gov.uk site which explains that they do have to disclose that. However, from a safety case standpoint, it will almost certainly be expected. so that's my interpretation of the text. in part because it explains how certain safety cases can be addressed and it might also provide a very important trend line, for instance If there is a reduction over time in the amount of remote assistance supervision or interventions, then the company will have a more credible case that we can get it they can get it to a very low number or to even to zero. Where if there is a flat number or an increasing number when you know it rains in the UK or that it hails or the ice gets slippery, well then that will be the sort of evidence that the VCA And the DVSA will consider. Daniel Abreu Marques: Makes sense. And for every service, there's also the operational design domain and there's the area which the permit allows. So maybe just explain for the audience the difference between the two and also who draws each boundary. Tym: So I believe the operational design domain will be a concept very familiar to your audience. it is the boundary in which a autonomous driving system can operate. and it's often used in any application. It's often sometimes even explained to investors as to what is technically capable for a vehicle to do. So it might drive between these two points and it might only be able to drive reliably at this hour or this level of light or this sort of weather condition. The automated passenger service permit area is very similar. It will also have definitions alongside the ODD. So just as the ODD will define location, the service area will define location. And just as it defines hours of operation, it will define the hours of operation. By nature it will have to be a subset It will have to be smaller than the ODD. And therefore, the DVSA will be very interested to hear about the ODD from the operators and from the autonomous vehicle technology providers. And that's the difference. the ODD is a industry term, whilst the permit area is specifically only a UK term, but will be, you know. A concept used across all licenses, I believe, at least Daniel Abreu Marques: Absolutely. Tym: in this level four environment. Daniel Abreu Marques: Yeah. And can a company also expand this permit area mid permit or does it then need a new application when it wants to expand Tym: It will require a new application and in preparation for this I also researched it seems that it will also require new consent from the consenting authority. So not only will they have to resubmit an application with DVSA saying, hey, we want to expand our permit area here and here's the ODD evidence and here's our safety case evidence that this can be possible. They'll also have to do that not just to the DVSA but in London's case to TFL as well. So then TfL has to be convinced. By definition, that these two authorities have to be addressed separately, that implies that they'll have separate bars of consideration. One of them might have a judgment that's a little bit harder to pass than another. But maybe they will have the same bar. Well it's that is not entirely clear. Daniel Abreu Marques: then there's also the service side which is less about the software also more about the people and we have now the vca and dvsa and I think they can run in parallel probably and what do you think realistically how long is it from an full application to a granted permit Tym: Unfortunately, from a full application to a granted permit it's not just DVSA and the VCA to consider, it is also the local consenting authority. So the VCA probably will move the fastest. They check that the vehicle is built in a safe fashion and that all copies of the vehicle will be built in a safe fashion and they'll assess the software. Once they do that assessment, the VSO will be granted. And they might provide some evidence as well for the DVSA to consider. Now the DVSA will also consider the operational side. Ultimately, however, the bar can also still stop at the local consenting authority. In this case, it is TFL. When a permit gets finalized and the permit service area gets finalized, the DVSA will submit that information to TFL. They'll say, hey TFL. Here's our idea for this new service. There's gonna be 15 Wayve cars, and we're gonna take all the drivers out and they're gonna operate here in North London. And this is what they're gonna do when there's a power cut, and this is what they're gonna do when it snows. what do you think about that? Are you happy? TFL has six weeks to respond. If they don't, they say, well, you didn't say anything. you're probably happy then. Boom. Consent is granted if there's no response. However, if TFL gives a reason, they say no, and here's our reason. Have they considered accessibility requirements might come up. Then there'll be a conversation. From what I understand, a rejection with a reason will almost always lead to the application just collapsing. And then a new application would have to be set up. So that application would get rejected. Now Obviously, in bureaucratic terms, that's how it's defined in practice. If someone submits one line in rejection, I'm not entirely sure what happens. No, because we don't want to. I is that a good enough reason? I'm not sure. I feel like people will take respect to the system and not play around these edge cases. But the way that the law is written right now, that sort of edge case might appear at some point. however, that is currently the process. How long, right? Very politician answer. I still didn't give you a time. We don't know, is Daniel Abreu Marques: Mm-hmm. Tym: the blunt way to put it. There's no sort of base rate. There's once we get one application through, they'll say, hey, we applied eight months ago and now we got the permit. Then we'll know, it probably takes roughly eight months. However, if there is a lot of rejections at once, you know, you know, TFL is unhappy. provides evidence, it goes from the start. Maybe they have to change the service enough that a whole new operational sort of review has to happen from DVSA again. If they, you know, change a little bit, then maybe those first few stages will be repeated quite quickly. But if a lot of changes are required, then there could be delays for another few months. the expectation that a lot of companies set out should be probably delayed by at least Two to six months at any stage right now, at least from the current regulatory environment that we seem to be having. Daniel Abreu Marques: And the applications opened in May, but as of now TFL said it had received no APS applications or consent requests for London. So why has nobody reached that stage yet, in your opinion? Tym: At this stage, the process is a little bit opaque. The evidence gets passed from the applicant to DVSA, and that is not public information as to how much information has been passed and at what stage before the TFL consenting stage are we at. And TFL did not even have to admit whether they received any applications or not, right? Officially, legally, that's also a fairly, you know, not involved with the public process. Why haven't we considered it? I think we would imagine that the operators want to have an application that passes. They want to make sure for 100% that they get the consent and they're happy and they will get through. And in the application process, there are stages which allow for consultation, for conversations with all involved parties. Meaning that legally it's probably fine for these operators to talk with TFL. beforehand, at least through certain channels. So they might be having these conversations saying, Hey TFL, this is kind of the idea of what we would w want to submit to you. Are you happy with that? And then TFL might say no, might have already said no, who knows? And maybe that's why we haven't seen a full application formally, because informally there's not yet certainty that there will be that go ahead. And any rejection will definitely delay things for, you know, at least a month, if not multiple. So it's the operators being cautious and maybe they themselves aren't yet ready to even submit an application. That's another explanation. So once again, there's no set answer here, but there's some hypothesis we could set up. Daniel Abreu Marques: Yeah, makes sense and also this six week clock this probably only works if companies know what TfL wants to see more or less and Tym: Look, this probably only works if companies don't want to T F L wants to say Daniel Abreu Marques: TFL has also not published it consent guidance so to speak. So in July the director of strategy said TFL was not in a rush. So what is the practical effect on operators if they don't have like this this content guidance? Tym: Yeah, the consent guidance isn't a legal requirement. I believe this was some industry expectation, you know, quite publicly, maybe some industry players. And here I don't know, but one hypothesis is, Hey, TFL, can you like write publicly? Like what do you want? Because we understand that London's a very complex city and that you want to make sure that public transport that TFL operates is not disrupted by autonomous vehicles. For instance, it's currently unclear whether Current taxi lanes or current bus lanes would be allowed for autonomous private hire vehicle services. Some private hire vehicles occasionally in the past get to use bus lanes and these limited roads, but will the autonomous vehicle be allowed that? That should be publicly written, because currently it is publicly written for Uber and Bolt and these other service operators. The fact that TFL hasn't published this, and they've had Quite a few, you know, months to consider this, or over a year really, might suggest that there isn't enthusiasm, especially with that comment as well. certain industry observers would comment that maybe TFL doesn't want these services to be happening right now. And they're using all these different mechanisms that are legally formed or that are process formed, but not statutory, to make this a little bit slower of a process. they ultimately maybe believe that politically they can't stop it this forever, but actually legally they can actually delay this for a very long time. and ultimately for London forever, because they are the consenting authority. politically, however, there would then be other considerations that maybe the DFT would want to place. The Secretary of State has other ways of influencing TFL outside of this formal process. So if the Secretary of State really wanted these services to happen. there might be other pathways. Obviously we're very, very far from that sort of world. Daniel Abreu Marques: And another interesting point is that the chair Tym: And another interesting point is Daniel Abreu Marques: of the London Assembly Transport Committee said that the central government has stepped on London's devolved powers. So is there a real tension between the Department for Transport and Transport for London or is the system working as designed? Tym: is there real tension or is the system working as designed? You could say both. These are not either ors. I believe TFL is making clear through the different mechanisms and through the different bodies that it has influence over that the current process might be a little bit too fast or that they don't believe they have the power to define the service as they want. Right? Ultimately they might not. care about having the full consenting and permitting license, right? They might want to just say, hey, earlier on in the process, we want to say that there is some fixed requirements in the permitting scheme. But the demands haven't been made public. So we don't know why TFL is delaying processes or trying to delay processes or being perceived because we're not ultimately saying that they've delayed anything yet, because there's no clear evidence of that. But the industry might point to some tensions. ultimately, however, you know, these are individuals, they have the right to say the things in their individual capacity. Maybe there are no tensions at all. Daniel Abreu Marques: And maybe one last question to round this up. So I want to to look into the future with you and your crystal ball. So please predict for us the first APS permit in London. So which company, roughly when, and what will it look like? So how many cars maybe, what Tym: So how many curves? Daniel Abreu Marques: area? So give us your future read. Tym: Yeah, so two notes first. First, the Institute for Driverless Transport has no preferences for any commercial success in any of these companies. we are entirely neutral players, but we wish all of them to be successful and beneficial for British society. and secondly, I have written some predictions in the past at Self-Driving Insights. you can find predictions for 2026. Self-driving Insights 2026 predictions is the post. I recommend checking that out. But with that out of the way, my if I were a personal betting man, what does Tym believe? I believe Waymo has a pretty good chance of being the first to have a license. There's no reason for them to not use almost every single car that they already have available. They have a hundred cars testing in London. So I would expect 80, 90 cars, or even as many as 200. I think that's the range. I could see 200 cars being in the first license, or maybe even 250, or 80. That depends if Waymo feels like they want to ship more cars. If there's a very big commercial pressure, if Waymo believes that Wayve will get the license two months later or a month later, I would expect more cars. But if there's less just the same amount of as today, then you could probably imagine that the next APS Permit doesn't get granted because we care about market share here and we care about brand, right? So pace of deployment, Waymo will set their resources accordingly. After all, they have lots of cities in the United States where they have competition as well. so what else? Where? I publicly made this this prediction. It's gonna be North London, unfortunately, over South London. I say personally, unfortunately, because South London has less transport accessibility in the first place, a lot less tube stations, more buses, but the buses have less capacity overall than tube stations, tube rides. but there's more money to be made in North London. there's more offices, there's more high-paying jobs, and there's also more demand for existing private hire vehicle services. So Uber makes more money. in North London and north central London specifically, north of a river, I mean. So I'm expecting places like King's Cross, I'm expecting places like Holburn and St. Pancras, perhaps even Westminster. it would help in convincing more politicians that this is a very important, reliable service, because maybe the politicians would start using it. but also ideally the city of London, I think unlikely to see Canary Wharf, despite the fact that there's quite a lot of high paying jobs there, partially because the road density isn't great, and also it's quite far from depots. A lot of depots are currently being set in northwest London, around Parks Royale. So you also do not want services, Waymo, starting closer to Greenwich and closer to Canary Wharf, because getting back to the depot would be a long one hour and a half or two hours in traffic drive, which would be quite wasteful in electricity and time. So My prediction eighty to two hundred and fifty cars, North Central London, Waymo. Daniel Abreu Marques: thank you very much for Tym: prediction, so Daniel Abreu Marques: that prediction. We will keep you on that. So I we will track it and then yeah, revisit your prediction Tym: We'll see if my name Daniel Abreu Marques: afterwards. Tym: benefits or will get tarnished, but I believe more people should make predictions. and we can only become more accurate in our beliefs once we see our predictions fail or succeed, so then we can become more wise for the next prediction. Daniel Abreu Marques: we're at the end. Thank you so much for taking the time. This was very insightful and I highly recommend everyone checking out his substack and also his post about the permits and also the post about his predictions. And yeah, Tim, thank you so much again see you soon. Tym: Daniel, thank you, it's been an honour