Discussion session at Traffex organised by ITS United Kingdom.
Participants: Jo White, Highways England; Andy Graham, White Willow Consulting; Pri Boyd, Siemens; Darren Capes, City of York Council; Ian Patey, WSP
Report by Jennie Martin, ITS (UK) Secretary General
The opinions expressed are those of the individuals concerned and not necessarily those of their employing organisations
The panellists made opening remarks:
Jo White explained that she was working on a number of CAV trials for Highways England, enjoying working on the topic and pleased to be working in the field at this interesting time.
The CAV work should be seen in the context of safety being Highways England’s top priority and their KPI of reducing numbers killed or seriously injured on their network by 40% by 2020. CAVs are key to delivering this target. HE is working with vehicle manufacturers on trials in the Midlands on the UK CITE project with JLR and Siemens, and also on platooning with DfT on the M2 A2 corridor. Again, safety is key, and possible infrastructure changes are part of the research.
Andy Graham said that he recommended thinking in terms of connected and pause … autonomous vehicles. More work is needed to fully understand the benefits of connected vehicles. Vehicles with the capacity to act in a connected way are already present in our fleets and more are being added every year. We can do a lot with connected vehicles in order to realise a wide range of benefits.
Pri Boyd is product manager for connected mobility with Siemens UK. She added that not just infrastructure but also traffic management systems need to be linked into the utilisation of connected vehicles and all aspects of autonomy. We should be focusing on connectivity, not just in new vehicles but also in our existing, older fleet. Pri has also worked on the UK CITE connected vehicles project in the West Midlands and works with the teams around the Vienna to Rotterdam connected corridor project, and on the Newcastle University and Newcastle City Council collaboration in the COMPASS4D project. She stressed the importance of partnership working to making the best progress in this field. She believes that more progress has been made in the last two or three years than in the previous ten.
Darren Capes represents the IET in thinking about how technology can be used in transport, and is also Transport Systems Manager at the City of York Council. He explained the Local Authority view of connected and autonomous transport technologies. He was overseeing some interesting work which would involve trials combining connected vehicles with urban traffic control on a site in York. It would be a priority for this project to get more data from vehicles. This is probably cutting edge in terms of what Local Authorities are doing. There is a gap between what LAs are doing very well now and their skills for instance in terms of traffic management, and these coming technologies and the capabilities of the devices citizens routinely carry around in their pockets. The reduction in funding in the public sector has led to deskilling, early retirements of knowledgeable staff and redundancies. LAs are not as well equipped now to write the right briefs and buy the right kit for these coming new technologies as they would have been a decade ago. This is a big challenge as the Department for Transport and the OEMs encourage a move towards a CAV future – how will LAs cope with this, with possibly no in house understanding of CAVs and no funding to recruit experts to handle the necessary work?
There followed a general discussion of LAs and CAVs including input from the audience.
One simple barrier to CAV deployment can be that the current vehicle models, such as Teslas, need white line road markings in good condition in order to function properly. However, authorities also need to keep doing the day job and serve all sorts of vehicles and road users, not only new connected ones. Good road markings and traffic signals are actually useful to all road users. Highways England are taking the opportunity offered by their trials to include work on line markings and road sign in the autonomous vehicle context, and they are also working vehicle manufacturers (OEMs) on possible improvements to headlights to assist optical sensors. However, relying on these is fundamentally unsafe in conditions of very low visibility such as heavy fog or blizzards. We also already know that the speed roundels carried on the back of some foreign HGVs confuse AV systems, who read them as speed signs which apply to them.
Cars will not necessarily always rely on the road markings – some projects are looking at using digital maps instead. Other approaches not currently under consideration will perhaps also develop over the next few years. As automation gets more advanced, we may opt for using a combination of maps and sensors and maybe future new technologies to guide our AVs. This requires data to be constantly updated. Vehicles will probably eventually use a whole suite of technology to navigate. It was felt that in the area of speed limits, having accurate digital map data including speeds would be superior to relying on individual vehicle systems reading speed limit signage.
Some of the glossy and ambitious video clips produced by those working on CAV trials are unrealistic since they do not show how to manage the coexistence of these vehicles with pedestrians and cyclists who from the LA point of view will continue to be very important.
Having virtual traffic signals and other forms of traffic management would be excellent from the roads authority point of view, but we will always, or at least for a long time, have to serve unconnected road users and accommodate them with signage, line markings and traffic signals. This diversity in the fleet will continue for a long time if not for ever. It is quite unlikely that there will ever be no non-autonomous vehicles on our roads. In any case, the average age of vehicles in the UK is just under eight years, so there will be a long transition period. People working in standards groups on line markings and road studs know that the CAV context is now important, but are finding it hard to engage with OEMs regarding this even though a dialogue would be very useful to both sides.
The smart motorways electronic signage is designed for human beings and uses pulsing LEDs which are currently accepted as the best way of communicating visually with human drivers. However, sensing systems are better at reading fixed signage. Vehicle speeds and weather conditions also have significant impact on existing camera technology. Highways England’s demonstration project on the concept of feeding signage information into vehicles demonstrated that it is quite feasible to do this. The UK CITE project is also working on this concept, with all information provided via gantry signage converted to feed into cars including into route planning services. The first step is to get messages into the vehicle and then the second step is to translate this into route planning and advice.
A new issue was raised regarding how AVs will interact with road works. An example was quoted of a Tesla which kept trying to overtake a rolling road block vehicle. How do we integrate AVs (which are not connected) with road works? The OEMs see the responsibility to lie somewhere between themselves and road operators. It could be that the responsibility gets allocated to the roads authority to provide correct data about the road works, rather than to the vehicle manufacturer / sensor supplier to enable the vehicle to directly detect the road works.
Another issue which needs resolving is how AVs will interact with police officers directing traffic. They need to be able to drive the wrong way along a one-way street if the emergency services so direct. This could be resolved by using V2V communications instead of I2V or sensors in the vehicle. Similarly, AVs must interact appropriately with emergency vehicles, another probable scenario where V2V will be most effective.
The uncertainty in all these cases is caused by the sensors and software. We need to engage with their manufacturers in order to find out what road operators need to provide to suit their equipment and this engagement is somewhat slow in happening.
Maybe we could restrict ourselves to improving the passive network enough to support CAVs in order to save money, instead of completely instrumenting the infrastructure? On rural roads it is particularly unlikely that creating an intelligent infrastructure would be cost effective.
The Department for Transport’s code of practice for AV trials was noted regarding its approach, that of providing advise regarding holding trials and suggesting a safety risk management system. It was agreed that this was a suitable approach to an important issue. The results of these trials in terms of technical successes and failures will be invaluable in terms of making progress, but so also will outcomes regarding how to manage and minimise risk. It is very important that publicly funded trials such as those currently underway in England with funding from DfT or Innovate UK are required to report publicly, comprehensively and in a timely fashion. This will help the UK progress in this field where there is strong international competition.
The vehicles themselves can generate data and share this, including reporting changes in road layout and condition to whoever needs to know. There is no reason why data generated by vehicles in this way should not feed directly into digital maps. This will definitely be faster and probably also more accurate than sending a person to survey.
We do not have to rely on the latest most expensive vehicle models to start to work in this way. By using currently available dongles in the OBD 2 Ports which are normally present in all vehicles from the late 1990s onward, we can already collect and transmit a lot of useful data. This could be verified / augmented by six monthly surveys.
For example, local authorities could use rubbish collection vehicles as probe vehicles to report pot holes and other defects. These vehicles travel the exact same routes on the same days of every week so such surveys would be very regular.
It is becoming increasingly important to give proper consideration to how we analyse data, with clear objectives in mind, since we are fast approaching a situation where volume of data has the potential of simply overwhelming us unless we start being selective about what data we process and why we are processing it. We also need to come to a consensus about who owns data – the generator or the collector? – and who has what rights over it. The automotive industry thinks of the vehicle data as belonging to them and intend to commercialise it for their own benefit. But the vehicle owner, driver or passenger may have other views.
Privacy concerns are not always logical or aligned to technical realities, and addressing them can have unintended consequences. But they still have to be addressed in a democratic society.
It is a fact that anybody using Facebook, Twitter or other social media platforms is already publicly broadcasting where they are and what they are doing. In addition, using google, trip planners, digital map services etc tells your service provider where you are, and the network you use at home or at work or while making a coffee last two hours at Starbucks also locates you geographically as soon as you do anything on line. The clever work done by for instance Transport for London to work out the actual start and end points of the journeys made partly on their public transport network shows how a partial set of location data can be combined with other data and accurately recreate a complete pattern of movement. But none of these facts alleviate concerns citizens have about being “tracked” while travelling by car.
In real life, just looking at somebody’s smartphone without using any fancy analytics will tell you nearly everything there is to know about them. Commercial transactions often harvest unrelated data to capture you. We need to ask the right questions – what is reasonable, what are your intentions with the data, what is reasonable and unreasonable surveillance? This is all about information. If you explain why you want the data and how you plan to use it, then people can take an informed decision about whether to share it.
Local Authorities would like reasonably priced access to useful private sector data. Data formats need to be better standardised in order to be useful; this is probably more important as a barrier to effective use than who owns the data.
It is not a question of connectivity versus autonomy. Connectivity must be sorted out first if autonomy is ever going to be realised. Completely independent, self sufficient autonomous vehicles are certainly a technical possibility, but whether they would really be an overall improvement on the vehicles we have now is open for discussion. Connectivity, on the other hand, promises a steady development towards safer, cleaner, less congested and more accessible network. This evolution is also likely to end in autonomy but it will be connected and interacting in a beneficial way with both other vehicles, the infrastructure, and other things you may want to link it to, such as schools, hospitals, large centres of working or retailing, etc. In order to realise this scenario, the IT, automotive and roads industries must work together.
There is an image problem in that (at least according to this gathering) connectivity is more important but autonomous is perceived as more cool and interesting. Connectivity is not new, our current discussion is just about extending what we can already do in this field into the car, which is nowhere near as exciting as automated vehicles to non-experts in the field.
Since we expect to rely more and more on continuous connectivity, we need to agree the detail of who is responsible for delivering seamless and continuous connectivity. This surely can only be whoever owns the network, i.e. the mobile comms provider or similar organisation. A city or a roads authority may choose to take on this responsibility in order to support what they view as essential services, and then the responsibility passes to them. In either case, the responsibility must be accepted, clear, and known to all parties.
In cities and towns connectivity is getting better in any case, but in some rural areas it is likely that authorities such as Highways England will have to step in if they want to realise the potential of CAVs. 5G will be partly about the vehicle and partly about cellular. Some areas of the UK do not even have medium wave radio reception. Another way to think about this is to say that coverage really does not have to be universal, as long as the public accepts the need to take some responsibility for itself. As one delegate put it: “Don’t just sit in your Tesla watching everything go pear-shaped.”
Vehicle manufacturers have always been in the business of selling vehicles, but now we can sense a shift to selling a service – mobility – which should make it easier for them and road operators to cooperate and share data, for instance. OEMs also make more money from parts and services than from vehicle sales, so they want a relationship with their vehicle owners and they want to know what they are doing with their vehicle. This creates a valuable revenue stream for the OEM. We can see from advertising how this is changing – the abundance of information about financing has disappeared, and now marketing is more about connectivity and services. This is what now attracts people to new vehicles. In technology terms the difference between higher and lower end vehicles is not that big. So you need to differentiate on other areas such as services, since the vehicles are so similar.
Smart parking is an interesting if fragmented area. There are many operators, many parking apps and many car parks. So a vehicle manufacturer can sell an attractive service by simplifying this – leasing a car with parking included – which may give them a commercial edge. Parking is the undiscovered country as far as mobility is concerned. It is a big pain point for motorists and making it a service would be very attractive.
Parking for AVs is also an area needing more research. How will they find their way in multi storey car parks where satellite services are unavailable? How will they park in cities and towns? What happens when they are required to park in usually off limits locations like fields at events, where they need to leave the network they are programmed to use? The general view is that AVs will free up masses of city centre land since they can park elsewhere. But this needs more work in detail.
If we revisit the film Blade Runner which was set in 2017, we are reminded that it featured driverless cars but no mobile telephony, which tells us that predicting the technical future is hit and miss. We now have ubiquitous mobile telephony, to the point that the rural areas of developing countries are leap-frogging fixed telephony entirely, but we do not have automated vehicles in public use.
Let us not be bound by how we do things today, or we will miss opportunities. Change is fast and there are lots of opportunities – we just need to assess them accurately and without bias. Yes we are nearly there yet. ◆