Search

Safety by automation – not autonomy?

stein@opensky


I have been following with interest the vision of connected and autonomous cars for a while – and I recently participated in a couple of webinars addressing intelligent mobility and various degrees of autonomous vehicles, the latest one on 17 Sep being “V2X - A Roadmap for Saving Lives” from Reuters Events. The GSMA also has had various activities in the area for a while – and also had a similar webinar on 1 July on C-V2X, which is a set of (cellular) interfaces standardized by 3GPP. From a technical standpoint, a lot of communication interfaces for V2X have been defined, be it vehicle-to-vehicle, vehicle-to-infrastructure, vehicle-to-network, vehicle-to-pedestrian etc.

It has amazed me how great the hype is around autonomous vehicles in the context of 5G – something I commented on as one of the potential 5G use cases already in a previous article “5G will save the world – or will it?”. Most om my comments in that article were really around Level 5 autonomous vehicles, i.e. fully autonomous L5 “steering wheel optional” – which seems a very long way away for me – or which might never happen (?) When that is said, there are various lower levels of automation that will be very useful.

If I go back to the latest webinar, while the topic was really about how V2X could improve people safety in various ways, my key take-aways were really 1) that the whole area seems to be technology looking for a use case – and 2) that business models are totally unclear (at this stage).

In the webinar, a poll on believed availability of V2X technology as standard fit in cars in main markets, indicated a timeframe of 5-10 years from now. Just today, I also saw an article in a Norwegian technology magazine forecasting “at least 20 years until full autonomy” (see also the reference in my earlier article showing that Norway is doing live trials with autonomous buses e.g. in Trondheim). These predictions may be true or false, but car manufacturers and the wider automotive industry have been implementing automation into their vehicles for years already - and are continuously developing enhanced functionality for their vehicles. Further, the 5G Automotive Association (5GAA) has a very wide membership including mobile operators, car manufacturers, telco network suppliers, device manufacturers and the wider ecosystem – all working hard to make V2X a reality – and if I might add, it seems mostly from a technology perspective (at this stage).

From a business perspective, it all seems very unclear, however - and not only from this webinar. For the full extent of V2X to become a reality, many things need to be in place:

  1. For V2V communication to be useful, there needs to be V2V technology implemented in most cars – so a network effect is needed depending on the number of cars actually having implemented the technology.

  2. For communication with the mobile network (i.e. V2N), there needs to be base stations covering all the roads – and for it to be useful, the coverage needs to be very good. Depending on the broadband needs and spectrum in use (i.e. sub-6 GHz or mmWave), the number of base stations needed will vary significantly, but in any case a very large number of base stations for road coverage need to be in place – and there should not really be many “notspots” – and mobile operators are the ones needing to build out these base stations. See also my previous article with practicalities on mobile network buildouts.

  3. Communication with the road infrastructure (V2I) also needs to be built out – i.e. a task for the road operators. In cities or along major highways, roads are likely to have lamp posts or similar that can be used for this purpose, but along minor highways or rural roads, there might be no such street lighting at all. In any case, it will be a major challenge and a huge cost to implement such V2I infrastructure and systems for any road operator – most likely governed by national budgets. A potentially simpler task, however, would be to implement functionality in traffic lights – to warn cars about upcoming intersections, red lights etc.

  4. In all V2X cases, very low latency may be needed to ensure the needed safety – and in any communication beyond V2V this may potentially require edge computing. This would clearly apply for 5G V2N and probably also for V2I – and someone needs to provide the functionality.

From a commercial or business perspective then, there seem to be several chicken-and-egg problems – and in many cases, good technological solutions are often inhibited or totally stopped simply because commercial interests or business models from different players conflict with each other (check also out this article on business and “technology” wars) – or because there is no funding.

Mobile operators will most likely not commit to building out coverage along all major roads to the degree that is required for useful V2N capability unless there is a business case for them in it – or a regulatory requirement to do so – but then the requirements must be clear – and the SLAs, accountability for safety and much more. In any case, “nationwide” rollout of 5G will take years – whatever “nationwide” means.

At the moment, car manufacturers seem to be willing to implement V2X functionality in cars on a proactive basis, however, will they make it available as safety tools if there are not 1) sufficient mobile coverage or 2) a sufficient base of other cars in the market with the same functionality? They might implement it – but with no guarantees. For road operators to implement large amounts of roadside infrastructure for V2I, I can only imagine the planning periods, budget discussions and time frames required to get this done. It will probably never happen nationwide in any country, however, there could be something on traffic lights though (?)

Enough of the party pooping. On a more positive note, V2X promises increased safety – and the GSMA indicates a yearly benefit by 2025 from connected cars of 11,000 lives saved and 260,000 fewer accidents. Having worked with health & safety for five years of my life, this sounds very promising.

The automotive industry has a roadmap for implementation of V2X in vehicles based on three levels of automation, i.e. “Act, Warn and Inform”, in that order. “Act” is mostly based on on-board sensors and is to a large degree available now. As an example, my current Volvo can detect change of lane, cars in front or on every side, objects behind etc – and does e.g. intelligent cruise control, parking assistance and a certain collision avoidance, based on this functionality. This is all good, but I don’t use all of it.

The next level “Warn” is for risk mitigation and depends on implementation of V2V and V2I functionality (in my view, probably mostly V2V) – and examples of this might be more of the same – plus enhanced communication between cars – and potentially also with traffic lights. The final level “Inform” is for risk avoidance and requires V2N (and V2P?) functionality, e.g. information about traffic jams, car incidents, road works, pedestrians along the road etc – and the combination of all V2X may be used for autonomous driving – if it ever happens.

To improve safety and saving lives, V2X has great potential and can be gradually implemented and taken into use. There will, however, be many challenges along the way. The easiest bit will be for car manufacturers to implement functionality in cars, however, they will not be able to make any guarantees – simply because the effectiveness of the functionality will depend on various other parties in the ecosystem, be it mobile operators, road operators, other car manufacturers, the number of consumers having bought cars with the same functionality etc. V2X may be used as sales arguments for car manufacturers – but the impact on society will be very gradual and for the long-term.

As soon as V2I, V2N and V2P come into the picture, the automation concept will go beyond car manufacturers and the whole ecosystem will get more complex. There will be a need to define services and there will need to be some form of Service Providers – and there may be different types of services - possibly fully autonomous driving at the end stage. In my earlier article I commented on Level 5 autonomy and who would be the Service Provider in such a case. I am not sure a mobile operator (or anyone) would be able (or want) to take on such a role. The liabilities of a Service Provider in such a case could be enormous and are yet undefined. Who will be liable if an autonomous car crashes and kills someone? The driver? The car manufacturer? The Service Provider? IS there a driver – or is there only a passenger? Does the person in the car need a driver’s license? By the way: Who shall the car hit if there is a choice? etc …

Going for Level 5 service provision is probably a very long way out in time - if it will ever happen. There will probably also be a large number of regulatory questions to sort out. It might be easier to offer other services combining different V2X functionalities – but without safety guarantees. A question still remains for me, however, on who would be best suited as a service provider in those cases. Could it be the mobile operators? The road operators? A government agency? Someone on behalf of the automotive industry? … or some form of new player? It is not clear to me. It will of course depend on the services provided – and the business models and regulatory framework around it.

In principle, governments would like the concept of improving road safety though vehicle automation. There could also ultimately be various other societal benefits like improved traffic flows, reduced congestion, reduced CO2 emissions and more. However, there is a risk it might all stop when it comes to budgets and business models eventually. I hope not – but it seems difficult!

To summarize, intelligent and connected cars are in the pipeline. However, it might take a very long time before the ultimate vision of fully autonomous cars is a reality (e.g. 20 years as referenced above) – if it will ever happen. Automation is, however, being introduced in cars on an ongoing basis. Personally, however, I would not trust a fully autonomous car (yet), like many others – as I need to feel in control. V2X (and most likely C-V2X) will be important for road safety going forward – but I am afraid the promise around 5G and autonomous cars is overhyped. I believe, however, that the main safety benefits from V2X will be through automation – not autonomy. Autonomy could actually provide the opposite.

On the ecosystem part of this, even if I have highlighted many of the challenges around V2X, with automation and specifically autonomous driving, it seems clear that quite a few players are betting on it, mobile operators and others. Vodafone seems to be one of the operators with a keen interest in it – and they of course put themselves in the centre of the ecosystem somehow. I am sure that also other players would want to take on roles in this ecosystem – and it will be interesting to see how it plays out – if I am still around …

25 views

© 2019 by OpenSky Consulting AS.