Reinventing the wheel: Distributed Ledger Technology as Enabler for Autonomous Vehicles
Roughly twenty years after the first automobiles appeared in the streets of the early 1900's, the business model of horse-drawn carriages, lasting for thousands of years, was completely disrupted.
Fast forward 100 years and the automotive industry, once disrupting global means of transportation, is at the brink of being disrupted — ironically — by itself.
Next to the current transition from combustion engines to electric ones, computational advances might make fully autonomous vehicles a reality as early as 2020. Irregardless of such possibly overly enthusiastic expectations, vehicle manufacturers are going to face dramatic economic and technological changes in any case, as described below.
While for some, autonomous vehicles may still sound like an episode from the Jetsons, this article isn’t about a far away future. It is a rather dry outlook, based on cause and effect, starting with the challenges and efforts of manufacturers we are witnessing today.
But before we get to why Distributed Ledger Technology is the most obvious solution for a lot of challenges mobility providers are facing, let’s first have a look at the challenges themselves.
The inevitable disruption of a business model lasting for a century: car sales
Nowadays, vehicle owners rarely trust strangers with their cars, fearing bad driving behaviour and accidents, resulting in cars being parked 95% of their lifetime.
As strangers would not drive autonomous vehicles themselves but be mere passengers, risk is reduced dramatically and would thus very likely enable a true sharing economy.
At the same time, „sharing“ of autonomous vehicles could become a new source of income: Elon Musk recently predicted that an autonomous Tesla could earn up to $30,000 annually, on its own, without any human intervention.
If correct, owners of autonomous vehicles could effortlessly refinance the acquisition cost of their vehicle within one or two years. After that they would purely profit from owning one.
Owning a vehicle able to generate revenue by itself, naturally would be much more attractive than owning one that only costs money, especially when taking into account that vehicles currently aren’t used 95% of their lifetime anyways.
Given their long development cycles, out of fear to become obsolete, nearly all major vehicle manufacturer currently invest heavily into the development of autonomous vehicles. They can’t wait because they wouldn’t be able to catch up. By joining in into the R&D craze they are raising the overall chances that autonomous vehicles will become a reality. Through their efforts they are therefore essentially digging their own graves:
Vehicles are currently only used 5% per day by an average of 1.4 persons. If automated robo-taxi’s would fulfil the transportation needs of just twice as many people (~3 people) per day instead, half the amount of cars would be needed and vehicle sales could get slashed by 50%.
A 50% decline in sales would be devastating. Yet the “shared” robo-taxi’s still would only have been used an additional 5% per day; and still remain unused 90% of the day.
Taking this further, if autonomous vehicles would be used 50% per day instead of just 5%, they would fulfil the transportation needs of 10x as many people as today, resulting in 10x less vehicles required – or in other words: 90% less cars needed, built and sold. Yet the robo-taxis would still remain parked 50% of their lifetime.
You probably see where this is going. F for vehicle manufacturers it gets even worse:
The prospect of making money while you sleep is a very powerful one. Once the first manufacturers start selling fully autonomous vehicles, it is therefore highly likely that the amount of robo-taxi’s being sent onto the streets by their owners to offer their services and make money will continuously increase.
Who wouldn’t want to earn money without having to do anything in return, assuming maintenance and expenses would be covered through the automatically generated revenues?
At the same time, an increasing amount of robo-taxi’s would also increase the competition amongst them in order to acquire passengers. The most basic economic laws indicate that where-ever competition increases, price usually decreases: Increasing competition would most likely lead to decreasing transportation costs.
As long as maintenance and acquisition cost would be covered by the generated revenue, vehicle owners probably don’t care as long as they keep continuing to earn money effortlessly.
The decline of transportation costs would only stop at a level slightly above the acquisition and ongoing maintenance cost for vehicles. Slightly above it, owners would still profit from a robo-taxi. Below it, it would become uneconomical. As a result, enabled by pure greed, because who wouldn’t want to make money without raising a finger, passengers would eventually pay the lowest possible fares, slightly above the acquisition and maintenance costs.
Once this level is reached, transportation is close to a Zero Marginal Cost Society, as described vividly by economist Jeremy Rifkin e.g. in his talk about the “Third Industrial Revolution”.
At this point, constant and immediate availability of a vehicle would likely remain as the sole reason to own a vehicle for exclusive personal use.
But constant availability is a very luxurious reason if one requires a vehicle only 5% per day on average, especially when the exact same service is offered or 90% less.
Speaking of cost: what could it be?
Let’s do a back-of-the-envelope calculation:
- Vehicles are used an average of 5% per day, covering the transportation needs of an average of 1.4 people
- A personal vehicle currently seems to carry an average total cost of ownership (TCO) of 600–800 Euro per month
- If a robo-taxi would be used 50% per day, it would thus fulfil the transportation needs of 14 people
- 800 € divided by 14 people is ~57 € per person and month
While this calculation is anything but sound, it isn’t outlandish either.
Even if the cost would be 30% higher, for example to cover higher maintenance expenditures, ending up at ~80 € per month, it would still be cheaper than a monthly metro pass in most western metropolitan areas. And most importantly, it would be up to 90% less than owning a personal vehicle.
Expenditures for personal transportation of 800 € versus 80 € per month is again bad news for vehicle manufacturers: At these price levels, probably only very wealthy individuals would consider the luxury of owning a vehicle.
But the few cars they buy wouldn’t make up for the millions of vehicles currently sold to anyone in need to get from A to B once in a while.
Hence, if only a fraction of today’s volume would be sold in the future, business models of vehicle manufacturers currently profiting from selling hundreds of thousands or even millions of cars would completely change. They are therefore in dire need of finding new revenue sources.
In summary:
- Autonomous vehicles would enable a sharing economy by excluding the human risk factor
- Given their development cycles, manufacturers can’t wait with R&D because their competition is already well on their way.
- Cars making money while you sleep will be preferred over those who can’t, leading to less “dumb cars” being sold.
- If offered as robo-taxi‘s, much less vehicles could easily fulfil the transportation needs of many, leading to less vehicles being needed overall.
- Less people would want to own a vehicle for exclusive personal use due to dramatically reduced transportation cost, a result of an ever increasing robo-taxi competition.
Let’s just continue to sell dumb cars
Manufacturers are currently caught in a prisoners dilemma (see this article by Daniel Stricker for a nice description)
- If buyers would favour vehicles enabling them to effortlessly generate a profit, any manufacturer that doesn’t (or can’t) offer autonomous vehicles most probably would be pushed out of the market by competitors that do offer them
- If manufacturers do offer autonomous vehicles, one vehicle would fulfil the transportation needs of many, thus resulting in less vehicles being sold overall, while simultaneously leading to overall decreasing transportation costs, which in turn again leads to less vehicles being sold in total
The dilemma of vehicle manufacturers is that, irregardless of the decision of any individual manufacturer (wether or not to invest in the development of autonomous vehicles), the impact on sales will be dramatic for each and every one of them if just one manufacturer succeeds.
In order to sustain their business, by researching and eventually developing autonomous vehicles, manufacturers are spiralling into destroying their current business model. Austrian economist Peter Schlumpeter describes this as “creative destruction”:
“a process of industrial mutation that incessantly revolutionizes the economic structure from within, incessantly destroying the old one, incessantly creating a new one”.
Manufacturers basically entered a race to the bottom. It’s not a question whether, but when autonomous vehicles will become a reality. Ignoring them will most likely lead to the downfall of any vehicle manufacturer.
The coming decade will most likely represent the biggest disruption any industry lasting longer than a century has ever seen. The New York Times recently titled “The Car Industry is under siege”:
“It’s going to be the biggest change we’ve seen in the last 100 years, and it’s going to be really expensive even for the biggest companies” […] Major auto companies will spend well over $400 billion during the next five years […]
They must retool factories, retrain workers, reorganise their supplier networks and rethink the whole idea of car ownership. […] this upfront investment is a matter of survival. If they don’t adapt, they could become obsolete.
The saucy questions we are currently left with are who will offer autonomous vehicles first, who will be too late to the party and what will the impact on our society, economy and environment be?
Now let’s have a look at the requirements of autonomous vehicles and why Distributed Ledger Technology is especially beneficial to manufacturers.
Data isn’t just the new oil, it will replace oil.
Autonomous vehicles produce a massive amount of data and require a constant stream of reliable, external data. The need to properly secure this data is beyond doubt. Here’s why:
In order to serve their purpose well and solve tasks autonomously, especially while humans are unable to intervene, as vividly demonstrated with the latest iterations of autonomous vehicles that don’t even feature a steering wheel, autonomous vehicles will have to constantly communicate with external endpoints.
- „Which route to take to my destination?“
- „Where is the next charging station with an available spot?“
- “Where can I find a free parking spot?”
- „Make way for an ambulance coming up!”
- “What is the (credit) history of this potential passenger?”
- “Caution, there’s a totally wasted human in a Winnie the Pooh outfit stumbling alongside the road ahead”
These are just easily relatable examples. The actual amount and frequency of environmental data required and generated by vehicles already exceeds several terabytes per day.
If malicious actors would find ways to tamper with data used by autonomous vehicles, “bad data” could e.g. lead to people being trapped in vehicles driving around in circles because the car thinks all other roads are blocked, or could lead to vehicles circling because they don’t find a parking spot, or them having to wait with a near-empty battery in a line on end at a charging station until it‘s finally their turn to charge up.
But unless Winnie the Pooh decides to take a nap in the middle of the road and is accidentally being run over, probably none of the above examples would be life threatening for humans.
Nevertheless, „bad data” would most likely lead to a high degree of inconvenience for autonomous vehicle owners as well as passengers, and therefore for manufacturers — because no-one would buy an autonomous vehicle if it is incapable of fulfilling its purpose reliably.
To safeguard the business model of “vehicle sales”, manufacturers are therefore in need to find solutions ensure the authenticity, origin and integrity of data.
Distributed ledger technology provides manufacturers with the opportunity to outsource security without having to develop and maintain their own, proprietary, non-interoperable infrastructure.
While it would of course be possible to rely on centralised setups like e.g. public key infrastructures, interoperability with external endpoints would be a nightmare to set up and to maintain (see further below) and therefore be much more expensive. And with dwindling sales, cost is an extremely important factor in their calculation.
Additionally, any proprietary system would automatically represent a single point of failure and therefore carry a huge red bullseye on its back:
A successful hack of a centralised architecture could potentially render all vehicles of a particular manufacturer useless for days, weeks or even months.
If relied on distributed ledger technology on the other hand, it would require individuals with malicious intent to successfully attack each car individually, or the DLT itself.
This was about low risk data. But what about high risk data?
High risk data and and high risk communication
When it comes to high risk communication and data in autonomous vehicles, the need for proper security is beyond doubt.
Finding patterns and algorithms for any imaginable situation an autonomous vehicle could run into is hard because those potential situations are infinite.
Letting every vehicle learn to deal with any possible situation itself would take very long while leading to unnecessary human casualties.
Having decisions made centrally while a car moves, for example in a data centre of the manufacturer, also most likely isn’t an option: Network latencies, even on 5G, are likely too large to send data about an unknown situation to a central service, have it processed, sent back and executed upon accordingly when driving at speeds of a hundred miles per hour, or faster.
The risk of vehicles having already crashed into a brick wall before a decision on how to react could have been communicated back to it is just too high to take.
And of course there also can’t be guarantee that vehicles would always be able to reliably communicate with the manufacturer in order to retrieve instructions on how to handle life-or-death situations.
Furthermore, if a constant and reliable data connection would be a requirement to enable autonomous vehicles, disrupting cell tower connectivity would probably become the easiest target for means to disrupt society:
Bringing a large area to a complete standstill via autonomous vehicles blocking every single street by cutting a mere data cable of a cell tower could be as cheap as a 20 € purchase of an axe.
Despite popular belief, autonomous vehicles will most definitely not rely on constant 5G connectivity. Autonomous vehicles will be autonomous.
The best approach to tackle the learning challenge of autonomous vehicles seems to be the one-to-many approach, also known as crowdsourcing:
Any new situation encountered by any individual vehicle is communicated back to the manufacturer, analysed in order to derive appropriate patterns, algorithms and directives which are being distributed back to all vehicles or even sold to other vehicle manufacturers or other interested parties (insurers, municipalities for traffic management, etc).
Any vehicle would thus “learn” quickly on how to identify certain situations and retain directives on how to react appropriately. By learning through the experiences of their brothers and sisters, vehicles would become exponentially smarter.
Yet with all knowledge devised and retained by the manufacturer, waiting to update a fleet of autonomous vehicles with new insights, patterns, algorithms and directives until their next physical inspection or maintenance cycle would potentially threaten human life. Wo would want to board an autonomous vehicle if it hasn’t received the latest “learnings”?
Any new information has to be distributed to all vehicles immediately in order to reduce risk and in turn potential liabilities. Autonomous vehicles will therefore receive a constant stream of over-the-air (OTA) software updates from the manufacturer or even third party “artificial intelligence maintainers”, or even insurance companies.
This data communication poses the highest risk. Not properly securing the authenticity and integrity of newly derived patterns, algorithms and directives being distributed to all vehicles would make autonomous vehicles an ideal target for cyber warfare or terrorism.
How would one stop a hacked, rogue, autonomous, 1-ton metal box on wheels going 150 miles per hour in the centre of a major metropolitan area, maliciously instructed to run over any human it can find?
Maybe with a rocket launcher?
In a worst case case scenario, a single, hacked autonomous vehicle could cause dozens or more deaths, hundreds of injuries and rack up millions in damages until it could be safely stopped.
Having to somehow stop a whole fleet of autonomous vehicles that are out on a killing spree, be it one type or even all models of a particular manufacturer, could take the authorities weeks or even months.
While low risk data like e.g. environmental traffic information might only lead to inconveniences, the need to secure any software update impacting the AI ensuring the safe operation of autonomous vehicles is beyond doubt.
Vehicle manufacturers can neither risk to damage their reputation (and thus future sales) nor would they probably be able to sustain the liabilities following a compromised fleet of autonomous vehicles.
Hence, they will integrate the most secure technology helping them to safeguard the logic retained in vehicles manufactured and sold by them.
Furthermore they require a solution for an immutable audit trail as autonomous vehicles don’t file a police or insurance report after an accident occurred. Instead, any accident would most probably automatically trigger a forensic data analysis. Excluding the question of whether logged data might have been altered, a solution proving beyond doubt “what happened at what time”, would thus be very beneficial to them.
An immutable audit trail stored in a distributed ledger offers exactly that. Manufacturers only need to equip their vehicles with a secure, tamper-proof element guaranteeing the origin of the data. And those are already easily available.
Given the cost of issuing transactions on public DLT’s, logging 4 terabyte per day and vehicle would amount to several billions of dollars annually for every vehicle manufacturer. Using DLT to distribute OtA software updates and/or create an immutable audit trail, inputs, outputs and decisions made by every vehicle would therefore only be feasible if transactions on a DLT would be cost neutral.
Given its fee-lessness and ability to communicate data, it therefore isn’t surprising that, among the thousands of different DLT’s, IOTA is currently the one being at the center of attention by the majority of global premium vehicle manufacturers.
Autonomous vehicles will need to become their own economic actors, in need of their own wallets
“Oh, I am sorry, I was having lunch and wasn’t looking at my mobile”
Whenever an autonomous vehicle earns money as a robo-taxi for its owner, relying on human intervention isn’t an option. Individual owners and even companies wouldn’t be able to make decisions 24/7 in a timely manner, by for example clicking on an “accept that passenger”-button.
Vehicles will have to be enabled to make decision on their own. They will have to make offers to potential passengers, vet passenger histories (credit scores, prior disputes, damages, etc), settle fees or chargebacks for their services, acquire energy, pay for road tolls or parking, cleaning and maintenance services, as well as most likely hundreds of other transactions and scenarios.
Temporary loss of network connectivity or a human agent that is supposed to make decisions taking a mere biological break could disrupt the utility of autonomous vehicles and additionally put a huge burden on public infrastructure:
Waiting for a human to remotely recognise and click on a “pay road toll now”-button while the car blocks a toll booth simply isn’t viable as „autonomous“ vehicles would constantly block public infrastructure.
In order to enable vehicles to interact fully autonomously with their environment they would have to be enabled to settle transactions themselves. Unfortunately, cars are currently not recognised as a legal entity (which of course might change in the future). As of today, cars don’t get bank accounts.
Of course, they could be connected to the bank accounts of their owners in order to settle transactions. The necessary application program interfaces (API’s) that would be required for the large part already exist. But then again, laws and regulations that would allow it don’t.
Existing financial regulations in most jurisdictions are based on the principle that any transactions, be it directly or indirectly through an API, have to be issued directly or indirectly by its owner. Allowing an artificial intelligence of an autonomous vehicle access to a bank account belonging to a person or a company might actually be illegal under current law in many jurisdictions. But even if legal, there are still hundreds of unasked questions and unsolved challenges.
Who would for example be liable in case of a dispute for an incorrect transaction: the vehicle manufacturer providing the software of the vehicle or the owner of the vehicle whose bank account got cleared out due to a software error?
Any forays into questions like these probably require several years of research, studies and eventually new laws, which would need to be passed and afterwards interpreted by courts, as well as implemented and tested as new financial services by the according institutions.
To put the effort of such an undertaking into perspective, the timeframes most jurisdictions (except the USA) required to determine under which conditions it would be legal to allow credit card payments to be initiated through mobile phones (see Apple Pay, Android Pay, Samsung Pay, etc). Mind that this rather minor novelty was a mere a matter of storing credit card numbers, traditionally printed on a plastic card, in a secure element of a phone instead. It took the vast majority of countries several years to pass legislation and regulations allowing it.
So, if we started today to discuss whether an artificial intelligence should get its own bank account or be allowed to initiate transactions on connected bank accounts of its owner, while we are not even completely sure what the exact requirements would be nor what even constitutes an artificial intelligence, it could likely take a decade or more until it could come to first trials.
That might be too late for vehicle manufacturers, especially when taking predictions like the former one from Elon Musk into account, that (fully autonomous) level 5 vehicles might hit the road as early as next year.
Whatever the timeframe, any lack of clarity would eat into the sales of vehicle manufacturers trying to sell (or operate themselves) autonomous vehicles in need of an interoperable settlement solutions.
Luckily, cryptocurrency, while still being very young and ridden with challenges, is already at their doorsteps. While the legal status of cryptocurrencies by far has not been concluded in most jurisdictions, lawmakers are at least debating them and international bodies like INATBA are actively driving regulation forward.
It most likely will at least require several years until cryptocurrencies have been properly regulated. But the discussions of regulators revolving around cryptocurrencies at least are already well under way, while regulation of AI’s hasn’t even really started yet.
For vehicle manufacturers cryptocurrency might therefore be a potential solution, much closer to manifest itself than the question of the legal status of AI’s.
Interoperability, standardisation and open source vs. proprietary, closed source architectures
Ease-of-use and the overall user experience have a direct influence on marketability and are thus two key elements for successful adoption of any new product or service.
But as of today, customers trying to fulfil their individual transportation needs while being able to always select the best possible option would need to register with at least a dozen services in most metropolitan areas of the western world.
Why would one register with BMW or even acquire a “BMW coin” if no BMW is in sight? Would that person quickly register with Mazda, if only a Mazda is available? Or even hold a Toyota coins, just in case?
It’s pretty straightforward: Mobility providers able to integrate user-facing interoperable services would gain an advantage over their competition, whereas any provider trying to create a moat around itself would lose out, eventually.
Cost as the decisive factor: Proprietary solutions are expensive to develop and maintain
Today, customers are required to register with every single MaaS operator in order to be able to freely choose between different offers fulfilling their individual transportation needs. Adding offers of private robo-taxi owners to that equation would add several levels of complexity for settlement methods and a staggering amount of registrations/identity solutions, management of payment information, payment processing and settlement solutions, handling of personal data as well as data privacy issues coming along with that.
In addition, every individual manufacturer as well as any service operator (electricity, cleaning, tolls, etc) has to ensure interoperability with all other connected systems, meaning hundreds, thousands or even tens of thousands friction points — while every interaction of course comes with a fee. This is anything but cheap, eating into profits of mobility providers.
It is thus unsurprising that we can already witness MaaS operators being well aware of the requirement for consolidation, interoperability and standardisation.
While BMW and Mercedes, for example, were in dire competition for the better part of a century, they recently joined hands and consolidated their short-term rental services Car2Go (Mercedes) and DriveNow (BMW) into a new service called SHARENOW. Just ten years ago it was completely unthinkable that those two competitors would even talk to each other. But times, they are changin’.
SHARENOW is a prime example of consolidation and interoperability: Mercedes and BMW now both save cost by sharing one infrastructure, backend, mobile app, payment processor and likely overhead like marketing expenses, while effectively having access to twice as many customers that in turn now have access to a much broader, easier and more compelling product. As a result, SHARENOW dominates the vehicular MaaS market while other vehicle manufacturers still seem to try to get at least a foothold in the market.
Another prime example for synergetic consolidation efforts is the recent announcement of BMW and Daimler to jointly develop driver assistance systems (level 4 autonomy) for highly automated driving and parking.
Given the monumental task, it seems natural to seek alliances in order to speed up development and share cost. The same applies to Volkswagen and Ford, who joined hands to cooperate in e-mobility and autonomous driving. There are likely more manufacturers who are already crafting alliances.
Interoperability and standardisation aren’t two other factors. Any vehicle manufacturer that would, for example, want to integrate functions for automated road toll payments into cars would currently have to integrate 27 different standards — and that’s just for the European market.
If taking into account that all those systems are constantly changing over time, and that maintenance for all of them has to be provided, integrating 27 different solutions is a rather big challenge that keeps costing vehicle manufacturers money while there’s no return on investment for them. Also, every vehicle manufacturer has to start from scratch, figuring out every nick and cranny itself. They can’t share the cost.
Going one step further, if a vehicle manufacturer would want to integrate automated payment solutions for electrical charging, they currently would face 400 different charging pole types and their individual settlement solutions, in Germany alone.
By just looking at these two, simple use-cases, it’s safe to assume that without any standardisation, there will be no automation in future mobility solutions.
It is nearly certain that mobility providers and especially vehicle manufacturers will face dramatic changes in the near future. It is also clear that manufacturers will have to rethink “security” in future mobility solutions. And it’s also clear that automation requires standardised, interoperable and cheap solutions.
IOTA is the only DLT in existence offering immutable data transport capabilities at zero transactional cost, while also offering value transfers at zero transactional cost.
IOTA is also in the process of becoming on open standard for the IoT. According to Richard Soley, Executive Director of the Object Management Group, an international body driving the standardisation of IOTA, an IOTA standard is expected somewhere around the end of 2020.
IOTA is also an early member of the “Mobility Open Blockchain Initiative” (MOBI), representing roughly half of the world’s largest vehicle manufacturers and mobility providers, while it also has more direct ties to the global top premium vehicle manufacturers than any other cryptocurrency.
The future isn’t set in stone. The road leading to it will in any case be a very scenic one. From today’s perspective, DLT seems like an interesting solution, solving many challenges vehicle manufacturers are going to face. And IOTA seems like a well suited contender.
Whenever it will be, until autonomous vehicles become a reality I will personally refrain from investing in any automotive stocks.
About the author:
My name is Ben. For the better part of the last 20 years I worked as a digital product developer at the intersection of IT, analysis, concept, strategy, business and communication for multinational corporations in Europe as well as a bit in Asia, being tasked with developing answers to the questions of what, who, how, where, when and why. Coincidentally i worked around 8 years for a german premium vehicle manufacturer.
Caring deeply about technology, I spend my spare time exploring new technological concepts and architectures like IOTA, investigating its meaning and its potential influence on business models and society. This is a result of that.
The iotaarchive.com, where most of the references in this article originate from, is myprivate repository tracking the progress of IOTA in different business verticals. New archive entries are published on https://twitter.com/_iotaarchive as-it-happens, whenever i find the time, in case you are interested.
I wrote this article over the course of half a year, spending several days on it. If you liked it and feel generous, feel free to send me some IOTA: QIXHISSVMPUYKTMTCECFVBJSR9TGJI9D9YNHTWMNXQFYELG9INHUAHTHKJBYMWA9GPWXAYSYVTEHFR9ZYYCVSAGOKD
