Technology surrounding autonomous cars that will influence its diffusion

On the technical level, just about the replacing human drivers with artificially intelligent robots but about the replacement, improvement and possibly creation of several (new) technologies as well as the involved mechanisms and patterns.

navigation system: http://www.osnews.com/story/29107/Self-driving_Cars_User_Interface_Will_Be_The_Key_To_Success, interface (voice, HoloActive, buttons, some conversational functionality,), interiours, exteriour, assisttant/personalization/emotions (important t increase trust: http://www.sciencedirect.com/science/article/pii/S0022103114000067)



New technologies can come from inside or outside the traditional automotive chain. Shows the importance of industry outsiders.

Assuming that AC technology is ready, is all the other tech ready?

Whereas ACs are getting better it is too early to consider them public ready and from my point of view — or rather gut feeling — I would say that it won’t be ready by 2020/2021, but assuming that it will, we have to make sure that at all the other technology around it is ready as well.

The connected car and assisting driving systems are user preference enablers (drive down costs while driving up performance through increased sales).

Even if go for mixed operational mode in the future, “Nearly error free decision making is a very hard problem but it needs to be solved before an autonomous vehicle system that provides ‘Full Driver Assist’ is ready. Predicting when it will be feasible is merely guess work, but a roadmap would still be useful in approaching it in a comprehensive and systematic fashion. ” (Rupp & King, 2010).

Will they slower than MC in the beginning?

Issue: “random non-repeatable events” (Rupp & King, 2010) + previously undriven roads (Rupp & King, 2010) + … (Rupp & King, 2010) à will AC ever be 100% accurate? + one big issue = unpredictability caused by human factor

Software

Will there be an “OS war” amongst OS for in-car technology? Toyota, for example, is working together with, amongst others, Mazda, Suzuki and Daimler AG on an open-source OS for in-car technology.

In regards to autonomous driving software, i. e. the software that makes the autonomous driving decisions, it seems that each company is working on the own algorithms. In regards to V2V and V2X communication at least some common standards will be necessary. And in regards to maps, several companies are jointly working on them.

The use of similar software in mapping and communication and a possible interchange of autonomous driving software between manufacturers poses a dangerous dependence. If all cars of the future rely to the same extend on the same code, all these cars will be affected simultaneously. Whereas we have such issues with traditional cars as well, I believe that car-interconnectivity will make matters more complicated with autonomous cars.

Combating boredom through in-car entertainment enabled by improved interior

What I find crazy is how much people like commuting. The chart from a study show that even if a commute takes more than half an hour or traffic delays it by more than 15 minutes people still like it.

Changing attitude towards commuting based on traffic and commute duration (Source)

Rupp & King argue that this is because commuting gives people “quiet or alone time”. Considering this, it seems that getting rid of traffic isn’t necessarily something people would want (😳) but rather enhance the experience while in it.

From another perspective I am curious whether making a commute is really desired by so many people because I think that firstly the “quiet or alone time” it gives them is really valuable and secondly that the combination of low key physical exercises and limited required attention creates an enjoyable environment. I think taking this away from people can only happen based on two conditions (applicable separately).

Firstly, people will look for something external to get back the “car thinking time”. Secondly, in-car entertainment (and improved interior) will compensate for the lacking motoric activities. In regards to the second I think that in-car entertainment technology will be important because people will need to have something to do while “driving autonomously”. Mereceds’ Vitality Coach (an AI-based coach in the car aiming at helping the driver to relax or exercise) and I think that more in that direction will be required. People could, of course, just continue using the gadgets they usually do (smartphones, laptops…). In either case, a different interior will be required to enable that kind of use (see aesthetic and attitude sense).

In this context we have one “landscape development” meaning that the “high consumer demand for infotainment in the car” correlates to people’s desire for “relaxation through music or conversation” (Rupp & King, 2010) This is especially interesting as traffic jams are expected to rise.

However, “Gen Y” might not be so ok with driving at all: “Regulation keeps trying to say texting is distracting to driving but for the consumer it is really the driving that is distracting to texting” (Markus Maurer, 2016). This leads me to the consideration that using AC to get rid of commuting might not be the best idea after all.

Motorization and charging infrastructure

Assuming that AC will entail them being electrically powered we will certainly need more capable batteries than we have today including the necessary charging infrastructure. In the context of this assumption I am curios, from a non-expert POV, whether car companies have enough ressources to work on AC as well as electric engines (as well as all the other technologies required).

Technical infrastructure

One aspect of technical infrastructure where we will need additional work is Vehicle-to-Vehicle and especially Vehicle-to-Everything communication to enhance the car’s communication with the outside world. Interestingly, people conceive this topic in a negative way, by perceiving the technical infrastructure to be insufficient as shown in a survey conducted by Eva Fraedrich and Barbara Lenz and published in Autonomous Driving: Technical, Legal and Social Aspects***.

The diffusion of AC also requires the introduction of new infrastructure. Here we are talking about charging stations (in case of electrically powered AC) and broadband internet structure for computations. + road stuff


Tier three technologies

Transformation from horse-drawn carriages to cars has shown that technologies from outside the regime might be necessary to for the core product to diffuse. Besides the obvious innovations required from outside the transportation landscape (e. g. communications infrastructure), I would go one step further and argue that new technologies might arise in the course of the AC’s diffusion that could either solve existing bottlenecks or that might enable previously unexpected solutions (consider, for example, what Blockchain could enable).

Human-machine interaction (HMI)

If we look at some aspects that AC entail we can look at HMI in regards to internally as well externally oriented HMI.

HMI required due to

  • Internally oriented
  • Externally oriented
  • more free time

Internally oriented

Refers to …

Assuming that ACs won’t reduce congestion but rather make it more pleasurable, the question arises what people will do with all the free time they have? Will whatever in-car enables be better than driving?

Here we have to consider what type of HMI we will have. (Rupp & King, 2010) argue that BMI (Brain-Machine Interface; establishing a direct — wired — connection between the brain and an external device) may be the best type of HMI for ACs. Although I imagine it to be a little bit scary, it sounds awesome 😎. Alternatives would be voice, touch, gestures… . On a somewhat more granular level we have to consider what type of interaction we will have content-wise. If we eventually want to have a conversational way of interacting, further work will be required to bring AI to the level where it can understand humans’ conversational input. Siri and lately Alexa and Co. are examples that show that we are still (far) away from that dream. In this context we have to ask whether the HMI will be annoying. This can be caused by the car’s (justified or not) actions or corrections (e. g. adjusting driving speed) that might be perceived annoying be the driver. This is, by the way, another reason why “mixed autonomy” (manual and autonomous driving) might not be the best idea (because that half-developed technology might be more annoying that useful). Other reasons for annoyance are the “typical” issues we experience with our current devices such as imperfect autocorrection. Will cars have gut feeling as humans have or cut people off (e. g. when they learn that behavior from their driver or when they deem it necessary to catch an exist?)

On a yet more granular level the discussion centers around the type of relationship we will have with the car. Toyota, for example, emphasizes strongly the interaction on a partner level as also demonstrated by KITT a fear years before. I think that one can go even one step further by looking at a cummulated and repeated set of interactions leading to a human-machine-relationship rising the question what kind of bond the driver will have towards the car. Will drivers treat car similarly to human beings (already today some people invest considerable ressources in cars) and possibly even grief if the car dies (i. e. will the loss of an “anthropomorphised” AC be equivalent to losing a human being)? Consider in this context the following extract from Isaac Asimov’s short story Sally, where Sally is an autonomous car:

“When you turn off Sally’s motor, you hurt her. How would you like to be kicked unconscious? That’s what you do to Sally, when you turn her off.” — Isaac Asimov’s Sally

Whereas this level of human-like treatment of vehicles might sound pathological, it is not even unperceivable considering that I have seen people wait a few minutes before turning off their car (apparently to conserve the motor after a longer driver), that people apparently have more than just practical interests toward the voice assistant Alexa (people apparently love Alexa and have sexual fantasies with it) and that something similar has actually happened with the introduction of tractors in the Punjab (see for details on this story Rogers, 2003).

The introduction of tractors in the Punjab
The tractors were introduced as a replacement for the bullocks used for farming and transportation. Although the tractors were adopted very willingly by the farmers, the tractors broke down relatively quickly due to overheating caused by farmers covering the tractors with blankets. As these farmers interpreted the tractors as similar to the bullocks they covered the tractors with the very same blankets they used to cover their bullocks for the very same reason, namely in order to protect them from freezing. These examples coalesce around the question which level of anthropomorphism AC have.

In this context it makes sense to ask:

  • Will companies be able to created consumers’ desired level of anthropomorphism?
  • How much of an obstacle will the uncanny valley truly pose? (the hypothesis that humans won’t accept all level of anthropomorphism but — broadly speaking — either full or no anthropomorphism) Apparently people trust AC more when they are “anthropomorphised”

At this point — at least that’s my impression — neither consumers nor producers are sure which way they want. Whereas Mercedes doesn’t anthropomorphise it concept car F 015 (although Mercedes’ Vitality Coach as part of the company’s foray into the car as the “central health hub” goes somewhat into an anthropomorphic car), Toyota’s concept the Concept-i demonstrates a highly anthropomorphic driving experience with its personal assistant Yui. I’m additionally clueless whether we might end up having both designs, anthropomorphic and not. An analogy from today might be highly emotionalized brands (premium cars) vs. “no-name” cars. Furthermore, “anthropomorphic vs. not” could possibly also become a feature in the future where you would be able to add an AI-based PA as an equipment feature to your car.

Understanding

In this context we have to ask whether it will be ok for the driver if she/he doesn’t udnerstandt the car’s action. Rupp & King, 2010 argue that “the potential for automation success increases” when

  • the driver knows concretely and on time what the AC is doing and is about to do
  • the driver knows how the AC operates in general (“having a mental model of the car’s working”)
  • the AC’s actions are consistent with the the driver’s “prior direction”

Based on this the consequences arises whether the driver will “understand” what the car is doing now (and why) and what it will do next (and why). Whereas one might consider this easily achievable, it seems that even pilots have difficulties understanding what “their” autopilot’s actions, despite having extensive training and the autopilot’s past improvements:

Experience with automated aircraft cockpits reveals that operators are often uncertain about its ‘behavior’. What is it doing now? What will it do next? How did I get into this mode? I know there is a way to get it to do what I want, but how?

On the literal other side, the car will have to understand whether the driver is capable of taking over control (e. g. due do drunkeness).

Driving mode

If we consider that the driving mode of AC will be — at least temporarily — a mixed one where the car and driver will have to exchange the driving role (equaling level 2 and 3 autonomy; e. g. when the car is incapable of handling a situation and has to hand over the driving to the driver — hand off, requiring certain actions from the car or taking over complete control) it becomes clear that the “car of the future” will have to become better at the interaction with the driver. Amongst other things it will be required to have greater situational awareness (which cars are where around me, what is the speed limit, what is the car trying to do, what are other road participants — drivers, pedestrians — doing or intending to do…) and also it will be required to adequately (timing- and type-wise) communicate that information to the drive. This interaction might show itself in, for example, alerting about possible accidents or slight involvement (e. g. correcting the car’s position when the driver crosses a line).

Learning

Drivers will have learn that they will become operators when in the “on-demand autonomous driving mode”. This means that the driver’s task will be more strategic (e. g. where to go and which to road to choose) than tactical (e. g. which lane or gear to choose) (based on Rupp & King, 2010).

Furthermore, the question arise how much do cars need to learn from a driver’s past driving experience? It seems to me that after buying an AC for the first time we will have to spend some time waiting for the car to learn our driving patterns before its autonomous technology can serve as a companion. In the case of Mercedes’ ATTENTION ASSIST (a feature capable of measuring a driver’s level of tiredness) needs, for example, 15 to 20 minutes to learn the driver’s driving patterns. This also entails the question of a “driver profile”: can we already now with our MCs start collecting information about our driving pattern and then just feed it into the AC when we buy it? Can we transfer one profile from one car to another or will these profiles be proprietary? Will we have a profile at all? Will that be a field where one company will be able to establish a standard? If a “standard war” breaks out or standards are required how will that influence the AC’s diffusion (possibly slow it down).

Finally, assuming that the car will require the driver as a techaer, what will happen if we teach the car the wrong things?

Externally oriented / Interaction w/ outside world

Today we (all road users) understand other drivers’ intentions by establishing eye contact with them. With AC that will obviously change and rises the question how we will cope with that. There are two “hows” to consider. On the one side we have to look at we will solve it technologically and on the other side we have to educate consumers somehow. I can imagine that especially consumer education would take a significant amount of time.

Mercedes, for example, “uses” a of projected crosswalk (lights built in the car’s front project the crosswalk onto the streets) supported by auditive indicators when it is safte to cross (see video here and image below). Further, I I am by the way curios how feasible such projectors are already today.

Mercedes’ self-driving concept car F 015 with a built-crosswalk (Source)

Additionally, it this example merits to think how this would change people’s respect towards car. From Mercedes’ example we can derive that they are prioritzing pedestrians over cars. If pedestrians understand this behavior and AC become better at recognizing objects than human drivers (i. e. they can predict with higher certainty whether somebody is going to cross the road) will that make pedestrian more reckless knowing that — legal consequences aside — nothin can happen to them because the car knows that they are about to cross the road and that there might be no “social consequence” imposed by drivers shouting at you.

From the driver’s side, I am curious whether they will be angry because their cars is stopping “everywhere”.

Another challenge for AC’s behavior in regards to new or only slightly familiar road participants. How difficult will it be for AC’s to interpret a traffic pilot’s gestures or how to interact with completely unfamilar vehicles?

All that considered, assuming that the core AC technology and all the other tech is ready, we must further look at whether people and society are.

Bullying by other drivers: https://www.theguardian.com/technology/2016/oct/30/volvo-self-driving-car-autonomous

I’m less confident in their ability to avoid non-lethal catastrophes. Consumers will not have a lot of patience for cars that take their occupants to the wrong location, attempt to drive on footpaths, try to board non-existent ferries, or get stuck at intersections by being overly cautious in response to other road users, strand them in dangerous neighborhoods, or hit buses while negotiating road hazards. People will get very impatient indeed if their cars drop them off to go shopping then never return because somebody put a cardboard box in front of the space it’s stopped in

Additional technology depending on dominant design

Then, depending on what the AC will entail several other technologies might be required. The “office AC”, for example, might need more sockets and a better ventilation system than current cars.

The importance of industry outsiders in driving new technologies and their production as well as innovation

There are two ways how outsiders show their importance. One the other side, industry outsiders might be important because they provide technology that is necessary for the “core tech” (here AC) to diffuse but which companies from within the industry cannot provide.

The importance of industry outsiders for new technologies and their production

Think, for example, how telecommunication infrastructure providers will be important for V2X communication as part of an autonomous world. I have outlined both perspectives here in more detail (see “Influence by tier two companies” and “Influence by tier three companies”)

Basing on the above discussion about exteriour and interior design I am assuming that new materials will be required to build all these visions. Further, I will also assume that broad scale production of AC will require changes in production processes (OTA updates, short release cycles because people expect the newest technology which they learn about through increased car/ridesharing or are used to by the technology they are currently using such as smartphones)

The importance of industry outsiders for innovation

One the one side, industry outsiders can initiate strategic shifts (new business models, technologies…), for example, because they are not invested in the established technology.

Path of technological progress

We need a dominant design — people need to know what an AC is — all questions have to be answered

In this context, the fact that several AC designs exist and consumers haven’t decided on a design yet (which, of course, they couldn’t because AC are not widely sold yet) is another indicator for me that the AC diffusion will still take some time.

planes @ war were either fuel efficient || had rapid acceleration

Talk about public perception: do we have one on AC ? #tbd à be4 read the rel. cpa @ AC book (some are there)à nevertheless: bottom line is that IMHO this “perception shaping” point is important + à is this important as a general concept (the concept of showing people how an innovation should look like & what it should be capable of à ppl. Learn what to expect from tech à was mentiond in one of the pahsed identifei dby Geels if am not mistakne. )? Do we generally need to have some “studies” showing what something should look like (e. g. F 015 shows how AC should be?) Google wining @ GO shows what AI is capable of (& then it lands @ consumers’ minds…) à expand on this

Co-Evolution

Co-Evolution of Multiple Technologies shows that it will considerable time to make self-driving cars a revolution-

The fruits of this new deal should show up within the next five years, according to BMW’s CEO: “We have a clear objective: to be the technology and innovation leader for individual mobility in the digital age. To achieve this, we are establishing selected partnerships. Together with Intel and Mobileye, we aim to develop the technology for autonomous driving within five years. All three partners will pool their know-how in the fields of automotive engineering, technology, computer vision and machine learning.” (http://www.bmwblog.com/2016/08/02/bmw-ceo-wants-autonomous-driving-cars-within-five-years/)


Examples:

(a) incremental innovations alongside radical innovations: new materials, entertainment in cars, batteries……

(b) cross-sectoral innovation: aerodynamics, petroleum

Also, assuming further that AC entail them being electrically powered and our current battery technology not yet capable of supporting long-range driving distances we have another obvious technical challenged to solve. Furthermore, if AC lead to more driving time this would require a different interior as people will be spending more time in the car. à #tbd link to blog article

#tbd: exterior

#tbd txt from somewhere els

The diffusion of AC also requires the introduction of new infrastructure. Here we are talking about charging stations (in case of electrically powered AC) and broadband internet structure for computations. Also, AC will call for new traffic rules, especially for the coordination between pedestrians and the cars.

require co-evolution. navigation system: http://www.osnews.com/story/29107/Self-driving_Cars_User_Interface_Will_Be_The_Key_To_Success, interface (voice, HoloActive, buttons, some conversational functionality,), interiours, exteriour, assisttant/personalization/emotions (important t increase trust: http://www.sciencedirect.com/science/article/pii/S0022103114000067)

Though I don’t give much credence to dwelling on the nightmare scenario, considering how much safer even the most rudimentary self-driving car technology is going to be, I’m also confident that car company engineers are going to do a pretty good job of making their cars react to dangers.http://www.osnews.com/story/29107/Self-driving_Cars_User_Interface_Will_Be_The_Key_To_Success

I’m less confident in their ability to avoid non-lethal catastrophes. Consumers will not have a lot of patience for cars that take their occupants to the wrong location, attempt to drive on footpaths, try to board non-existent ferries, or get stuck at intersections by being overly cautious in response to other road users, strand them in dangerous neighborhoods, or hit buses while negotiating road hazards. People will get very impatient indeed if their cars drop them off to go shopping then never return because somebody put a cardboard box in front of the space it’s stopped in.

incremental innovations will be important

The two important technologies driving this trend includes visual analysis/image processing and natural language processing. Though both are becoming realistic but the 100% accuracy in analysis is still a long way. The accuracy come from the quality of image sensors, which captures the scene as good as eye or even better than that. There is significant improvement, there are image sensors now available which can produce colour images of the scene, even in extreme poor light conditions such as night. So in some sense, image sensors are matching up to human eye. Also sometimes image sensors can outperform human eye by taking infrared light based thermal image.

http://www.eeherald.com/section/news/onws20141026003a.html

Some of these problems will be solved with better databases of real world conditions and better AI to analyze unfamiliar situations, but perhaps the most effective way of building these databases and training these AIs will be to utilize the faculties of the passengers who end up experiencing sub-optimal navigation from cars they’re in. “Damn it, Siri, that’s a one way road!” is a very useful piece of data, if the car is equipped to use it. And I’d suspect that a lot of the user feedback is going to sound more like “Whoah, whoah, whoah, WOAH, AAAAAH!” http://www.osnews.com/story/29107/Self-driving_Cars_User_Interface_Will_Be_The_Key_To_Success

A truly useable self-driving car is going to have to be part of a complex ecosystem of technologies: hardware, software, automated services, and real-world support.

call a guy 2 ask what the stuff in front of me is (if the car cannot move), call a special service 2 be able to repair stuff

http://www.osnews.com/story/29107/Self-driving_Cars_User_Interface_Will_Be_The_Key_To_Success

maybe it will draw from other technologies? e. g. drones @ war

Interfaces:

- Car<>passengers

- Car<>pedestrians

F 015 has signs on the street, but also talks, conversational interfaces (bots)…


see (Rupp & King, 2010) @ 2.4 for techn. Progress in r 2 tech. + link that 2 niches: niches will emerge alongside the techn. Possibilities.

Secondly, in regards to the car at least two diffusions took place. Firstly, as prerequisite for the car’s diffusion, the right car engines (steam, electric or gasoline) had to diffuse. This diffusion was made possible by technological advances (#tbd if !). Secondly, the car in itself had to diffuse. à look @ what enabled the car, what has to be enabled for the AC + role of catalysts

I think it makes sense to assume that the transition from manual cars to fully self-driving cars (level 5 autonomy) will take similarly long as the transition from horse-drawn carriages to manual cars (70 years) (Geels, 2005).

Whereas technology nowadays is accelerating faster than in 19th and 20th century there is scientific evidence that the diffusion rate of a new technologies is the same as it is for the old technology it is replacing (Elie Ofek, 2016).

Also, AC have actually been diffusing — as mentioned above — since the 1920s à post about the history of the autonomous car until from a sociological POV + wikipedia

By the way, for the last 100 years the self-driving car was always supposedly 20 years away (Markus Maurer, 2016). #tbd exact source

One way to track commercial AC readiness is to track the price/performance ratio of the involved technology. à link 2 how shitty expensive AC are supposed 2 be + how much “state of the art” AC cost today

For involved technology see (Rupp & King, 2010) @ 2.2

On the technical level getting the overall difficulty of having fully autonomous cars that can react to every traffic situation (e. g. heavy traffic) under any driving condition (e. g. heavy rain and bad sight) is possibly the biggest problem to overcome.


9. Rapid upscaling: Refers to the issue that taking too big steps in the development of the new technology might be risky if the performance is not yet stable and result in fatal accidents. In self-driving cars this might be the case when companies move too fast to too high levels of autonomy. Further being too far ahead / taking large steps might be negative because ppl. != ready 4/used 2 the tech (see “add-ons”). I consider big bang approach unrealistic.

Use info from (Rupp & King, 2010) @ 2.2 r diffusion of ACC as analogy + this from (Rupp & King, 2010): “Anti-lock braking systems were introduced in 1971, and reached 86% market penetration only after 37 years, in 2008. Compare that to Electronic Stability Control (ESC), introduced in 1995. Although the industry already had an implementation plan, the U.S. National Highway Traffic Safety Administration (NHTSA) accelerated penetration by mandating standard ESC in all new vehicles by 2012, less than 20 years later. NHTSA has included Forward Collision Warning and Lane Departure Warning in the ratings for the Active Safety New Car Assessment Program. The European Commission is considering mandates for Collision Mitigation Systems on light commercial vehicles. Non-governmental organizations such as the Insurance Institute for Highway Safety and the Consumers Union (publishers of Consumers Report magazine) have started to address CADS technologies, raising consumer awareness. Insurance companies are considering lower rates for vehicles with CADS features. „ + automatic transmission+ eCall

How long did that take?

“always in 20 years”

half acing ac issue

 → try bass





I am by no means an expert in autonomous driving, the automotive industry or anything related to it so I have to believe the experts that autonomous cars will be commercially available by 2020. However, as I do have thoughts on the tech behind autonomous driving it and value Gill Pratt’s opinion when he argues that “none of us in the automobile or IT industries are close to achieving true Level 5 autonomy”.



Use info from (Rupp & King, 2010) @ 2.2 r diffusion of ACC as analogy + this from (Rupp & King, 2010): “Anti-lock braking systems were introduced in 1971, and reached 86% market penetration only after 37 years, in 2008. Compare that to Electronic Stability Control (ESC), introduced in 1995. Although the industry already had an implementation plan, the U.S. National Highway Traffic Safety Administration (NHTSA) accelerated penetration by mandating standard ESC in all new vehicles by 2012, less than 20 years later. NHTSA has included Forward Collision Warning and Lane Departure Warning in the ratings for the Active Safety New Car Assessment Program. The European Commission is considering mandates for Collision Mitigation Systems on light commercial vehicles. Non-governmental organizations such as the Insurance Institute for Highway Safety and the Consumers Union (publishers of Consumers Report magazine) have started to address CADS technologies, raising consumer awareness. Insurance companies are considering lower rates for vehicles with CADS features. „ + automatic transmission+ eCall +

The novelty starts as an auxiliary to the incumbent technology

Usually a new technology (in the context of a “system technology”, like transportation) doesn’t compete with the incumbent technology from the beginning but they instead engage in an collaborative relationship. In this relationship the novelty serves as an add-on to the established one the result is a hybrid version of these two competing technologies. The hybrid car and voice-UI (Siri as an add-on to the iPhone’s touch-based input) are two examples of that. Similarly to that, I don’t see any other way for the AC’s diffusion. In the beginning I can only imagine that we will have “mixed roads” (ACs and MCs). Similarly then, to how voic could completely overtake touch (at least in theory) as the default UI, AC could eventually fully replace MC. The connected car and assisting driving systems are technological enablers (drive down costs while driving up performance through increased sales).

This path is important because it increases the chance of a successful diffusion and although such a incremental introduction appears to take longer than a “big bang” approach, it will in effect be even quicker because it lowers the chance of consumer backslashes, errors… . There are four reasons for that increased success:

  1. Hybrid version lower the opposition by peopled invested in the “old” technology
  2. Gives the AC technology time to gradually improve
  3. More time for the ecosystem to emerge
  4. Helps people gradually get used to it: It goes without saying that consumers won’t immediately adopt AC. One reason for that is missing compatibility with people’s current experiences. Through gradual introduction people’s experiences towards cars will incrementally be autonomous and thus rise compatibility with the AC. The failure of the steam car in the transformation from horse-drawn carriages to cars was due to a familiarity issue. When steam cars were first introduced people rejected them, amongst other things, due to their high-speed being incompatible with what they were used. It was only after the e-tram familiarized people with higher speeds that the ICE car (speed similar to steam cars) was accepted. Today’s manual cars equipped with self-driving tech, like autonomous parking or line assistants, are one way of familiarizing people with an AC world.

One example for an auxiliary technology is the emergence of the first motorized car or rather motorized carriage. It was in 1885 when Daimler added created that vehicle (car/carriage) by adding a gasoline engine the carriages (see image). In this case adding the gasoline engine to a carriage created a hybrid between horse-drawn carriages and the first car based on today’s criteria (the Mercedes from 1901, see other image below).

Daimler’s horseless carriage (1885) on the left vs. the Mercedes from 1901 (first car based on today’s criteria) (Source)

The importance of deliberately slowing down the technology’s diffusion

Although it might sound counterintuitive at first, it might actually make sense to slow down a technology’s diffusion. One reason for that is the above-mentioned need for people’s familiarization (see “Helps people gradually get used to it” under “The novelty starts as an auxiliary to the incumbent technology”). Another reason is that taking too big steps in the development of the new technology might be risky if the performance is not yet stable and thus result in fatal accidents.

Enablers

as opposed to sttback


Ride sharing as an enabler

Ride sharing’s role as an autonomous car enabler is another reason meriting its mentioning in this context. It shows its role as an enabler in two ways. Firstly, proving the cars’ convenience (see above) without owning a car and secondly, familiarizing people with the ideas of not owning a car and slowly convincing them of the advantages of not having to drive. Showing people the perks of passive driving and thus familiarizing them with the idea of not driving might be of great important considering how much people like driving, even when stuck in traffic (see “combating boredom through in-car entertainment enabled by improved interior” here). Moreover, the pure absence of having to take care of a car in such scenarios might be another “breeding ground” for AC. The working on cars and the fixing of these was by the way one of the reasons they started diffusing in the first place because that appealed to the first buyers (see Technological Transitions And System Innovations: A Co-evolutionary And Socio-technical Analysis for details on that). In the long run such “pre-diffusion” could lead to easy and quick consumer acceptance of fully autonomous taxis.

For all these aspects (Consumer changing preferences and expectations towards cars) the standard is set by car/ridesharing and today’s consumer technology. Through sharing people are more likely to be exposed to new technology faster and more frequent.


In this sense, however, I don’t think that today’s cars equipped with considerably improved in-car technology will slow down the autonomous car’s diffusion, on the contrary, such improvement might actually contribute to consumer adoption. On the one side it is slowly familiarizing consumers with high-tech in-car technology and on the other side AC — as we are envisioning them today — can only exist once such technology is widely available through a mainstream price/performance ratio (see also “Co-evolution of technologies” here)

Setbacks

Sailing ship effect + “competing technology diffusion

Sailing ship effect

This phenomena describes a situation in which an incumbent technology improves when it is challenged by a new one. Whereas I don’t see this happening directly for AC, I can see how that could slow down the AC’ diffusion by slowing down the development of the ICE. Nevertheless, this can only work until a certain threshold is reached beyond which further incremental improvements won’t be able to keep up with the advantages of the radical innovation, the EV in this case. I think another area where we could see the sailing ship effect is in-car entertainment. Considering people’s “need” for in-car relaxation and entertainment (caused, amongst others, by traffic) AC could serve that need. However, a continuouos improvement of these technologies among existing cars could lead to them increasingly getter more comfortable, without never reaching the “next level of comofort” which is only possible through the the technological (disruptive) “leap” made possible be AC. A similar development can be observed using the example of driving assistant features such as line keeping assistants. Consumers might actually consider this “good enough” because it is a sweet spot between driving on their own and having a better driving experience thus render the use for an AC non-existent. → copy to clarifying article

“Competing technology diffusion” based on the transformation from horse-drawn carriages to cars

In addition to seeing in-car technology as an alternative technology to AC as a solution to traffic jams we could look at Elon Musk’s electric skate (underground tunnels where cars are being transported automatically).

Alternatives to traffic problem are, for example, flying cars. Uber, Google and Airbus are already working on that.

So, the current situation is that people want to be able to get directly to any place at any time, with full control over the environment but also without the annoyed of traffic jams and the flexibility to do other things than driving (like texting or sleeping). In other words, people want (“need”) a “personalized, on-demand, door-to-door, automated transportation system”.


With “Competing technology diffusion” I refer to the situation where we have one problem and two possible solutions (technologies) for that problem where the more suitable solution diffusion will diffuse.

The transformation from horse-drawn carriage to cars (see Technological Transitions And System Innovations: A Co-evolutionary And Socio-technical Analysis) is a real-life example that shows a similar pattern. The e-tram and the car first appeared roughly at the same time (1880s/1890s) and diffused, among other things, due to them being a solution to a then prevalent problem. The problem was “overurbanization” (overcrowding and similar of cities) and their solution was that they enabled suburban living by enabling faster and wider travel than previously possible with horses. However, because the time was more suitable for e-trams than cars, the e-tram diffused before the car. (The diffusion was, by the way, only of temporary nature and in effect that e-tram “paved the way” for the car’s diffusion).

Applying this concept to the autonomous car’s diffusion, we could frame “traffic accidents caused by distracted driving due to cell phones uses” (number one cause of traffic accidents) as the problem and autonomous cars and by law required in-car radio jammers as the two competing solutions.

As hinted above, or outlined in another post of mine, traffic accidents might not be a problem in the first place in the context of the autonomous car, but due to the medial perception of this issue and the repeated mentioning of it in regards to autonomous cars it is well suited to outline my argument. This being said, we could imagine, in a very theoretical world, abandoning the idea of autonomous cars as the solution to our fatalities problem and solve it via the other technology, jammers. Whereas this idea might sound absurd at first sight (as well as at second) there is more to it than it being a pipe dream of mine.

Supporting the “competing technology diffusion” based on the eCall and the “first” autonomous car

With the eCall something similar to the above-mentioned jammers is being implemented right now. eCall is a system required to be embedded in all new cars after April 2018 that will automatically call the European emergency number 112 and transmit accident-related and rescue-relevant information in case of a “serious accident”. This won’t solve the solution of traffic accidents but it might decrease their consequences if helpers can reach the location quicker and better prepared. Admittedly, the comparison is a stretch but it makes sense because the two technologies (jammer and eCall) are conceptually insofar similar as they are both add-ons to the car and that they are required by law.

Another reason why the jammer (or any other technology) might diffuse instead of the car is again provided by history, in this case by the first autonomous car first diffusion. The first autonomous car in form of a remotely controlled driverless one appeared in the 1920s (see image).

A picture of the first autonomous car in form of a remotely controlled driverless vehicle (Source)

Notable about this, amongst others, is that the idea of an AC emerged as a response to increased traffic fatalities as a consequence of mass motorisation receive. As the human factor was considered cause number one for that, removing the driver from the equation seemed like the best solution (infrastructure and car design as causes for fatalities entered the equation only at a later stage, see Autonomous Driving: Technical, Legal and Social Aspects for details on that)

Further example for “competing technology diffusion” and relevance today

I am not saying that the AC won’t reduce (or possibly completely remove) traffic accidents, especially as it is clear that the human factor is causing them, I am saying that it is possible that the AC’s diffusion stops again because we encounter alternative solutions to the problems the AC is addressing or alternative satisfiers for the advantages the AC is promising (e. g. increasing driving comfort). At this point of the AC’s diffusion process, however, I don’t (yet) see this happening because the (social) problems the AC is trying to solve (the mentioned traffic fatalities, congestion…) are not as pressing as they were in the above-mentioned transformation from horse-drawn carriages to cars with “overurbanization” and because the AC’s advantages are more technology push (by companies) than pull (by consumer needs) (which in the case of something as transformational as cars might be a problem as well).

Imagine, however, another example, where we frame the AC as the solution to innercity congestion. In this case the AC would compete against other solutions such as day-based driving restrictions, Elon Musk’s electric skate (underground tunnels where cars are being transported automatically) or people’s frustration with being stuck in traffic could be mitigated to some extend by improved in-car entertainment and wellness features (see “Contribution to traffic optimization” here)

I think that these two theoretical examples (autonomous car vs. day-based driving restrictions or underground tunnels against innercity congestion and autonomous car vs. by law required in-car radio jammers against innercity congestion) and the upcoming of the first AC show that once problems or needs that could be satisfied or solved by AC arise, we must not expect the AC to diffuse as a solution/satisfier for that. In fact, we could go so far and argue that visions created by AC (such as the “vision” of less traffic accidents was created by the introduction of the remotely controlled car) would end up changing consumer preferences so strongly that other solutions for that arise and we again lose interest in the AC (as it was with the AC’s first try).

Ride sharing as a competing technology

I have examplained that stuff here:https://medium.com/p/df8a6d023712/edit This need can be fulfilled by two products or rather “solutions” (see “Validating the “leapfrog strategy” by framing the autonomous cars as a solution and not product” here, where I explain that the autonomous car is a solution and not a product). One the one side we have autonomous cars and on the other side, ride sharing. Although autonomous cars and ride sharing are different products they should be considered simultaneously because conceptually they are very similar. Furthermore, considering alternative solutions (possibly competing ones) to the autonomous car makes insofar sense because we must not take the diffusion of the autonomous car for granted). On the other side, ride sharing might slow down the diffusion because its value (the advantages — personalized, door-to-door transportation at low upfront investment, maintenance and possibly operation costs — minus disadvantages) is bigger than with AC.

Although China’s “car-eating” bus in specific seems to be a failure, it is another example of a technology competing with autonomous cars.

Dominant design

The transformation from horse-drawn carriages to cars has shown how four dynamics about the dominant design concept*:

  1. The dominant car design as a prerequisite for diffusion
  2. A dominant design might stretch to accommodate the other market niches’ requirements while the basic technology stays the same. In regards to cars ICE cars have the same tech but they come in a variety of forms and shapes.
  3. The perceived dominant design can change over time thus making enabling the recognition of a dominant design only in hindsight
  4. “perception shaping” through publicly demonstrating dominant design ideas is necessary

A survey conducted by Eva Fraedrich and Barbara Lenz and published in Autonomous Driving: Technical, Legal and Social Aspects found that people in fact do care about that topic. **

The missing dominant AC design posses a hinderance to its diffusion

The fact that several AC designs exist and consumers haven’t decided on a design yet is an indicator for me that the AC diffusion will still take some time. In the context of the transformation from horse-dran carriages to car, dominant design was twofold, namly technological and aesthetic way. In the case of AC the situation is similar.

Technological sense

For the first car the major technical dominant design was the establishing of the right motorization (the internal combustion engine won steam and electric engines) a . For AC we might as well motorization as establishing aspect but also less fundamental technologies such as the right UI (voice, touch, gestures…), anthropomorphic or not (see below), user profiles file type (see below) or the type V2V communication (think protocols).

There will, of course, be some technological designs that will differ from manufacturer to manufacturer and cases where the dominant design will “stretch” (as explained above), but there will also be aspects on which consumers as well as companies will have to agree upon. One example for that is, according to Jeffrey D. Rupp and Anthony G. King, AI in regards to which they argue that it is still unclear how it should evolve (“nondeterministic, implementing stochastic type algorithms of learning, optimization, decision making, planning, and goal formation”). They explained that situation back in 2010 and I admittedly have not found out what the progress was in that area.

Aesthetic and attitude sense

if smarpthones/lapots; seating position

Aesthetic sense refers to the AC’s interior and exterior appearance and attitude to what the car represents and which function or use case it serves. As these factors (aesthetic and attitude) are correlated, I will look at them simultaneously. The above-mentioned Mercedes from 1901 established the dominant design for cars with its “nose” (the nose was required to accomodate the additional technology required such as the gearing technique).

Differing exterior across manufacturers hints as missing dominant design 
“Today”, the images below of Mercedes’, BMW’s, Chrysler’s and Toyota’s self-driving cars show that the AC visions of the manufacturers sometimes differ quite significantly.

Left: Mercedes’ and rigth Toyota’s self-driving car concept (Source Mercedes, Source Toyota)
Left: Chrysler’s and right BMW’ sself-driving car concept (Source Chrysler, Source BMW)

Curious that these concepts don’t differ as significantly from MCs as it was the case with cars and carriages. Admittedly, I can imagine that these exteriors can already by considered dominant and just by being very similar to each other representing a design stretch (see above). However, as the interiours and their functions based on that differ more significantly from each other I am not sure if the car of the future will really stretch that wide.

In regards to the first dominant design in an attitude sense, long-rand and high-speed driving were considered the standards for the car. Today’s cars of the future hint at quite different things (except for BMW maybe). Daimler frames it car as mobile office (see interior below), Toyota focuses on a “KITT” like car where the car and driver are “teammates” (I have analyzed their concept in more detail here but only in German) and for Chrysler the car of the future is modular car with focus on virtual and real-life human interaction (see the interior on the picture below; I have analyzed their concept in more detail as well but again only in German). In regards to BMW I am not 100% sure what their AC represents but it for most it seems like the are sticking with their “driving pleasure” slogan (original phrase: Freude am Fahren).

Left: Mercedes’ and rigth Chrysler’s self-driving car concept from the inside (Source Mercedes, Source Chrysler)

The perceived dominant design can change over time thus making enabling the recognition of a dominant design only in hindsight

The transformation from horse-drawn carriages to cars provides an example for that. Before the diffused as all-purpose vehicle it was considered a societal car, an urban car and also a tool (one at a time). With this in mind we can firstly be assured that we think an AC is for will change over time and secondly that we can expect the AC to enable completely new use cases.

“perception shaping” through publicly demonstrating dominant design ideas is necessary

In the course of the AC diffusion it will be important to show people how ACs look like and what it could be capable of resulting in shaping people’s perception (i. e. influencing what people expect from an AC). The presentation of ACs — such as the Chrysler Portal or Toyota’s Concept-i — at the CES is one of doing so. Similarly as Google’s AI winning in GO has shown the the public that computers can indeed be intelligent, the DARPA Challenges and making AC failure rates (disengagement rats) publicly availabe, have shown that cars truly can drive autonomously. Through all these demonstrations organization demonstrating not only how ACs should look like but what they should be capable of. As of now, I doesn’t seem to me that there is common sense among consumers which features will be relevant for an AC but safety is definitely among them. Concretely in regards the “public” will have to demonstrate failure rates and agree upon how safe is safe enough. Gill Pratt, CEO of Toyota Research Institute’s, explains that due to empathy people tolerate other people’s mistake. As according to Pratt this empathy seems to be missing towards machines, people will emotionally only tolerate an error rate of zero, although from a rational standpoint it would make sense to hold the AC to the same standard as humans. Whereas this argumentation makes sense, I could imagine that beyond a certain level of anthropomorphism we might treat our cars as buddies and forgive their mistakes. This consequently raises the question which level of anthropomorphism is human-like enough and whether the car might be at risk of falling in the Uncanny valley (the Uncanny valley refers to the phenomenon that a certain level of anthropomorphism in technological devise might cause repulsion by people).

Additional technology depending on dominant design

Then, depending on what the AC will entail several other technologies might be required. The “office AC”, for example, might need more sockets and a better ventilation system than current cars.

The influence of companies’ strategic actions on the technology’s diffusion

I have written about this topic in a separate post here with the main message being that the micro developments represented by companies’ strategic actions (such as past unamortized investments, employing a “wait and see strategy” or opposed to that engaging in innovation races) can influence the diffusion on macro development and in the worst case slow it down.








Notes

[1] see Technological Transitions And System Innovations: A Co-evolutionary And Socio-technical Analysis for details on that

  • ** “Technology & vehicle design” was found to be of concern to people (although only to some people). The survey analyzed 827 comments on autonomous cars made across 16 articles published in leading German and US newspapers (Germany: Bild, Die Welt, Frankfurter Allgemeine Zeitung, Heise online, Spiegel Online, Süddeutsche.de and Zeit Online. USA: Los Angeles Times, NY Daily News, The New York Times, San Francisco Chronicle, The Wall Street Journal and The Washington Post)
  • ***The survey analyzed 827 comments on autonomous cars made across 16 articles published in leading German and US newspapers (Germany: Bild, Die Welt, Frankfurter Allgemeine Zeitung, Heise online, Spiegel Online, Süddeutsche.de and Zeit Online. USA: Los Angeles Times, NY Daily News, The New York Times, San Francisco Chronicle, The Wall Street Journal and The Washington Post)

Disclaimer: I am currently working for BMW (on project basis, not employed by BMW) and this article is in no way connected to BMW and any focus towards BMW — which I doubt exists in this article — happened purely subconsciously and without any intent.