Thursday 25 July 2013

Autonomous Vehicles: An Inconvenient Truth (Continued)

It's an Inconvenient Truth, but vehicles capable of driving unmanned could be with us by 2017.  That's based on what Ron Medford of Google said at the recent TRB Workshop on Road Vehicle Automation, and on what Sergey Brin (Google co-founder) said at the signing of the California Autonomous Vehicle Bill on 25 Sept 2015.  (Since confirmed on numerous occasions - and in particular by Anthony Levandowski of Google to a question that I asked the panel at the California Public meeting on the Friday after the main Workshop).

On 22 May I presented on this 'Autonomous Vehicles: An Inconvenient Truth' subject in association with Barrie Kirk and Globis Consulting.

My key 'Inconvenient Truths' were:
  1. We aren’t planning for exponential
  2. Much sooner than you think
  3. Business models will be impacted if not disrupted (public and private sectors)
  4. This is but the Dawn of the Robot Revolution 
But since May I have chosen to add some more to the list - see below.

So let's unwrap that a little bit, bearing in mind that there shouldn't be much of a credibility gap with this technology any more, because at the TRB Workshop the White House saw this as possibly the hottest technology policy issue, and the Director of California DMV who does not like hyperbolae saw this as a 'game changer':

1. We aren't planning for exponential

We have four years and small change from now (July 2013) until our 'five year warning date' for the possible deployment of Google self-driving car technology.  If you understand 'exponential' then you will appreciate that in the next four years we are going to see as much technological advancement since 40 years ago - 1973.  I will leave you the reader to get your head round that one - think about how much things have improved that haven't seen a paradigm shift - like cars.  Then consider areas where we have had a paradigm shift - like telecoms - where we have seen four paradigm shifts in that time (Landline-Cell-Smartphone-iPhone/Apps-Social Media Revolution) and are expecting the fifth this year with the release of mainstream wearable computing in the form of Google Glass.

The problem with the transportation mindset, is that the last paradigm shift on the roads started back in 1868 when Karl Benz patented the modern motor car.  Since then we have seen numerous incremental improvements, but no paradigm shifts.  So don't be surprised that the telecoms sector gets what that means, but that anyone in the transportation sector has never experienced it on the job.

Billions of dollars worth of infrastructure projects are being planned or built at the moment, designed to function for 20, 30, 40 years or more.  Have any of the designs been made with a cognisance of the impacts that deployment of autonomous vehicles could cause?  I'll go out on a limb and say the answer is a resounding 'no', and add that it would be great if a reader of this blog could correct me.

2. Much sooner than you think

Just in case you missed it.... 2017

I bet you, your business and the government services around you are not planning for a paradigm shift in how we 'do' road transportation, and all that means to society, to happen in about 4 years time.  But there it is - Google probably know better than any of us the socio-economic impacts of what a self-driving car means, and they have fulfilled their social compact obligations to us, by giving us as much warning as they can that the paradigm shift is coming.  It's now up to us to sort out what all this might mean.

3. Business models will be impacted if not disrupted (public and private sectors)

In most developed nations every jurisdiction is required to produce a Long Range Transportation Plan, because mobility is a key part of a healthy and functioning society - this is basically a business or operational model intended to ensure we plan wisely how our tax dollars are spent.  As many of the key autonomous vehicle developers have facilities in the Silicon Valley, and as Google Self-Driving Cars have been driving on Californian Roads, then you might expect that the San Francisco Bay Area Plan (Draft) to probably be leading the world in what this technology means....  Well, page 125 notes that they exist and that they will be researched.  Yep, that's it - not much help to their planners and engineers, or anyone else for that matter.

Now 2017 may turn out to be an underestimate of how long it takes to have this technology ready to be used safely by the public.  But surely it is best if we all plan for this date, including all businesses and public sector organizations, as the downsides and unintended consequences of not planning could be severe.

Those that don't plan risk becoming collateral damage in terms of business disruption and possibly even impact on your personal situation (e.g. if you are planning on buying a new or replacement car in the next four years there are certain things you would benefit from knowing).  There is more on some specific business models that will be impacted in 6. below.

4. This is but the Dawn of the Robot Revolution 

I may sound nuts (well I think I sound nuts!), but check out the latest DARPA Robotics Challenge (DRC).  Then look at the DARPA Grand Challenge of 2004 that 13 years later may see self-driving cars on our roads.  Then take note of No. 1 on my list above - technology is exponential.  Then consider that as a result of the DRC we could see robots doing manual work outside of factories in maybe 2021-2023.

My guess is that the Artificial Intelligence Operating System (AIOS) being developed for the first autonomous car will likely be the base AIOS for many future robotic developments.  This might dominate the robotics market just as the Microsoft Windows OS has dominated the PC market.

I am also willing to bet that not too many long range plans by municipal and regional governments are taking this robotic future into account.  It likely leads to the de-construction of capitalism as we know it - our wealth could be measured by how many robots we own...

I am not making this up - robotic development is happening at a staggering rate.  Check out some Boston Dynamics videos as an example of the public/civilian face of what is possible.


I'll add an extra Inconvenient Truth for fun:

5. We don't have a consistent name for this technology...

So what's in a name?... I appreciate that this is a US-centric view:

The law uses autonomous - currently Bills passed in Nevada, Florida, California and District of Columbia all define 'Autonomous' vehicles.  I am fairly certain that the bills in process in 15 other US States also do too.

The latest NHTSA definitions use 'Automation' - which was pushed at the first TRB Workshop for Road Vehicle Automation in Irvine in 2012, and 'stuck' with the help of the the NHTSA regulations, SAE and others at the second TRB Workshop in Stanford.  I do find it odd that not even the public bodies can agree on a common name here.  I can imagine the highway designers, having 'Automated Vehicles' in the design codes, and trying to interpret that with the Law that refers to Autonomous Vehicles.

Google use 'Self-Driving', which is simple and descriptive but as they are the principal ones using it, it makes it hard for regulators to adopt it at the moment.  Plus it's an extra word.  On a recent poll on LinkedIn it came out ahead of the other names (but that was in the 'Self Driving Car' group! - who'd have thought!). 

Driverless is probably the most colloquial term that returns the most hits from the search engines.... along with stories of vehicles out of control with no driver present...  Maybe not!

None of the above are unique to this technology - although there are plenty of others that are unique but have not gained anything like as much traction - e.g. Brad Templeton's 'Robocars' - initially I disliked it, but it is uniquely searchable and right now that would be really useful.  My own attempt with 'Autonome' has issues and no traction.  We ideally want a single unmistakable word that is easily searchable - or a unique acronym that everyone can agree on (I will not list the various acronyms - I am at over 20 thus far...).

Next time you get a document that you think should contain a reference to this technology - say a long range transportation plan.  See if you can find it using the search terms above.  I found one document that used none of the terms above and needed me to search for 'Advanced Vehicle Technology'.


My next Inconvenient Truth is deliciously ironic when you consider that many observers agree that autonomous vehicles will see trillion dollar money flows:

6. If this is so important, so soon, so transformational and so disruptive, then why isn't there any money flowing around it yet?

Sure there is lots of private money being spent by the autonomous vehicle technology developers, and a few researchers and University off-shoots getting funding to develop the technology.  But what about the government departments and the businesses that will be affected?  Why aren't they spending money to understand this better and making plans for impacts and potential disruption?  Is anyone concerned that we are in the processing of committing billions of dollars to infrastructure projects with no knowledge if their design is robust enough to accommodate the deployment of autonomous vehicles in as little as four years time?...  Aren't my tax dollars potentially being wasted by ignoring the possibilities here?

My glib answer - there is very little research money flowing around autonomous vehicles because it doesn't have the word 'Connected' in it.  There is a lot of history and politics and vested reputations and vested interests here, so as an example I will cite the US DOT ITS Joint Program Office presentation at the TRB Workshop on Road Vehicle Automation, slide 11:
"Unconnected, automated vehicles could negatively impact road network operations"
and the pièce de résistance:
"“…driverless cars will only arrive if and when all cars are connected to one another and the infrastructure.” - Strategic Analytics"
I would very much like the US DOT to present the evidence to back up those statements - particularly the second one. My research tells me that the second is an incorrect statement (note how polite I am being), and Brad Templeton who is a Google consultant was also very keen to point this error out on the open mic after the presentation. 

But there are plenty of other reasons why autonomous vehicles don't have any money flowing yet into research and revision of operational, business and revenue models.  The main one at the moment being because no one else is spending any money on it.  If no one else is, then why should I? (Errr... Maybe because you stand the risk of going bust, or wasting a lot of money and resources, if you do nothing for too much longer?).  The pressure is building behind this dam, and when it breaks then just about everyone will want studies and all sorts doing to give them confidence moving forward.

Knowing what I know, here are a very few examples where I would be doing my due diligence now:

Public sector:
  • transportation projects or infrastructure, or projects with a surface transportation component
  • operational, business and revenue models relying on predicted road traffic flows more than four years out
  • energy generation and distribution
  • health service provision - particularly trauma and critical care and organ donations
  • financial/treasury models relying on revenues more than two years out
  • Public Private Partnerships (P3s)
Business sectors - expect significant impacts or even disruption to existing operational, business and revenue models:  
  • Trucking
  • Taxis / Limousines
  • Vehicle Rental
  • Car Share / Ride Share etc.
  • Car Parking
  • Auto Insurance
  • Postal / Parcel Delivery
  • Public Private Partnerships (P3s)
  • Auto Body Repair
  • Any vehicle fleet operators
  • Disabled/Seniors services

Finally - this is an extension of No.6 really, but it is a huge challenge

7. We need a new paradigm in government where policy precedes technology

Hopefully this is pretty much self explanatory.  Either we make plans now on the assumption that this tech will 'do what it says on the tin', or do what is time-proven prudent and we wait until the tech is 100% safe in pilot form, and then wait some more for the 'early adopters' to go up the learning curve before we jump in and get the 'early majority' benefits

The danger of making plans now is that it is all wasted as the tech deployment and market penetration and impacts are nothing like expected. In fact it could do an 'electric car' on us and despite promising much, it then fails to deliver.

The danger of following the tried and trusted prudent method is that the early adopters are proved right and gain a massive business or operational advantage that quickly becomes so insurmountable that our jurisdiction loses out big time to its neighbours.  Businesses shrink or go bust, the economic vitality is lost and it all goes downhill from there.  Consider this statement in light of the 'trillion dollar money flows' argument - this won't be one sector suddenly struggling, it will be multiple sectors with a magnified impact as a result.

Because autonomous vehicles have the potential to transform society, and because I have demonstrated some money flow scenarios where market penetration is very rapid, then I recommend that each jurisdiction look very carefully at the pro's and con's of putting policy before technology.  With this tech, the normal prudent paradigm of steady and cautious may prove to be our un-doing.

There may be a middle ground that is the optimal solution - I just raise a warning that it is an inconvenient truth that technology is developing so fast, and its impacts could be so great, that a new paradigm in governance may also needed to better cope.


Whatever autonomous vehicles bring to society, I can assure you that it will not be business as usual.  

Is that another Inconvenient Truth?...

Monday 22 July 2013

Why Automated Vehicle Zones (AVZs) Might Develop



This is a poster that I presented at the TRB workshop on road vehicle automation at Stanford University (15-19 July 2013).  Please excuse the poor quality of production, but rather note the concept which generated a reasonable amount of interest amongst those TRB participants who stopped to discuss my poster.

In it I propose that if you follow the business models that result in the inevitable rise of fully automated (NHTSA Level 4) taxi fleets, and combine that with the aspirations of urban planners (reflecting the desires of many mayor's and councillors in major cities) then AVZs are a natural development of the convergence of these streams.

On the poster I assumed that Google would release their technology in 2018, with it being certified safe for unmanned use by 2020 - thus leading to the first AVZ, somewhere in the world around 2023.  But in a public session on the Friday, I asked the panel when NHTSA Level 4 technology might be deployed - and Anthony Levandowski of Google referred me to the comments made by Sergey Brin at the California Autonomous Vehicle bill signing ceremony on 25 Sept 2012.  At that time Sergey Brin intimated that the Google tech would be in public hands within five years - so in July 2013 we are looking at 2017 - that is 4 years and small change.

In which case, if I am correct with my business models and estimates for market penetration, we might see the first AVZ in 2020.  The competitive edge that any city might gain by switching to an AVZ is considerable - not to mention the quality of life, safety and emissions benefits.

Thursday 18 July 2013

Google aspire to bring Level 4 Automation to market - maybe within 5 years

At the TRB workshop on road vehicle automation yesterday (Weds 17 July) Ron Medford made it clear that Google would like to bring a NHTSA Level 4 automation (i.e. capable of unmanned) product to market.  There are clearly many issues and hurdles, so no guarantees, but the Google aspiration is for this Level 4.

Last year, 25 Sept 2012, Sergey Brin intimated that they would have SDC tech in public hands within 5 years.  Earlier this year two separate Google SDC team members have stated that they would like to have the SDC tech in public hands within 5 years.

We shouldn't 'hang our hats' on these aspirations, but at the same time I take this as a very clear advance warning signal that we are now potentially running out of time to prepare for this.

For me this is a potential 'wow' moment.  Level 4, unmanned is the 'game changer' - which is the language used by the Director of the California DMV stated at the conference the previous day.  The clincher for credibility, in my opinion, came from there being a  Senior Researcher from the White House who spoke off the record.

Level 4 is the 'quantum leap', the 'paradigm shift', the 'game changer'.  If it is less than 5 years away then we don't have much time to prepare.

The Workshop has been incredible thus far - significantly more attendees than last year and the level of knowledge and awareness is so much greater. The public and private sectors are well represented and the language/mood of some has moved from guarded/skeptical to open and even excited.