Tesla Autopilot Controversy: My Personal Experience – I Love It!

For all of the controversy over Tesla’s Autopilot features, I find it a big help. For example, I was taking an evening course at Swarthmore College on How to Talk to Climate Change and Evolution Deniers. The drive was an hour and 45 minutes each way. Without Autopilot I arrived home exhausted from the strain of driving at night. By contrast, with Autopilot I arrived home alert and almost fresh. It takes a lot less cognitive and physical energy to supervise the driving process than to do it.

No I’m not one of those crazy people who sleep in the back seat or watch videos or text while driving. You have to keep your hand on the steering wheel, exactly as Tesla demands, both in the instructions every time you turn Autopilot on, and because it yells at you if you leave your hands off, and then will turn off Autopilot until you restart the car.

Even though Autopilot is designed for highways, I use it extensively in local driving – the vagaries of local roads can confuse Autopilot, especially missing lane lines and sharp curves over a hill. But it still saves my personal energy, and is fun.

Coming out in the morning to see what new features Tesla has downloaded overnight is always fun. For example, now my Tesla automatically opens the garage door when I arrive home, and closes it when I leave.

Conflict: Who Is Going to Win the Race to Build Self-Driving Cars?

Wired published this article on March 3, 2017: Detroit Is Stomping Silicon Valley in the Self-Driving Car Race, and the next day, the New York Times published this article: Tesla Passes Ford in Market Value as Investors Bet on the Future. Clearly Wall Street doesn’t agree with Navigant Research, although of course their objectives are different. Navigant placed Ford as the leader in the race, with GM a close second, and Tesla in 12^th place. I’ll make a wild guess that the big car companies funded Navigant’s study J.

Of course Navigant didn’t have access to Tesla’s 1Q2017 results: a record number of vehicles produced, 25,000, for a 69% increase over last year. And they probably don’t believe Elon Musk’s plan to build 500,000 Teslas next year. Bloomberg is skeptical that Tesla can ramp up production that quickly, even though Elon Musk says orders are going out to suppliers for Model S production of 1,000/week starting in July, and ramping to 5,000/week by the end of this year. Bloomberg notes this would mean outselling BMW 3’s and Mercedes C’s combined.

This conflict is really amusing, especially given my previous post!  I can hardly wait to see how this turns out

“Silicon Valley Decides It’s Just Too Hard to Build a Car” – What About Tesla?

Bloomberg just published an article Silicon Valley Decides It’s Just Too Hard to Build a Car, by Keith Naughton , Alex Webb , and Mark Bergen, October 25, 2016. That make Elon Musk’s achievements with Tesla Motors that much more amazing! Tesla is a Silicon Valley company, and Tesla is producing amazing cars. The article points out that the Silicon Valley companies can’t get the real-world driving data that the conventional car companies have. And conversely the conventional car companies can’t hire the software experts to build the systems needed for the autonomous cars of the future because Silicon Valley firms are more attractive places to work. Yet attempts at partnerships between companies in the two industries remain stalled.

Meanwhile, Teslas have driven over 100 million miles with Autopilot active, so Tesla is gathering far more detailed data than the conventional car companies. By contrast Google’s much-touted self-driving car program has driven 1.5 million miles over its entire history – Teslas log 2.6 million miles a day on Autopilot. And Google says Google’s Robo-Cars Hit 2M Miles, Confirm Driving Is Dadgum Tricky. That’s why our proposal is to have Autonomous Vehicles run in enclosed Autonomous-Ways, or A-Ways, excluding human-driven vehicles, pedestrians, animals, debris, and weather. A-Ways greatly simplify the requirements on the autonomous driving system, and greatly improve the safety – in 94% ± 2.2% of crashes the driver was the critical reason, NHTSA DOT HS 812 115, February 2015. And even better, the A-Way can provide power directly to the vehicles, essentially eliminating batteries, which are such a drag on electric vehicles.

Having worked in Bell Labs, which at the time was part of the largest private company on earth, AT&T, and subsequently in several startups, I have a perspective on the challenges facing the conventional car companies. I even directed modeling efforts as part of studies leading up to the AT&T divestiture in 1984. Bell Labs was a great place to work back in “the day” but now startups are far more fun and satisfying.  

My view is that it is incredibly difficult for a large company to make the transition to drastically different technologies. On March 28, 2008, I was in Sacramento, California to make a statement at the California Air Resources Board meeting on the Proposed Amendments to the California Zero Emission Vehicle Regulations. After the session I was being interviewed on video, and I predicted that the big 3 car manufacturers would go bankrupt, as I had been predicting quietly for several years – the interviewer promptly left, presumably because he thought I was a crank. Thirteen months later I was proved right, ok 2 out of 3 isn’t bad, and Ford only dodged the bullet by presciently mortgaging all its assets in 2006. My comments weren’t based on the economy, that was just the trigger, but that the companies had ignored the changes taking place in the car industry, especially the need for higher quality, and the difficulty in changing a large company.

Now I predict that the conventional car companies will have great difficulty adjusting to the combination of autonomous driving, and the smaller numbers of vehicles due to sharing. I don’t know whether they will go bankrupt or just fade away to bit players. All those companies are making claims they will have autonomous cars by a variety of unrealistic dates, yet have no real experience. I’ve been managing research and development for my entire career, and built two high tech companies, and Tesla is the only one doing it right, IMHO.

Tesla tries things in the real world, and then rapidly modifies their system, based on the new information. That rapid cycling is essential for building complex systems. The conventional car companies are still stuck on annual model cycles, which might be appropriate when you are just bending metal, but don’t apply to high tech systems. Look at how long it has taken the car companies to come clean on the airbag and other disasters, and how long it takes them to actually fix the problems – this is a recipe for doom and oblivion.

Chevy now says the Bolt won’t be production limited to 30,000, and they could meet demand of 50,000 cars per year, reasoning that the annual sales of electric vehicles in the US so far is barely 30,000. Did they somehow miss the 276,000 people who signed up for the Tesla Model 3 in the first two days? Or perhaps they are siding with the naysayers on whether Tesla can meet the demand, despite the videos of their automated production line? Or are they conceding that they can’t compete with Tesla? Go Elon!

Innovation: Continuous Convoys and En Route Sequencing

Story: While riding the train from Long Branch to New York City, I was annoyed with all the stops along the way. Some of the time I gained by not driving and being stuck in traffic, I spent thinking of different ways to make the trip faster and more efficient. I imagined a wide variety of schemes, including: enclosing the tracks, building a second level above the tracks for high-speed vehicles, and so on. My thinking stuck in the form of the existing trains: an engine pulling a dozen cars. After several years of interesting, but fruitless, ideas, it suddenly hit me that if I just made each car an Autonomous Vehicle, there are many new possibilities.

Follow a similar approach to freight trains: connect many vehicles together into a “Convoy” of Autonomous Vehicles. As you saw before, having a long Convoy of vehicles is much more aerodynamically efficient than separate vehicles. 

One of the efficiency challenges of conventional trains is that the whole train stops at each station. This slows down the everything that isn’t exiting at that station, and is also less efficient because of the mass of vehicles stopped and then reaccelerated needlessly.

To achieve the advantages of non-stop service, while also providing many intermediate stops, we have invented, what we believe is a new approach to fast, efficient transportation.

Innovation: Continuous Convoys

We call these Continuous Convoys because the Convoy as a whole never stops (except at the end of the A-Way), only individual vehicles stop. As the Convoy approaches a Stop, the last Vehicle in the Convoy (A) detaches from the Convoy and decelerates to arrive at the Stop. The other Vehicles in the Convoy (B-D) continue at normal speed.

Simultaneously, as the Convoy approaches the Stop, the Vehicle waiting at the Stop (E), accelerates to join the front of the Convoy.

The next diagram shows two consecutive Convoys passing through the Stop. Note that Vehicle A, which was the last Vehicle in the first Convoy, and stopped at the Stop, joins the front of the second Convoy.

You have probably noticed a challenge with this approach: if you want to get off at a Stop, you have to be in the last Vehicle in the Convoy.

Innovation: Continuous Convoys and En Route Sequencing   

The solution to this challenge is to have people, and other loads, within the Convoy Vehicles move to the appropriate Vehicle for their Stop – we call this Continuous Convoys and En Route Sequencing.

As the Convoy approaches a Stop, everything destined for that Stop is in the last Vehicle in the Convoy (Stop 1, Purple in the diagram). At the appropriate time, the last Vehicle detaches from the Convoy and decelerates to arrive at the Stop, and the loads exit. The other Vehicles in the Convoy continue at normal speed.

Simultaneously, as the Convoy approaches the Stop, there may be a Vehicle waiting at the Stop, carrying loads for various Stops ahead. At the appropriate time, this Vehicle accelerates to join the front of the Convoy.

Once this Vehicle has connected to the front of the Convoy, the contents of this Vehicle move to the Vehicle destined to go to their particular destination (Stop 2, Blue; Stop 3, Green; Stop 4, Orange; Stop 5, Red) – we call this En Route Sequencing.

Multi-Layer Nesting and En Route Sequencing

We showed smaller Autonomous Vehicles inside other larger Autonomous Vehicles in the diagram of Nesting. These smaller Vehicles can also use En Route Sequencing inside the larger Vehicles; then the larger Vehicles use En Route Sequencing inside the Continuous Convoy Vehicles, so that everything arrives at the proper Stop.