October 12th, 2019 by Frugal Moogal
I mentioned in a recent article that I added to my stock position in Tesla for a variety of reasons. Before I go on to discuss Smart Summon, I wanted to address a complaint in the comment section of that article.
It’s true — I tend to write Tesla articles. I use my business experience to examine stock price, public statements, press releases, and the health of the company. As I try to state in every article I write, I am a random voice on the internet under a pseudonym and you shouldn’t trust me without researching and verifying what I am saying with other facts — just like any investment you’re making. In fact, one of my first articles highlighted who writes Tesla articles and why they are published.
I do follow other clean technology companies, with my largest stock holding being Brookfield Renewable Partners (BEP). I’ll write about them sometime, but they aren’t doing anything like Tesla is with Smart Summon, so let’s get to that!
After my most recent article, and the article before noting that Smart Summon is actually doing a great job, I started emailing back and forth with CleanTechnica’s Director, Zach, who noted that he thought that I was overlooking the marketing aspect of Smart Summon.
And, if you have a car that can Smart Summon, you know Zach is totally right. I’ve mentioned it to a few friends, and was then met by disbelief, and requests to see the feature at some point in the future. It makes people hear about Tesla as a company and get more interested in Tesla’s products.
I think that it is also doing a few other things that are perhaps even more important — a few things being overlooked by those who are making comparisons to the chairs carrying humans in Wall-E — and I think those are the things that matter the most. The truth is, the success of Smart Summon was the biggest factor to me in deciding it would be worth investing more in Tesla, beyond just the marketing. I’ll explain why.
The first major factor is that if Tesla can get to the point of turning its vehicles into a robotaxi fleet — and I believe Tesla can — the thing those vehicles need to do the absolute best is pick people up. Let’s put it like this — if you for the first time summon a robotaxi and it drives to pick you up in a parking lot while looking like it’s going to crash into a curb, are you going to trust getting in that vehicle and taking it for a ride? Smart Summon will help refine what is ultimately the most important key to getting people into robotaxis in the future — the first impression.
Normalizing Autonomous Cars
As all of the articles declaring that Smart Summon was causing panic and hysteria proved, people aren’t yet really ready for this. Some people are certain that Teslas using Smart Summon are running people over left and right — yet, amazingly, two weeks after release, I haven’t heard of any additional issues beyond the initial three that all the “CHAOS!” articles linked two, none of which were the vehicle’s fault.
That’s a huge win, and rather amazing by itself. Tesla weathered the expected “sky-is-falling” phase of Smart Summon, and the feature is still working. As it gets better, and as Tesla produces more vehicles, more and more people will see the feature in use, and at some point in the relatively near future, cars driving around in parking lots without people sitting in them won’t be surprising. Which will make acceptance of a car driving on the highway without a driver a lot more acceptable when that starts to occur.
Parking Lots Are Hard
This isn’t just Musk saying this. Parking lots are a very difficult problem. There are often islands, different cement colors, shadows, parking spaces that may or may not be angled, people, curbs, and many other factors in parking lots. They are significantly more complex than an average road. While the car cannot currently park itself, navigating a parking lot autonomously — with a human watching it and holding a button down — is a huge accomplishment.
It also allows for the user to help Tesla teach the neural network. During Tesla’s Autonomy Day presentation, Andrej Karpathy noted that they would search for particular images to denote what was going on in them to better train the neural net. Parking lots are similar to roads, and it may be difficult for the car to learn which is which. Every time you press Smart Summon, though, I’m sure your vehicle is noting that data as parking lot data, so it can use that to learn about the different edge cases within parking lots.
As an aside — I heard about Smart Summon, read about it in the test fleet, and figured that the earliest it would really be pushed to the fleet as a whole would be in the spring of 2020. I figured those early vehicles would need to keep training the neural net to look for the huge number of edge cases out there and to solve them, a task I expected would take much longer than the optimistic time frames Musk and Tesla had given us. But, not just did it come out, the lack of videos of Tesla vehicles crashing into things is a huge sign that they can train the network faster than I predicted could be done.
Smart Summon is not perfect, but it works significantly better than I expected, and that has given me considerable faith that they are moving into fully autonomous driving at a pace so much faster than anyone else.
Smart Summon to me proves just how much further along Tesla is than anyone else in all aspects of self-driving vehicles. It may seem like a relatively useless gimmick right now, but it is one that I feel is critical to the future of autonomous driving and the future of Tesla as a company.