Breakfast Bytes Blogs

The first half of my predictions for 2021 was yesterday. You should probably start there if you ended up here without having read it.

Automotive

This seems like this is going to be the year that autonomous vehicles start to go somewhat mainstream. Not level 5, with no steering wheel, and probably not in all road and weather conditions. Waymo (Google/Alphabet) is deploying autonomous taxis with no safety driver in Phoenix. Phoenix is almost a best case, with low-density, broad roads, and it rains rarely enough that even if taxi service is suspended then, that is a tiny percent. Waymo has driven over 20M miles, mostly with safety drivers until recently.

Tesla has millions of miles with its SDC (self-driving-computer, see my post Tesla Drives into Chip Design for more details). Tesla has driven over 2B miles autonomously (end of 2019).

Cruise's numbers, again for 2019, is that they drove nearly 1M miles autonomously that year, with one disengagement per 12K miles (for most people, that would be less than one disengagement per year). It said in October that it would introduce autonomous vehicles with no safety driver by "the end of the year". So, for all practical purposes, 2021.

In computer animation, there is something called the "uncanny valley", between where the graphics are obviously just graphics (like Toy Story) and where the graphics are so good that you can't tell real people from CGI people. In between is this weird gap where the graphics are almost good enough—but not quite, so it feels like something is a bit off. Autonomous driving seems to have its own uncanny valley. The autonomy needs to be good enough that either the car can handle it, or it can give the driver a lot of warning that something is wrong. What it can't do is tell the driver to take over ("disengagement") at five seconds' notice, since a person who has not been watching the road in detail just can't switch on that fast. So autonomy either needs to be more like ADAS (Toy Story) or good enough to get through the uncanny valley that any issues can be smoothly transitioned.

IoT

To be honest, I don't know that much about IoT. It is such a fragmented market, and only a few markets such as smartwatches can support the cost of designing a special chip, so from the point of view of the semiconductor market, it is a standard product market with all the smarts provided in the software. Everything says that the most important markets are industrial IoT, since there is a clear value proposition there. But I know even less about predictive maintenance. I have recently been taking an interest in agricultural automation, since I worked a few summers on a farm as a teenager (see my post Agricultural Electronics). But I don't know if that counts here, or if it should be counted under autonomous vehicles.

The big category is smartwatches, with Apple far and away the market leader, and fitness trackers as the second category. Fitbit is the market leader and Google has announced an intention to acquire them but it is tied up in regulatory approvals. But let me make a prediction. This year will be when someone solves the holy grail of medical devices and creates something like a smartwatch that can take continuous blood pressure measurements without requiring a cuff. I've seen it done with a sensor on a smartphone (see my post Léman Micro Devices: Blood Pressure on Your Smartphone) but I've not actually seen any phones using the technology, but with MWC Barcelona virtual this year, it is hard to wander around and try and pick up on trends.

How about robots? That is for sure IoT. If you count all the fulfillment robots inside Amazon warehouses, they are probably the market leader in practical robotics (or do they count as autonomous vehicles? These categories are so fluid).

Boston Dynamics is some sort of leader in the robotic space, but struggle to be profitable. They were founded way back in 1992, acquired by Google in 2013, sold (or a controlling interest) to (Japanese) Softbank in 2017. Softbank has announced that they are selling a controlling interest to (Korean) Hyndai. But this will require regulatory approval since there are lots of ties between Boston Dynamics and the US military, although Google cut a lot of that back, one reason that Google never managed to make them profitable. But they produce the cutest videos! Here's this New Year's one, Do You Love Me?

Security

A couple of days ago, I wrote about the SolarWinds exploit. See my post The Biggest Security Breach Ever for more details. I think that this might be enough of a wake-up call to cause some legislation. Although I thought that when the NSA had to admit that it had been breached and the attackers had stolen a lot of its hacking tools, or when all the security clearance paperwork of the OPM was stolen. If the NSA can't keep a secret, what chance do commercial companies or less specialized government departments have? This latest breach is going to take years to determine what happened, and it is probably impossible to fix completely.

So look for lots of security news this year going higher profile, and mainstream journalists starting to pay attention. Although, as I pointed out in the post linked to above, you often have to look to the foreign press to find out what is going on in the US.

Part 3

The third (and final) part tomorrow.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.