Autonomous vehicles seem to continue to fascinate entrepreneurs, hardware developers, and automobile manufacturers around the globe, although it also seems obvious that such vehicles might not be completely ready to roam the open roads, and since June of 2021,  ADS crashes involving autonomous systems must be reported to the National Highway Traffic Safety Administration if the system was in use withing 30 seconds of the crash and property damage or injury occurred.  Requirements get more stringent with each level, although there are so few vehicles with level 3+ systems due to state and other restrictions that there is little crash data.
Regardless of the available data, it would seem that while ADS and ADAS (level 2 ) systems are getting more sophisticated, they still require both careful use and full driver management.  Level 3 and above need more time for refinements, particularly lots of time in a controlled environment.  We think of that test zone as a very large parking lot where there are a few vehicles with test drivers rather than a public road.  What if you could have a parking lot that was hundreds of miles wide?  Wouldn’t this be the ideal testbed for an autonomous vehicle?

Instead of a vehicle, how about a ship?  Not an aircraft carrier or a giant cargo ship, but smaller, scalable, and relatively inexpensive fully autonomous craft that could be trained and deployed on the high seas with little chance of potential accidents.  The US Navy has been working on such a program for years, although including air (drones) and ground systems, but the accelerated conflicts that the US Navy faces daily, such as those in the Red Sea or in the South China Sea, make the necessity for such systems an absolute. 
Recently a company, Blue Water Autonomy (pvt) came out of stealth mode after raising $14m of seed capital and indicated that they have been developing such a ship for the Navy.  The company, founded by a former Navy officer who became the founder of 6 River Systems[1] (pvt) (sold to Ocado Group (pvt)) after a stint at Amazon (AMZN) Robotics, a former VP of Engineering at iRobot (IRBT), and an MIT grad who helped Ukraine collect front line data with drones.  Given this level of expertise and experience, the company developed a concept ship design in less than a year and has already begun saltwater testing of a 100 ton prototype for the US Navy.
The idea behind the autonomous ships is to make the design scalable and low-cost, allowing for modular payloads, giving the ship the ability to perform a wide variety of services.  But why does it need to be autonomous? 
China is ahead of the US in naval shipbuilding, making competition at the carrier and cruiser level difficult, but by employing large numbers of smaller autonomous ships, long duration patrols, surveillance, border patrol, and electronic warfare (jamming, etc.) can more easily be accomplished, without risk of life.  Such ships could easily be outfitted for logistics or resupply, search & rescue, or even mine recovery, but there are some things that autonomous ships can do that are far different than what goes on in more traditional naval warfare. 
If you have seen New Year or 4th of July celebrations recently, you have probably seen drone ‘swarms’ that can spell out words or create images using hundreds of linked drones.  Autonomous ships, especially small, less detectable ones, can do the same, linking themselves electronically and moving together as if they are a larger vessel.  This technique can be used to fool adversaries into thinking they are outnumbered or outgunned, when the ships involved are small and less far less equipped.  This ability to link systems also allows these ships to operate without human supervision in difficult circumstances or when given a specific task, such as a general command like “Return to base, restock fuel and return to location to refuel XYZ”
While we understand that the lure of being the supplier of millions of autonomous vehicles is unbelievably attractive to both large and small companies, it is also one that will take years to refine enough to where autonomous vehicle accidents will not be any more significant that those caused by human drivers.  It seems a smart idea to work such systems in a larger ‘parking lot’ like and ocean, where the chances of hitting something are far less than on the East River Drive or the 101.  JOHO.


[1] Warehouse robotics

Xiaomi (1810.HK) is a diverse company, although best known for its line of smartphones which they becan marketing in 2011.  Since then the company has expanded rapidly through very aggressive marketing and has become the 3rd largest smartphone vendor globally, but Xiaomi has also entered a number of other markets including TV sets and a number of other consumer electronics items, but last year the company decided that it was also to enter the electric vehicle market by investing $10b US over a ten year period, so it is no surprise that the company has been using its Hubei Changjiang Industrial Investment Fund to invest in a slew of companies in both the automotive space directly and the semiconductor space as a source of silicon products needed for both the automotive side and the CE side of the company. 
Last year the fund, which is also funded by the Wuhan government, made investments in 42 companies, far above the few investments the fund made since its establishment in 2017, with two in 2017, two in 2018, five in 2019 and 41 in 2020. Among the investments are 24 semiconductor companies, 11 new energy vehicle companies, and a number of photovoltaic and AI companies, with only 3 investments in companies related to the smartphone business.  So far this year the fund has already made 3 new investments in Si Tan Technology (pvt), Foote Technology (pvt) and Chengxin Semiconductor (pvt).
Below is the list of companies the fund invested in last year.  We apologize for the few that did not translate or were translated incorrectly but you get the idea; Xiaomi is making investments in its future.

​Beijing based Xiaomi (1810.HK), primarily known for its smartphone business, is adding to its autonomous driving capabilities with the acquisition of ~71% of Deepmotion (pvt) common and ~29% of the company’s preferred stock for a bit over $77m in cash and stock   Deepmotion was formed by a number of Microsoft (MSFT) scientists in 2017 to develop object recognition and behavior prediction software for autonomous vehicles.  The company had been financed by Redpoint Capital (pvt), Andromeda Ventures (pvt), and Source Code Capital (pvt).  The company lost $1.8m last year and $2.7m in 2019.
Xiaomi has indicated that it has been aggressively recruiting engineers and experts after initiating its electric vehicle project in March, and has taken or raised stakes in Geometrical Pal (pvt), company that specializes in radar and light based sensing for positioning and route planning, and Zong Mu (pvt), a company developing localization and mapping software.  A number of Chinese CE companies have been branching out into the autonomous vehicle market over the last year or so, including Huawei (pvt) who has agreed to invest $1b in the segment, which has seen estimates grow from $54.2b in 2019 to $556.7b in 2026, and Baidu (BIDU), and with the challenge of trying to top the US and Europe in that segment’s development, China is certainly not going to hold back spending for technology.  While spending will certainly be a focus, we are far more interested in the progression of driver assistance software and hardware, which we see as a far more viable way of developing the automated driving market than working toward a standalone autonomous vehicle.

While the COVID-19 pandemic has been a dreadful experience for humanity, it has perhaps accelerated a few lifestyle changes that had already been gaining some momentum.  One such is grocery shopping, a tortuous task involving examining and touching things that many other people have touched, and trying to anticipate what you or others might be desirous of days in advance.  Of course there are some pleasurable moments during grocery shopping, such as discovering that Entenmanns is having a BOGO on chocolate donuts or finding mismarked artichokes selling for $0.28 instead of $2.28, but that is usually offset by screaming children in auto-like shopping carts or those wonderful folks who are on their phones while wandering aimlessly through the aisles, or those that pay with cash and coins and don’t bother to find their money until after everything is rung up and packed and the line stretches across the store.
Grocery home delivery services are certainly a step in the right direction, although leaving produce selection to a 19 year-old who has purple hair and two nose rings can be iffy, but even better are the services where you place your order on a site and they bring your groceries to your car while you remain in the parking lot.  This does give you some time to check for missing items or to actually speak with the person that collects your items, but soon that will all change if Elon Musk gets his way.
As part of Project Dojo, the designation given to Tesla’s (TSLA) AI chip development program, Mr. Musk introduced yesterday an enlightened version of  the Tesla Bot, a 145 lb humanoid robot, that is not only ‘friendly’, but can free you from ‘dangerous, repetitive, or boring tasks’.  The bot can carry ~45 lbs of ‘stuff’.  Intentionally, the speed of the robot max’s out at 5 mph so if the bot suddenly decides it is interested in world domination you can run away from it, and even if it catches you, you should be able to overpower it.  Rather than a face, the bot will have a screen to reveal ‘useful information’, will be built from lightweight materials and have over 40 actuators to allow freedom of movement.
Most important are the cameras, neural networks and systems used in Tesla’s ‘self-driving’ vehicles, which means it should be able to avoid running into almost anything other than emergency vehicles parked nearby with their lights flashing.  Mr. Musk expects that such a bot should be able to perform tasks without explicit pre-training, easily responding to the sentence, “Go to the grocery store and buy chocolate donuts and artichokes”, singling out the arduous task of grocery shopping in his presentation.
We are all for the technology, which unfortunately has no timeline attached, although we are a bit hesitant about using the same collision avoidance software and hardware that is used in Tesla vehicles today, but more importantly we want to know how these new Tesla bots will deal with their elderly brethren, such as Marty, the Stop & Shop™ robot.  Will there be a caste system where the Marty’s of the bot world will have to clean and oil their more sophisticated colleagues, or will there be a place where the ‘Marty’s’ go to retire, spending their days sitting by the window, waiting for someone to come by and spill a soda on aisle 14?  Hopefully the Tesla bot will also have ‘enlightened’ abilities that will give it a ‘humanized’ moral compass, sort of a ‘woke bot’ that will be able to see us humans as the nice folk we really are and not a race of mindless couch potatoes that spend all day checking their Twitter (TWTR) account, or maybe we should just stick with Marty and get our own groceries.  He is very non-judgmental as long as you stay out of his way and don’t block the doors to the stock room.

SAE, the name for what used to be called the Society for Automotive Engineers, is a well-known standards organization for the automotive space, and while such standards are not legally binding they are widely used across many industries in and around the automotive space.  One area that SAE has set standards for is Driving Automation, an area that falls into the realm of consumer electronics, or at least on the edge.  In May SAE released an update on “SAE Levels”, which it has proposed as the standards for defining levels of automation, which is a topic that will continue to garner attention over the next decade.  While we are not at a point to comment on the validity of the standard and how other countries and organizations might view it, we thought it interesting and educational.