Can Robotaxis Survive A Pandemic?

Almost all the self-driving car fleets are out of operation right now. The primary reason is that testing these cars is not an “essential service” and so the safety drivers who supervise them are not allowed to come to work. Most companies use a team of two people in each car, which would create a disease risk, and to top it off, the roads are empty and this unnaturally easy, making testing less valuable.

Questions arise about what happens in the robocar world if we need to suffer another pandemic in the future. The Covid-19 crisis took place before the commercial deployment of robocars, but that won’t be true later. I covered some issues in my article on delivery robots, but let’s consider what we’ve learned this year:

1. Uber/Lyft
   LYFT
   and Taxi rides are way, way down, and not just because all travel is way down
2. Transit ridership has cratered, ans service has been cut
3. Delivery demand has vastly increased
4. People worry if drivers might be infectious
5. People don’t want to touch things, from seats to scooter handlebars, that others are touching

All travel is way down, but people are particularly suspicious of travel with other people, be it transit, or riding with a driver. They are also concerned about sitting down in a vehicle where somebody else just sat. With parking plentiful, there are incentives to go back to using your own car even if you previously used something else. Scooter services like Lime and Bird have also suffered major declines.

As noted in the delivery article, while delivery robots are always good in a time when there is a massive surge in demand for delivery, it’s pretty easy for a driver in a van with an automatic door to never have to touch the packages, just like a delivery robot. With massive unemployment, human driven vans are probably the best answer to the delivery demand surge.

With taxi service like Uber or Lyft, which are almost never designed to isolate the driver from the passengers, removing the driver is an attractive proposition. There’s still the problem of the person who just sat in your seat and coughed on the armrest.

Robotaxis, it turns out, will need to be designed for long lifetimes —300,000 to 500,000 miles or more. Since nobody wants to ride in a car seat that’s had people in it for 200,000 miles, they will need to be designed so that replacement of the upholstery is a planned event. While most of the time, people will want comfy seats, at the start of any epidemic, the seats can be swapped out — if this is planned for — with easy to clean transit style seating. This is much harder to do in the privately owned cars of the Uber fleet. The cars can make regular visits, many times a day, to sanitation stations (formerly known as car washes) for a top grade cleaning, and in between, wipes and gloves can be provided for all passengers.

If that’s not enough — and companies will do what it takes rather than go out of business — experimentation can be done with the installation of things like intense UV light, or even spraying hot soapy water on the interior at high pressure, followed by a rinse and dry for self-cleaning. (Drain holes would be opened in the floors.) This means taking each vehicle down long enough to dry between rides, but demand will be down. Internal HEPA filters and the air recirculation system would also go into high gear.

There are also signs that certain coatings, such as nanoparticles of titanium dioxide, can make it easy to be rid of coronaviruses. Copper, which is very good for killing bacteria, still takes about 3 hours to dispose of SARS-CoV2.

When the crisis is over, the nice interiors can be restored.

It’s a bit tougher for the scooters. Being low cost is a big part of that business. Easy to clean handlebars and a box of wipes may be sufficient, but they just lose a lot of their value if roads are empty and travel is low.

Testing

There is an irony that, with all the work to make these cars safe on regular crowded streets, the simpler problem of streets as empty as these may already be solved. Almost, because right now there are actually more pedestrians on the streets than ever before in many areas, as going out for a walk is the only permitted activity for many.

Testing with two safety drivers creates a risk, but in theory, many teams are ready to test with just one safety driver. This is a controversial question, since having only one safety driver contributed to Uber’s fatality — the safety driver at fault would probably not have spent her time watching a streaming TV show if another person were in the vehicle. Even if this happened, the other driver might have been looking at the road and prevented the incident.

At the same time, Tesla
TSLA
cars in Autopilot are driving millions of miles with just one untrained supervisory driver, and putting down a reasonable (though not perfect) safety record. That’s at the simpler highway task – we may soon see how it works for Teslas on city streets — and it’s with drivers who know they are using a lower grade system that they should not be complacent with. Even so, it suggests that the Uber fatality was a truly negligent anomaly, and single-driver testing with reasonably mature vehicles (like Waymos) is OK. Indeed, every car has to go through a stage where one driver would be fine before it’s ready to go with zero supervision.

Of course, these vehicles still need occasional remote assistance. With work, the remote assist stations could be moved to homes with top quality bandwidth, though they are usually pretty fancy stations, not just a home laptop or desktop.

The bigger problem is that the purpose of testing, for anything but the youngest teams, is to go out there and have the car face challenging and new situations. Aside from the extra pedestrians, today’s lockdown roads are almost empty. Even if testing could be done with no virus risk, it still doesn’t meet the description of an essential service, unless it were taking people for rides — which has the virus risk discussed above.