What Are The Biggest Challenges Uber Will Face With Self-Driving Cars?

Uber is working hard to get self-driving cars on the road, but I’m not convinced it will be happening any time soon.  This week, in a coordinated PR move, Uber invited 50 reporters from around the country to Pittsburgh to test-drive their ‘self-driving cars’. It made for some great headlines, but these cars were hardly self-driving.

They currently require not one, but two Uber engineers in the car with you while you ride and every account I read had multiple reports of the human driver having to take over.  So although the storyline sounds cool, we’re still a lot further off from self-driving cars than Uber would like you to think, and here’s why.

Uber is currently working on self-driving cars, but there's no guarantee the efforts will pay off. Here's why self-driving cars are facing an uphill battle.

Updated September 22, 2020: Recently, GM announced an investment with Lyft to develop autonomous vehicles. We’ve updated this article to reflect the news.

General Motors will invest $500 million in Lyft, giving them a valuation of $5.5 billion. This helps push Lyft forward in their battle against Uber. However, Uber’s valuation currently stands at $62.5 billion. 

The plan is for GM and Lyft to work together to develop self-driving vehicles for passengers to call on demand, though when this grand future may happen is unclear. In the immediate future, GM will be providing Lyft with cars for Lyft drivers to rent short-term from various hubs around the U.S. 

Uber has already been investing heavily in self-driving vehicles, and the company seems to believe the realization of autonomous vehicles being active widespread on U.S. roads is likely going to happen within 5-15 years from now. 

GM President Dan Ammann said, “The U.S. is our home market and it continues to be our largest market and we think this is the right place to begin the journey.” 

One looming question, however, is will autonomous vehicles be widely accepted by the public within 5-15 years? With situations like what happened in Tempe, Arizona in 2018, it’s hard to say. 

The safety driver who was monitoring the vehicle at the time of a fatal accident was just charged with negligent homicide. The self-driving vehicle, monitored by a safety driver, struck and killed a pedestrian. It was nighttime and the pedestrian was walking a bicycle across the street at the time of the accident.

Of course, many would say there have been far more fatal accidents with drivers in the vehicles versus autonomous vehicles—whether they are monitored or not—but many considerations still need to be taken:

  • Will these cars be available in adverse conditions such as rain or snow?
  • What if the car takes a wrong turn or delivers a passenger to the wrong address?
  • What if the previous passenger got sick in the car? Who would clean it up before the next fare if no one knows it happened?
  • How will driverless vehicles handle hills or even mountains? Or especially twisty roads?
  • Will it know not to drive away if a passenger puts something in the trunk? What if a passenger forgets something in the back seat?
  • Does it know if a minor is trying to take the ride instead of someone 18 or older?
  • What if a drunk person passes out in the back seat and never leaves the vehicle at the drop off point?

 Waymo has been testing self-driving cars in Phoenix since 2017, sometimes with and sometimes without a safety driver. But, what are the conditions? They have mostly operated in a geofence of 50 square miles. They are allowed to drive at night, but not when it’s rainy or there’s a dust storm. 

Read about this person’s experience in a Waymo self-driving vehicle

Our Take

As an Uber/Lyft driver, I know a lot goes into each interaction with a passenger. Anywhere from helping an elderly woman get into the vehicle, helping load up luggage, cleaning up after accidents, and so much more. 

In a world where human interaction is stripped from so many of our lives, it would be a shame to take away even more. I haven’t loved every ride I’ve given and I haven’t loved every ride I’ve been on, but I appreciated having a lively conversation while I earned money to pay my bills. 

Autonomous vehicles remove the human part of the experience as well as possibly put people out of work. 

Personally, I’m also not quite ready to trust a vehicle to do its thing all on its own and get me where I need to go and feel safe while doing it. I would be watching the traffic around me like a hawk, not trusting that the car will see the person running across the street or the sudden stop of the vehicle in front of us or that jerk trying to back up when there’s no room. 

I realize change is coming and it’s likely going to happen during my lifetime, but it doesn’t mean I have to trust it just yet. 

Our previous article and analysis is below.

Self-Driving Technology Is Tough

As of today, Uber’s self-driving cars can’t handle bridges, pedestrians, snow, etc.  The list goes on and on but in reality, these problems are all very solvable.  It may take some time to figure it out but I think the technology portion of self-driving cars will actually be the easy part.

Uber is investing heavily on the technology side to solve these problems and recently acquired Otto, a start-up focused on retro-fitting trucks with self-driving technology.  Uber knows other companies like Google and Tesla have a big head start, so it’s a good bet Uber will be investing even more going forward.

In October 2020, Motional, a joint autonomous vehicle venture from Aptiv and Hyundai, announced it would be resuming its self-driving mobility service with Lyft in Las Vegas. Unlike the tests by Waymo in Phoenix, these self-driving cars will not be fully autonomous, and ‘safety drivers’ will be behind the wheel of every trip.

But How Will Self-Driving Cars Deal With Uber Passengers?

I’d say 95% of Uber passengers are great, but there are some who make the job beyond frustrating.  They don’t care about your car, your property or your time, and those are the ones who will cause some serious problems for self-driving cars.  Here’s what drivers have to deal with on a daily basis:

  • Incorrect pin drop: During the Party Hours (Friday and Saturday nights), it’s not uncommon for passengers to not be where they say are.  A lot of them expect you to drive to them and ignore all safety laws while you’re doing it.
  • Mapping issues: It’s pretty common in downtown areas with tall buildings for GPS to go haywire. Uber currently relies on Google Maps for routing/destination info, but if Google is developing its own fleet of self-driving cars, how long before they cut off Uber’s access?  Google made a big change this week to their rideshare option on the Maps app and they now also list competitors like Lyft and Gett under the rideshare option.
  • Contacting your driver: This is the part that scares me most since we all know how poor Uber’s customer service is.  What happens if there’s any type of problem with your ride and you have to get in contact with your self-driving car (see above)?  We already joke that a lot of Uber’s CS responses seem automated and robotic but now they really will be!
  • Dealing with drunks: I affectionately refer to Friday and Saturday nights as the Party Hours but for good reason.  Demand for rides peaks on the weekends and most passengers are out drinking.  That means lots of instances of riders making you wait, stuffing in more than four people, bringing in alcohol, opening the doors while you’re driving, fighting in the backseat, throwing things out the window…  See where this is headed?  A lot of times, drivers can be more babysitters than anything and, given free reign, I think passengers will really take advantage of being in a self-driving car with no supervision.
  • Lost items/stops/drive-thrus: It’s not uncommon for passengers to make the occasional request to stop for a few minutes, transport a package or hit the drive thru.  These situations aren’t common but I don’t know how easy it will be for a self-driving car to handle the litany of requests that a human driver can handle with ease.
  • Low ratings: Drivers are deactivated if they fall below a 4.6 star rating but what happens if a self-driving car falls below 4.6?  Will Uber deactivate it? [only joking, but seriously…]

Humans may never be able to match the driving skills of a robot, but there’s a lot Uber drivers do that can’t be programmed into an algorithm or replicated by a computer.  And I’ve always felt that because of the large disconnect between Uber corporate and their drivers, the former doesn’t realize much of this.  A self-driving car can be programmed to handle all of the above situations, but how will it affect the Uber experience and the Uber brand?  If there’s nothing to distinguish Uber’s brand from the competition, why would you ever take an Uber?  Why not take a Lyft or a Google rideshare or a Tesla rideshare?  That’s a big risk for Uber in my opinion.

And then there’s the safety aspect. A recent article from Andrew J Hawkins from The Verge unveiled some interesting statistics about Waymo’s self-driving car program in Phoenix, including:

Waymo said that it was involved in 18 crashes and 29 near-miss collisions during 2019 and the first nine months of 2020… The company said that no one was seriously injured and “nearly all” of the collisions were the fault of the other driver.

While it’s unclear (the article does not say) how that compares to how many crashes and collisions occurred in Phoenix in the same timeframe (and with human drivers), people will still be leery of any numbers like this.

Some People Might Actually Prefer Human Drivers

Believe it or not, some passengers actually like talking to their drivers, and I think they would prefer a human driver over a robot.  Unfortunately, based off how many Uber pax get in the car and go straight to their phones, I have a feeling that human drivers will be in the minority from a demand point of view.

Uber’s largest expense right now is the driver, and once they can replace them with a self-driving car, self-driving rides are going to get a lot cheaper.  So although there may be shortcomings to a robot driver, they’re going to be a lot cheaper than a human driver.  And when you compare the cost side-by-side, you can imagine which one the average rider will choose.

Luddites Will Be Pissed

Any time industries are disrupted or new technology threatens the existence of a class of workers, emotions tend to rise.  We’ve already seen this play out in the US with Uber upending taxis, but that was nothing compared to what protesters in places like Mexico City and France did to Uber drivers.

I’m surprised the future of driving jobs and more specifically the gig economy as a whole was not a bigger topic in this election cycle, but no doubt it will be in the future.  A majority of Uber drivers work less than 10 hours a week though, so it’s going to be hard for them to risk going to jail to destroy some self-driving cars.  However, there are a lot of people who do work full-time jobs that depend on this type of work, and as self-driving cars become more ubiquitous, who knows what they could be driven to do.

It’s easy for Uber to tell drivers, “sorry, this is the way of the future” but when you’re someone who depends on your Uber driving job to literally pay your rent and put food on your table, the last thing you’re going to want to see on the road is a self-driving car.

A New Business Model For Uber

Unlike a traditional taxi business, Uber doesn’t own any of the vehicles in its fleet.  Instead, they rely on individual drivers and fleet owners to provide a majority of the vehicles on the platform.  This allows them to avoid vehicle depreciation, insurance costs and even maintenance.  But with a shift to self-driving cars, that’s all going to change.  Uber will either need to become a fleet owner themselves or work with a third party fleet owner to contract all of their self-driving vehicles.  And they’re definitely not going to be manufacturing any vehicles since they’ve already started to partner with companies like Volvo on that.

So who’s going to own all of the cars in its fleet?  Uber could contract with a third party fleet owner to handle all of this but managing a fleet of hundreds of thousands of cars (much less self-driving ones) is going to be capital and labor intensive.  At just $50,000 a self-driving car, that would be $5 billion for 100,000 cars.  Why would anyone invest $5 billion in self-driving vehicles for Uber when they could just build the app themselves for a million bucks and reap all the profits?

The more likely situation involves Uber being a majority stakeholder in their self-driving fleet, but that would mean they’d now be responsible for:

  • Capital: Uber will need LOTS of money to finance the fleet which could be challenging but if the ROI from eliminating drives is there, there’s no reason they won’t be able to get it.
  • Maintenance/Repairs: A typical full-time rideshare driver puts 1,000 miles a week on their car and self-driving cars will likely do more.  That means frequent oil changes/brakes/etc – not to mention mechanical failures.  There’s a reason why rental car companies sell their out of service vehicles to private owners and dealerships.  Is there a secondary market for used self-driving cars though?  I don’t think so – so that’s another cost that Uber will have to eat and factor in to their numbers.
  • Service: People tend to treat rental cars like shit and you can imagine that passengers will do even worse to a self-driving Uber car.  So there’s going to be plenty of service needs required in order to keep these cars presentable.  Today, drivers are the ones who vacuum their cars before every shift, get a car wash once a week and sweep for trash after every ride.  But now that all has to be done by Uber, so there will have to be multiple ‘depots’ across a city where cars can get fixed.

For more reading on this new business model, see: Who Owns the Cars? The Billion Dollar Problem with Autonomous Taxis

Peak Demand Challenges

Whoever ends up owning this fleet is going to face one of their biggest challenges during times of peak demand.  In the past, taxis were never able to meet the highest point of demand on a Saturday night because if they had enough cars for Saturday night, then a majority of their cars would be empty the rest of the time.

Today, Uber solves that problem by leveraging the fact that a majority of their workforce are part-time drivers doing only 10 hours a week or less.  In essence, those part-time time drivers are subsidizing the vehicle costs for Uber since they’re able to come onto the platform to meet demand on a Saturday night, but there is no expense on Uber’s part when the car goes offline.

But when you have to own and maintain your own fleet, it makes more fiscal sense to have enough cars to meet average demand.  Uber could solve this by leveraging personal self-driving vehicles but now they have to pay a premium to an individual owner.  And at least at the beginning of self-driving cars when the technology is the most expensive, who would want to let a bunch of strangers ride around in their brand new and expensive self-driving car?

Regulation Will Be The Ultimate Barrier For Self-Driving Cars

When Tesla’s auto-pilot had its first fatality, it was a huge story.  There’s just something that doesn’t sit well with most people when a robot kills a human, but if you’re texting and driving and you kill someone, it’s not as big of a deal.  Personally, there’s no doubt in my mind that a network of self-driving cars will be magnitudes safer, but when self-driving cars do make those mistakes, they will be magnified by public perception.

Uber blazed a warpath in its quest to upend the taxi industry and overcome regulation and it worked.  But when you start involving real public safety and human lives, there’s no way Uber can take the same approach.  I think Uber is going to really struggle with regulators and city officials when you start to add in all the safety questions, job losses and more.

To put things in perspective, a few months ago I moderated a panel in LA on the Future of Transportation and one of the remarks that really stood out to me came from Hyperloop One‘s director of recruiting when he said, ‘The first Hyperloop will not be built in the US.  It will probably happen in the Middle East or some far off place like that.”

Regardless of your personal feelings on regulation, the reality is it’s a problem for technologically advanced companies like Uber in the US, and this time they won’t be able to ignore it.

What About Uber’s Human Drivers?

During Uber’s major growth phase in 2014-15, they were constantly touting the number of jobs they were creating in every place they landed.  But over time, that rhetoric has shifted with the push towards self-driving cars.  Here’s the blog post Uber sent to its Pittsburgh drivers:

We wanted to let you know that a small number of Self-Driving Ubers will start picking up passengers in Pittsburgh today. Drivers like you are an essential part of Uber and we know you’ll have questions about what this technology means for your future.

It’s still early days. Self-Driving Ubers have a safety driver in the front seat because they require human intervention in many conditions, including bad weather. Even when these technology issues are fixed, we believe ridesharing will be a mix—with rides provided by drivers and Self-Driving Ubers. This is because of the limits of self-driving software and the skyrocketing demand for better transportation that drivers like you are uniquely able to solve.

The past has shown that technology creates new work opportunities, while disrupting existing ones. Many people predicted that the ATM would spell doom for bank tellers. In fact, ATMs cut the cost of running a local bank so more branches opened, employing more people. Self-Driving Ubers, for example, will be on the road 24-hours a day, which means they’ll need a lot more human maintenance than cars today.

I have to commend Uber for sending this e-mail to drivers in Pittsburgh (although I think they should have sent something to all drivers) since in the past, they’ve basically swept these issues under the rug. As a driver, it’s a weird feeling knowing your job is going to be replaced by a robot some day. But according to this e-mail, that may not be the case.

But I would take anything Uber says with a grain of salt since when it comes to self-driving cars, they’ve changed their messaging more than once:

  • In 2014, Travis Kalanick said, “Look, this is the way of the world, and the world isn’t always great.”  Wow, could he be any less empathetic?
  • In 2015, Travis Kalanick said there were several “easy ideas” in play to assist drivers with this shift, citing “vocational periods, education, and transition periods”. He insisted “you’ll see Uber working on this way before the transition happens.” Someone in PR must have given him some tips before this interview 🙂
  • In 2016, Uber is now saying, “Even when these technology issues are fixed, we believe ridesharing will be a mix—with rides provided by drivers and Self-Driving Ubers.” Actually makes a lot of sense to me.

You can see some of Uber’s early messaging around self-driving cars was pretty ludicrous.  Vocational training for drivers?  Give me a break.  I also can’t imagine a bunch of Uber drivers want to go maintain and clean puke out of a self-driving car either.  Uber’s latest message, around a hybrid approach to self-driving cars, finally starts to make much more sense.

Learn more about the current state of self-driving cars below:

 

Recruiting and Retaining Drivers

As we’ve seen with the self-driving car tests in Pittsburgh, nobody’s going to wake up one day and see the streets filled with self-driving cars.  There’s going to be a transitional period, and they’ll also launch on a market-by-market basis.  But who’s going to want to work for Uber when you can look out the window of your car and see the robot that’s going to take your job?

Honestly, I think there are probably a lot of people willing to do that since there’s always people who need work.  But Uber tends to think of these drivers as data points on an Excel chart more than anything and it shows.  When Uber cuts rates, they know X % of drivers will quit, but they can replace them for a cost of $Y.  But what they fail to take into account is all of the monetary and even emotional investment that people put into becoming an Uber driver.

Most of the people flocking to my site are new at this and they’re investing a lot of time and energy into becoming drivers.  As self-driving cars hit the road, I’m not convinced Uber will be able to sign up enough drivers during that transitional period.  Why would anyone invest time or money into becoming an Uber driver when the gig might only last for a few months?  Who knows, they might even have to pay drivers more to get them to come drive for Uber.

Readers, what do you think of self-driving cars and do you think a self-driving car could handle the Friday-Saturday night crowd?

-Harry @ RSG with additional reporting from Paula Gibbins and Melissa Berry