San Francisco Is Teeming With Self-Driving Cars And It’s A Mess
Here in the heart of San Francisco this really feels like any other rideshare. But there's one key difference. There's no driver. Cruise says that the future is here and robotaxis are ready for prime time.
But local officials and first responders, they say not so fast. They want a slower rollout and they want to make sure it's done right. We just went around a car with its emergency lights on. It did it pretty smoothly, but now we're. Oh, now we're behind a car that's trying to park.
He got out. Oh my gosh. And he kicked it. I'm going to put up my window. So this guy just kicked the car and now we're calling support. This is stressful. Oh my gosh. Oh, he just spat on the car.
Hi. I can't hear you that well. Everyone's okay. Okay. Do you feel safe at this time?
Yes. Okay. Thank you for letting me know. I apologize you experienced this. Those were the most stressful few moments I've had in a car in recent memory.
I took this ride in a Cruise in September, but the company has since halted all autonomous operations nationwide. So what happened? It turns out they weren't ready for prime time. Self-driving cars without safety drivers behind the wheel have flooded San Francisco streets. Two of the leading companies, Waymo and Cruise, have been operating in the city for years, but in limited capacities.
In August, both companies were granted permission to expand operations, allowing people to hail a fully autonomous vehicle like an Uber. Waymo is still operating autonomous vehicles in San Francisco, Phoenix and Los Angeles. And Cruise had previously been operating in San Francisco, Phoenix, Austin, Dallas, Houston and Miami. Seeing cars navigate busy city streets on their own feels like we've stepped into the future, but the launch of autonomous fleets hasn't been without problems. Robotaxis have driven into fire fighting scenes, caused construction delays, impeded ambulances and even meandered into active crime scenes.
It's clear that there are still some glitches that need to be worked out, and this is only with a few hundred vehicles. The idea that thousands of vehicles could be hitting our streets in short order is what gives us concern. They're not ready for prime time. They interfere with our operations. They really have mucked some things up for us. It's impacting our public safety and that's what we need to fix.
In late October, the Department of Motor Vehicles suspended Cruise's permit to operate driverless vehicles in California. Following the announcement, Cruise halted driverless operations nationwide. I sat down with the Cruise CEO before the DMV suspension, and he was optimistic it could get past the hurdles. We're testing in 15 cities and when this vehicle behind us goes into production, we'll have a lot of capacity to expand. It will be very commonplace for people who are in major cities to get around town in a robotaxi.
California has been fairly open with autonomous vehicles, allowing dozens of companies to test driverless cars on public roads. In 2009, alphabet owned Waymo was the first to start testing its autonomous technology on San Francisco streets. GM-owned cruise has been operating in the city since 2015, but back then, vehicles were required to have safety drivers behind the wheel. More recently, the drivers were removed. Waymo was actually one of the first companies to start doing driverless tests on public roads, and launched some limited driverless ride hail services in 2019 2020. Cruise has been a little more recent and
middle of 2022, they got a permit to actually start operating paid rides. Fully driverless operations in San Francisco started in 2020 with limited rider programs. Rides were then expanded to include small groups of the public that have been invited or had signed up on a wait list. It costs similar to what you'd expect from Uber or Lyft. I took a ride in a Rruise a few weeks before the DMV suspended its permit.
So we ordered our car. Kyle Vaught, the CEO of Cruise, says that you always remember the name of your first car. So it's Bedrock, easy enough on the app, and then you basically pick a pickup spot. And I think it's circling the block and trying to get to me. Okay, here's our ride — Bedrock. Unlock doors. You can even honk. Oh, I guess the honking is so you can find it.
We've been in the car for about 30 seconds and we've already been honked at twice. A few things to remember. Please keep your seatbelt buckled. Always keep your hands and arms inside. Two screens for each passenger. There's a map of San Francisco. An end ride, a
help button. There's a trivia button and radio. That's kind of wild. Changing lanes with no driver.
Not a very successful lane change. So it kind of feels like a normal taxi or Uber ride. I have to note, I think it's the fourth time we've been honked at in two minutes of being in this car. It almost makes me nervous when we're really cruising. One thing you quickly notice is the number of cameras. They're always watching you. You've got one on the roof right here, and you've got one in the front to make sure that you're not interfering with the wheel or any part of the front of the car. They're pretty clear on what you can or
cannot do, even when you open your window, something comes up on the screen and says, "keep both hands and arms in car at all times." Pretty eventful ride, we went all throughout the city. Nearly got into two accidents. Got honked a lot. I won't say it was smooth, but it was interesting.
And the best way to describe it is like being driven around the city by student driver. We tried out Waymo as well, while Cruise had more robotaxis on city streets, Waymo has been testing longer. Both companies say they've logged more than 1 million driverless miles. It did start driving without me having my seatbelt on. Cruise would not do that.
So when you get in the car, you can push on a button that says start ride. And I would say so far we're only a few minutes in, but does feel a little smoother than the Cruise robotaxi we took a week ago. And similar to Cruise, it has sort of a navigation tells you where you're going. That is something
that a cruise car would not do. This car seems to drive much more like a human driver, whereas the cruise really did feel like a robotaxi. Oh, this is interesting. Cruise didn't have this something called camera privacy. We may use interior cameras to check on riders. Interior mics are on only when connected to
rider support. So I guess they're telling you they're not listening to you, but cannot turn off the cameras. That's interesting. It took the opportunity to squeeze
through a pretty narrow space and did so fine. Actually doing better than the drivers around us. I got to say one word to describe this ride. Unremarkable. We didn't get honked at once.
There was one incident where I thought maybe we cut someone off. It was uncertain, but it's been really smooth. You ask me whether this ride is ready for prime time, I mean. And would I take it again?
I think the answer is yes. While the Waymo ride was a smoother experience compared to Cruise, and it still has its permit to operate autonomously in the city, it too has faced challenges. Waymo declined our request for an interview. They've had actually some similar problems to some of the issues that Cruise has had in San Francisco.
With vehicles sometimes causing traffic backups, they would get stuck. There was a situation that was happening where Waymo vehicles kept turning down a dead end street. They've also had issues with emergency responders causing some congestion.
In August, the California Public Utilities Commission approved permits for Cruise and Waymo to charge fares and expand operations in San Francisco. It allowed them to significantly expand their fleets. And in fact, for Waymo, it was the first time that they were allowed to charge passengers for rides in California. But as the number of driverless cars on the road increased, so too did the problems. The uptick happened when they took the driver entirely out of the car. The uptick happened when they took drivers out of the car, and then it exploded after the California PUC okayed the expansion.
There have been 75 plus incidents. And so to me, it's like playing Russian roulette. We are the last line of defense for most people. Seconds matter if it takes us a minute to go around the block, because there's a vehicle in the way that puts the people in that building more at risk. It puts that building more at risk, and it puts my firefighters more at risk. And the problems aren't just happening in San Francisco.
In Austin, where Cruise launched at the end of the year, similar incidents occurred. We have been responsive. We have been making changes very rapidly to the software, probably much more than most companies are capable with because this is a technology and we can deploy changes to the entire fleet very, very quickly. You know, that said, I think there's more work to do. We're not done here.
In August, Cruise vehicles blocked an ambulance from leaving a scene where critically injured pedestrian had been hit by a car in downtown San Francisco. As part of its efforts to work with the city, Cruise and San Francisco fire reviewed footage from the vehicles to determine what might have gone wrong. My team was able to look at that video. We saw the same video, but we saw two very different things. They didn't see any problem with it, but it delayed us by a minute and 45 seconds, maybe two minutes. As the number of incidents throughout the city increased, Cruise reduced its fleet by 50% at the request of the DMV, just over a week after the Public Utilities Commission ruling. But the final straw
came after a pedestrian was struck in a hit and run by a human driver and then thrown into the path of a Cruise vehicle. Responding to the collision, the AV pulled over, dragging the pedestrian with it. In its investigation, the DMV claims crews failed to disclose that the AV executed a pullover maneuver that may have caused further injury, and concluded that its vehicles lacked the ability to respond in a safe and appropriate manner during incidents involving a pedestrian, and suspended its permit to operate in California. Cruise said in a statement, 'Ultimately, we develop and deploy autonomous vehicles in an effort to save lives, and that its teams are currently doing an analysis to identify potential enhancements to the AV's response to this kind of extremely rare event.' On the same day, General Motors CEO Mary Barra spoke about the technology safety on the company's third quarter earnings call. As Cruise continues to push the boundaries of what AV technology can deliver to society.
Safety is always at the forefront and this is something they are continuously improving. In fact, it's our zero crash vision that keeps us pressing forward, and we know from the data that cruise AVs are involved in far fewer collisions than human drivers. Did they ever approach you in this process before they had approval to run robotaxis or autonomous vehicles? Yes, and they did provide us with some training. And the training was, you know, these are our vehicles. This is how you would cut someone out with our jaws of life. But there was never any real
conversation about how to interact with our vehicles, how to stay out of the way of our vehicles when we're going to an emergency or when we're at an emergency. Is there collaboration now? My folks met with both Cruise and Waymo, and I'm hopeful that we will continue these meetings and that it will lead to fewer incidents. What I don't appreciate is when the CEO says that we are sensationalizing everything. I'd love to take him on a ride along on engine three someday, and he can see how challenging it is for us to get around the city as it is. Never mind when there's an autonomous vehicle that will not listen to the police or us. We just need to sit down and work together so they understand how we need to operate, what will impact us and what won't. And we need to be able to set up
geofencing as well so they can stay away from certain areas when we have incidents going on. When asked about collaborating with the city, the autonomous vehicle companies say they're cooperating with officials. We have a hotline for first responders.
That's one of the things we built in response to feedback. So both police and fire department can call a number, and we can very quickly do things like relocate a vehicle that's in the way. Data about every collision is reported both at the state and federal level.
We've shown video footage with first responders and things like that when it's appropriate, but that's got to be very carefully controlled in order to maintain the privacy of our customers. In California, the Public Utilities Commission is responsible for governing autonomous vehicles, but it's a state level agency. On the actual streets of San Francisco, where the majority of testing takes place, city officials there have no say. Members of the public think that San Francisco ought to be able to have some ability to influence how this industry expands on our streets. Part of the problem with this system, with both the DMV and the CPUC, and this actually also applies throughout the entire country, is that currently we don't actually have any safety standards that govern these vehicles.
From our perspective, it's not entirely safe yet. And this is why, when our California regulator made a decision to allow for unfettered, unlimited expansion of autonomous vehicles, our city has had some issues. You're pushing to limit the expansion. We have concerns, and our hope is that we can allay these concerns with proper data and ensuring that if there is an expansion plan, it's tied to safety and performance metrics.
Following the DMV suspension of Cruise's operations, the California PUC also rescinded its permit for Cruise, but not for Waymo. In fact, Waymo partnered with Uber recently in Phoenix to offer self-driving rides. A robot driver can't get drunk, can't get tired.
Is that an argument for having these on the road? For making the road safer? I think it's a really important consideration. Yes. And again, we don't have data. But yeah, I would love there to be no more drunk driving incidents. My folks are the ones that have to
respond to that and deal with the trauma and everything else from that. But Cruise commissioned its own study showing that its AVs are safer than human drivers. We analyze the first million miles of driving that we did in San Francisco and collected a lot of data on human drivers in San Francisco so we could make a fair comparison. And what we found isn't surprising given that AVs don't get distracted, drowsy, or drunk. We see about half as many collisions overall.
But more importantly, there's almost a 75% reduction in the kinds of collisions that cause injury. They're only sharing data around actual collisions, but certainly there are many other incidents that involve interfering with first responders impeding traffic flow. We have no standard on which to base whether these vehicles are actually as safe as humans, safer than humans or not as safe as humans, except to trust that these companies are telling us the truth about their safety statistics. AVs are ready for prime time.
Absolutely. The fire department. The police department, they say the exact opposite, that AVs are not ready for prime time.
Where are they coming from? This is technology that is sharing the road with human drivers. It's important that we have oversight, that we have scrutiny into how these these vehicles are performing. Part of the process of society transitioning to this safer form of transportation is us doing that work, having those conversations and improving the product to the point that it integrates really well with communities. And I think I
think we're doing a great job there, and it's only going to get better from here. The ultimate goal of autonomous vehicle companies is to expand to more cities. The high cost of research and development means that they will need to scale if they ever hope to reach profitability. Kyle Vogt, the CEO of Cruise, has set a target of having $1 billion in revenue in 2025, and that's an ambitious goal. It's very unlikely that they will be profitable for many, many years to come, probably not before the end of the decade at the earliest. It's essential
that they start generating some cash to help fund their ongoing development and expansion. And the same is true for Waymo. GM reported that cruise lost $700 million in the third quarter, and expects a similar run rate moving forward for. For a business like this, it's all about scale, and the good news is that we're operating in a massive market. San Francisco alone is $1 billion ride hail market, and we're already testing in 15 cities. The United States ride hailing market is estimated to reach more than $15 billion in 2023.
When they announced in September of last year that they were going to expand into Phoenix and Austin, Texas, that was the first time that they had gone into Austin. Within less than three months, they were operating a driverless ride hailing service in Austin. Cruise says removing the driver improves the economics.
When you have a system that doesn't have the driver, the cost structure drops and what you're adding back in are the additional costs of the technology and the cost to operate the service. As soon as 2025, we're going to start producing vehicles that hit the sweet spot of a dollar per mile from a cost standpoint. The only thing that stands between this business and profitability at that point is scale. In 2020, cruise unveiled The Origin, its purpose-built robotaxi that features inward facing seats and no steering wheel or pedal.
This take up space, weight, cost, all this stuff we don't need and so we've reclaimed that space for the passengers and also for your luggage and other things that you might want. The company is working with GM to manufacture it and says it's getting close to regulatory approval. We're really fortunate to have GM as a partner, and I think that's going to be a huge asset for us.
We are the only company that actually has this kind of connection with an OEM and can produce autonomous vehicles at scale. Waymo is also developing its own robotaxi with Chinese automaker Geely. It's currently expanding operations in Los Angeles with plans for Austin, Texas next.
Waymo has not been moving quite as aggressively now in Phoenix, they've gone from just the suburbs to basically almost the entire Phoenix metro area, about a 200 square mile area. They're also testing in the Los Angeles area, Santa Monica. And gaining more experience in more driving environments is important. People don't drive the same way in every city. It's unclear when or how cruise will be able to operate without drivers again, but as the companies continue to roll out robotaxis in new cities, regulators have the tough job of developing safety standards for these vehicles. Experts agree this is key to safe and orderly operations, preventing a repeat of what's been happening in San Francisco.
There is no standard governing how safe they have to be, and this is where federal regulators, who typically govern federal motor vehicle safety standards at the National Highway Traffic Safety Administration, has really fallen down on the job over the last 7 or 8 years. They have basically done almost nothing in terms of developing some performance standards for these vehicles. These systems are by no means mature or finished. It's not a solved problem.