San Francisco robotaxis are causing false 911 calls and other chaos

[Daniel Sims]

Through the looking glass: Sometimes unfortunate or sometimes deadly incidents involving self-driving cars happen. However, recent occurrences with autonomous vehicles in San Francisco have been downright bizarre. Not only does the software controlling the cars struggle to negotiate the real world, but companies and authorities are still learning about how humans behave around self-driving vehicles.

Recent formal complaints to California regulators reveal strange incidents that have occurred since fully-autonomous taxis started operating in San Francisco and Los Angeles. The cases mainly involve robotaxis disrupting first responders or otherwise wasting their time.

Wired notes that one letter sent to the California Public Utilities Commission claims that sleeping passengers have resulted in unnecessary 911 calls. GM-owned Cruise operates self-driving taxis in San Franciso, featuring two-way communication between passengers and company staff. On three occasions, passengers decided to nap during their trip, and Cruise staff called 911 when they became unresponsive.

San Francisco authorities described the events as a waste of first responders' time, requesting that the state of California slow down its adoption of driverless vehicles on public roads. It appears that robotaxi passengers are starting to behave similarly to how they would on buses, trains, or airplanes, using trips as an opportunity to get some sleep.

Other complaints cite incidents where autonomous taxis impeded firefighters. Soon after California let Cruise operate robotaxis in San Francisco last June, one ran over a fire hose during an active firefighting operation. Last week, firefighters had to spend over two minutes stopping an approaching robotaxi from running over their hose. The letter claims they eventually had to shatter its front window to force it to halt, but Cruise says it had already stopped the car by then.

These cases seem like an escalation from the incident last July where a test group of Cruise robotaxis stopped in the middle of the road for an unknown reason, obstructing traffic. Human engineers had to remove the vehicles after they ceased all functioning.

That incident is one of almost 100 that San Francisco agencies said occurred between May and the end of 2022 in which Cruise vehicles held up traffic by stopping at inopportune places and times. The occurrences disrupted public transportation and caused other human drivers to respond with sudden, potentially dangerous maneuvers.

San Francisco and Los Angeles transportation authorities are worried that self-driving technology isn't ready for their cities and are requesting more data from companies like Cruise before they expand their operations. Cruise states that it has to withhold some information for consumer privacy and to protect trade secrets. The company also points out that none of the noted incidents have resulted in serious injuries or fatalities.

The officials' letters aren't uniformly negative, however. They admit that driverless vehicles could increase opportunities for people who can't drive, including the disabled.

Permalink to story.

https://www.techspot.com/news/97418-san-francisco-robotaxis-causing-false-911-calls-other.html

 

[psycros]

You know what else can transport people who can't drive? Uber, Lyft, regular taxis and the bus. There's absolutely no need for dangerous robo-cars.

 

[VitalyT]

 

[passwordistaco]

Napping or nodding out?

 

[Hodor]

You know what else can transport people who can't drive? Uber, Lyft, regular taxis and the bus. There's absolutely no need for dangerous robo-cars.

I agree, however, a self-driving car won't try to rape a drunken teen girl, which in the case of Uber and Lyft has been happening.

 

[Endymio]

You know what else can transport people who can't drive? Uber, Lyft, regular taxis and the bus.

The nice thing about software like this is that, once an error is found and corrected, it never happens again, whereas human drivers keep making the same mistakes over and over. In 30 years, self-driving vehicles will be so safe and reliable that the thought of allowing someone to manually drive a vehicle will sound as insane as allowing a ten-year old boy to pilot a fully-armed F-15E.

 

[waclark]

You know what else can transport people who can't drive? Uber, Lyft, regular taxis and the bus. There's absolutely no need for dangerous robo-cars.

To be fair, I've had a couple Uber/Lyft drivers that were downright dangerous. One almost got us T-boned trying to make a left-hand turn. One was so herky-jerky that I had a sore neck for a week. Some humans should not be allowed to drive, period.

 

[Vanderlinde]

To be fair, I've had a couple Uber/Lyft drivers that were downright dangerous. One almost got us T-boned trying to make a left-hand turn. One was so herky-jerky that I had a sore neck for a week. Some humans should not be allowed to drive, period.

Weird. I think uber is the "better" and more stricter taxi type of thing service compared to lyft or bolt.

Out of the 250 ~ 500+ rides only once it went a bit odd; driver stopped at the gas station to have a break while supposedly bring me to my destination.

 

[Nobina]

The nice thing about software like this is that, once an error is found and corrected, it never happens again, whereas human drivers keep making the same mistakes over and over.

In dynamic situations like when driving it's very difficult to correct errors that haven't happened yet. As software gets more complex it becomes more difficult to find and/or fix errors properly. Humans are better in such environments.

In 30 years, self-driving vehicles will be so safe and reliable that the thought of allowing someone to manually drive a vehicle will sound as insane as allowing a ten-year old boy to pilot a fully-armed F-15E.

Probably not gonna happen. Most people's idea of an AI is what they've seen in sci-fi movies.

 

[Endymio]

Humans are better in such environments.

Humans have killed more than two million people attempting to drive vehicles, in just the US alone. And they're not getting any better at it.

I heard for 50 years that no machine would ever beat even an average chess player. Then it became none would ever beat a grandmaster. Today, such programs make the world's best player look like a 90 lb weakling getting sand kicked in his face on Venice Beach. A self-driving vehicle can see in the dark, in thick smoke or fog, in all 360 degrees at once. It never gets sleepy, distracted, or drunk. And once AI engines are fully training on these "dynamic situations", they'll never make the same mistake twice.

 

[waclark]

Weird. I think uber is the "better" and more stricter taxi type of thing service compared to lyft or bolt.

Out of the 250 ~ 500+ rides only once it went a bit odd; driver stopped at the gas station to have a break while supposedly bring me to my destination.

I had that as well. A gal didn't realize how far the trip was and had to stop for gas. My biggest complaint, recently, has been language skills. I've had a couple drivers in the past month or so whose first language isn't English. Not a big deal, but one driver didn't know where they were going and didn't seem to want to follow their GPS and it was difficult to give them instruction on where to go. The second one had a similar problem because the app was not working properly so we had to tell them how to get where we were going. I don't expect people to be fluent but you should understand things like, turn right, turn left etc. Overall, most drivers have been fine and I will continue to use the services.

 

[Nobina]

Humans have killed more than two million people attempting to drive vehicles, in just the US alone. And they're not getting any better at it.

I heard for 50 years that no machine would ever beat even an average chess player. Then it became none would ever beat a grandmaster. Today, such programs make the world's best player look like a 90 lb weakling getting sand kicked in his face on Venice Beach. A self-driving vehicle can see in the dark, in thick smoke or fog, in all 360 degrees at once. It never gets sleepy, distracted, or drunk. And once AI engines are fully training on these "dynamic situations", they'll never make the same mistake twice.

Chess has rules, limits and is very different from driving a car. Regardless, you think in 30 years these things will be fully fleshed out? If that happens, they're certainly not going to be affordable, car prices are getting out of control as we speak. Maybe the 1% will own them, the rest will drive their own cars. As long as that is the case and as long as people are willing to drive their own cars, humans will be able to drive and should fight to death anyone trying to take that right away. Probably what bugs me the most about this is how can anyone put their life in the hands of "AI" they know nothing about, made by a private company that has shown time and time again they're all about releasing unfinished products and making a profit. I can only describe those people as gullible or suckers.

 

[waclark]

Chess has rules, limits and is very different from driving a car. Regardless, you think in 30 years these things will be fully fleshed out? If that happens, they're certainly not going to be affordable, car prices are getting out of control as we speak. Maybe the 1% will own them, the rest will drive their own cars. As long as that is the case and as long as people are willing to drive their own cars, humans will be able to drive and should fight to death anyone trying to take that right away. Probably what bugs me the most about this is how can anyone put their life in the hands of "AI" they know nothing about, made by a private company that has shown time and time again they're all about releasing unfinished products and making a profit. I can only describe those people as gullible or suckers.

In 30 years no one will be driving a personal vehicle, except possibly in rural areas. Consider this. Assume that you can build a car that is pretty good at autonomous driving. Once you have that, it then makes sense to reduce the number of cars on the road so that there are fewer vehicles and therefore your self-driving car will have fewer obstacles to monitor and track. At that point, I suspect the government will start mandating use of high-occupancy vehicles such as busses or railed vehicles at least in urban areas where there is a high concentration of people. They will force this by making parking difficult and expensive for personal vehicles. There will be more toll roads and taxes for low occupancy cars and trucks. Look at what San Diego is proposing to make it more expensive to drive a car.

Logically, it does make some sense but the problem is that currently mass transit in the US is a sh*t-show. It's not ubiquitous, it's not clean, it's not efficient, it's not safe and it's not fast. Still, I think the government will push this under the guise of being environmentally better, safer for commuters and will result in cost savings by removing a high percentage of vehicles from the road which means we don't have to build more roads and maintaining existing ones will be cheaper.

 

[Sir Sparkles]

Personally I welcome our robot overlords. Resistance is futile.

 

[Freddie159]

Robotaxis could fix this simply by blowing the horn and punching the gas pedal and slamming on the brakes to 'jar' the passenger awake, not enough to injure them but enough to 'wake them'.

 

[yannus]

I have to say the article made me laugh even though it's not funny. It seems to me that people follow any trend they're told without considering that things that used to be good could still be good today. Technology is good but it always should be compared to what we already have. I'm sorry but the world we used to have in the 80's-90's was WAY better than what we have right now. A moderate and wise approach of technology and so called "science" should be considered.

 

[Avro Arrow]

"On three occasions, passengers decided to nap during their trip, and Cruise staff called 911 when they became unresponsive."

Can't say that I blame them.

"It appears that robotaxi passengers are starting to behave similarly to how they would on buses, trains, or airplanes, using trips as an opportunity to get some sleep."

Can't say that I blame them.

These "robo-taxis" are a terrible idea because they're taking away jobs that people need to live. If I were the California governor, I would never have allowed them to begin with for just this reason. If I were governor and they were already in place, the owners of said companies would be taxed so high that it would be cheaper for them to not only employ human drivers, but to pay them well.

I would allow maybe one of these things per ten full-time human drivers on any given company's roster without imposing the tax and those would only be permitted for use at night when human drivers would be better off sleeping or spending time with their families. The other upside would be that any screwups these automatons had would be mitigated by the fact that the city would be mostly deserted compared to the during the day. Less risk to the public and first responders wouldn't have to fight daytime traffic to reach them, whether their calls were genuine or false.

 

[Endymio]

"These "robo-taxis" are a terrible idea because they're taking away jobs that people need to live.

Right! And these electric lights are putting all the candlemakers out of business too. And don't even get me started on electronic calculators -- now one bookkeeper can do the work of 20. What are the other 19 supposed to do for a job?

 

[mailpup]

What are the other 19 supposed to do for a job?

Perhaps manufacture electronic calculators.

 

[BadThad]

Robotaxis could fix this simply by blowing the horn and punching the gas pedal and slamming on the brakes to 'jar' the passenger awake, not enough to injure them but enough to 'wake them'.

Rigging the rear seat with an electric charge would be much more fun!

 

[Q Wales]

Surely common sense is for the taxi to return to base and the snoozer gets billed for the time wasted and inconvenience