Search

Making Safe Self-Driving Cars Is Really, Really Difficult, So Lyft Just Sold Its Self-Driving Tech - Forbes

simpanta.blogspot.com

Full disclosure: I’m a non-active Lyft driver. I was plying the streets of Portland until mid-March of 2020, when the pandemic hit like a viral asteroid and everything just stopped, including my Lyft adventure. Why was I driving Lyft? It covered my car payment, it was fun (you meet some really interesting people at 3:30 a.m.) and because I wanted to see the rideshare business from the inside, as well as track Lyft’s forward motion on autonomous cars. Those self-driving Lyft cars were supposed to be on the road putting me out of business right now.

They aren’t. And they won’t be for a long, long time.

Quick overview of the deal: Lyft sold its autonomous vehicle division to Woven Planet Holdings, a subsidiary of Japanese automaker Toyota, for a little more than half a billion dollars. The deal is... complicated, but keeps the two parties in touch if anyone can crack the autonomous code in the future. Lyft stock popped a bit on the news to $65 per share before it slid back to $62, for a $20.5 billion dollar market cap. Not bad for a company that has yet to make dollar one in profit. Forecasts from two years ago - about when I fired up my fancy LED Lyft sign - had pegged 2021 as the year the rideshare icon would turn cash flow positive. Then: Covid. Buh bye profit forecast, passengers, Lyft drivers - and any hope of self-driving Lyft cars on the road anytime soon. Maybe someday.

The promise of self-driving cars has been with us for a while now, spurred on by Elon’s Autopilot Tesla tech, which has seen its fair share of bad news over the years despite being as close to actual available-to-the-public self-driving goodness as we’ve ever seen to date. Uber was also on the self-driving path, but bailed on their efforts and sold the program in 2020 while they were also taking a beating from Covid.

But the real reason we don’t yet get in our cars and blithely say “take me to the dentist” while poking at Instagram is because creating the tech smarts, vehicle tech bits and needed infrastructure to make it happen without mass daily mechanical murder of riders, pedestrians and insurance companies is because it’s probably one of the most difficult tech undertakings ever.

Making the cars is easy. Making the radar, lidar, cameras, sensors and computer gear is (relatively) easy. Perfecting the computer code that makes sense of what all the data means to our future robot overlords is extraordinarily difficult when the ultimate goal is to never harm any humans involved. As is often the case, the last part of the journey to the goal is often the most difficult, and computers, for all their speed, power and brilliant design are, to put it bluntly, sorta stupid when it comes to seeing things and understanding what they are and what they are doing. A five-year-old can out-think them. If that wasn’t the case, self-driving cars would be ubiquitous.

Early Tesla Autopilot fails include mistaking a semi truck trailer for the sky, resulting in a fatal 2016 crash. Chances a sober, attentive, experienced human driver would make that mistake? Low. Musk claims you’re safer using Autopilot than taking the wheel yourself, and in most cases, he may be right, but it’s the “most cases” part of the deal that’s the literal killer. It seems that when it goes wrong, it really all goes wrong. Uber scored their first fatality of the self-driving era in 2018 when a self-driving car in self-driving mode with a human along just in case still hit and killed a pedestrian. That incident paused Uber’s autonomous efforts.

The fact of the matter is that the human mind, for all its faults, is still vastly better at navigating clogged morning rush hour city streets to get to the nearest Starbucks than the best A.I. can hope to. For now.

I have no doubt our “driving future” will feature blissful, driver-free, worry-free trips in silent electric cars with 5,000 miles of range being guided by Matrix agent wannabes that digitally talk to every car, truck, bicycle, stoplight, streetlight, crosswalk and pedestrian tracking tag in our brightly LED-lit hyper-connected future. A car crash? Like what grandpa was in back in 2025? Never seen one.

It’s going to happen, don’t worry. It’s just a matter of time, money and the steady, inexorable march of technology. Lyft’s and Uber’s autonomous tech may change hands, but eventually, someone is going to get it figured out, and the sooner the better. But not this year. I better reactivate that old Lyft profile. People need rides from people.

Let's block ads! (Why?)



"difficult" - Google News
April 28, 2021 at 06:00PM
https://ift.tt/3xvVQoA

Making Safe Self-Driving Cars Is Really, Really Difficult, So Lyft Just Sold Its Self-Driving Tech - Forbes
"difficult" - Google News
https://ift.tt/2VWzYBO
https://ift.tt/3d5eskc

Bagikan Berita Ini

0 Response to "Making Safe Self-Driving Cars Is Really, Really Difficult, So Lyft Just Sold Its Self-Driving Tech - Forbes"

Post a Comment

Powered by Blogger.