3 ms·

The software and training data set an extremely high barrier to competition. Once you've solved self-driving taxis, you've also solved a large chunk of self-dr

The software and training data set an extremely high barrier to competition. Once you've solved self-driving taxis, you've also solved a large chunk of self-driving commercial transportation. AI trucks could kill off rail, in addition to manned semis. From there, it's a small leap to becoming the backbone of an ultra-efficient on-demand national distribution network. All without the massive operating overhead of human labor... No sleep, no breaks, no weekends, no healthcare, no HR, no wages, just the price of electricity and the low/infrequent maintenance costs of electric vehicles. Eventually. I think the taxis are 5 years away from 24/7 availability in NYC, maybe 3 for San Francisco, and the grand vision is probably 10 years out. But... Yeah... This is going to be huge.

rhino369 9y ago
>The software and training data set an extremely high barrier to competition. How many technological barriers to entry have survived very long? The industry may be huge, but it's not clear that only one company will crack self-driving. I think it's far more likely every major autocompany will develop good enough versions of self driving. Plus, these companies won't be able to hide their software behind algorithms. Someone is going to reverse engineer anything put into the open market.
CPLX 9y ago
> I think the taxis are 5 years away from 24/7 availability in NYC, maybe 3 for San Francisco lol That's another amusing point of view I see here a lot. The use case of taxis in a major city is by far the most challenging use case for autonomous driving. Mark your calendars for 5-10 years after you first see an actual autonomous vehicle having any type of real world application at all. Which you haven't. The whole idea that taxis are the vanguard of autonomous driving is obvious patent nonsense, why it persists remains a mystery to me.
- losteric 9y ago
  > Mark your calendars for 5-10 years after you first see an actual autonomous vehicle having any type of real world application at all. Which you haven't. What's your criteria here? Because I have the same opinion as you... Except I've literally seen experimental autonomous cars driving on the street... Hence +3 years for SV, and +5 for NYC. Taxis are not the vanguard, they're just the most visible. Autonomous trucking is underway and will likely see mass adoption before taxis. The revolution is close - we're seeing disruption everywhere. I could empathize if this was an isolated Google vaporware experiment... but every class of vehicle has promising automation, researched by dozens of companies around the world, and implementing multiple approaches. Most will fail, of course, but we are on the cusp of success.
skybrian 9y ago
This isn't a binary problem: "solved" versus "not solved". There are many parts of a solution at various states of development. Research can be replicated and reverse-engineered. It's also unclear that why there would be one vertical-integrated company in an industry built on outsourcing. There's a lot of room for collaboration at building different parts of a car. And why wouldn't a company that builds a self-driving taxi sell it to multiple operators?

rhino369 9y ago

>The software and training data set an extremely high barrier to competition. How many technological barriers to entry have survived very long? The industry may be huge, but it's not clear that only one company will crack self-driving. I think it's far more likely every major autocompany will develop good enough versions of self driving. Plus, these companies won't be able to hide their software behind algorithms. Someone is going to reverse engineer anything put into the open market.

CPLX 9y ago

> I think the taxis are 5 years away from 24/7 availability in NYC, maybe 3 for San Francisco lol That's another amusing point of view I see here a lot. The use case of taxis in a major city is by far the most challenging use case for autonomous driving. Mark your calendars for 5-10 years after you first see an actual autonomous vehicle having any type of real world application at all. Which you haven't. The whole idea that taxis are the vanguard of autonomous driving is obvious patent nonsense, why it persists remains a mystery to me.

losteric 9y ago
> Mark your calendars for 5-10 years after you first see an actual autonomous vehicle having any type of real world application at all. Which you haven't. What's your criteria here? Because I have the same opinion as you... Except I've literally seen experimental autonomous cars driving on the street... Hence +3 years for SV, and +5 for NYC. Taxis are not the vanguard, they're just the most visible. Autonomous trucking is underway and will likely see mass adoption before taxis. The revolution is close - we're seeing disruption everywhere. I could empathize if this was an isolated Google vaporware experiment... but every class of vehicle has promising automation, researched by dozens of companies around the world, and implementing multiple approaches. Most will fail, of course, but we are on the cusp of success.

skybrian 9y ago

This isn't a binary problem: "solved" versus "not solved". There are many parts of a solution at various states of development. Research can be replicated and reverse-engineered. It's also unclear that why there would be one vertical-integrated company in an industry built on outsourcing. There's a lot of room for collaboration at building different parts of a car. And why wouldn't a company that builds a self-driving taxi sell it to multiple operators?