4 ms·
> This is a truly big ask. From the article: The technology for an all-robotic car fleet is probably already available... if we take all the non-robotic cars o
by matheist 10y ago
> This is a truly big ask.
From the article: The technology for an all-robotic car fleet is probably already available... if we take all the non-robotic cars off the road.
Many of the challenges in autonomous vehicles (maybe even all the remaining ones) are related to modelling/predicting an uncertain world. Will a pedestrian jump out at me? Will a truck do a left-turn across my path? What in the world is that bicyclist doing?
Take all that out, and things get vastly easier.
- niftich 10y agoWell yes, it's the same sort of argument that teachers jokingly make, saying school would be so nice without all the students. It's been suspected for some time that a mixed environment of autonomous vehicles and human-operated everything else (including pedestrians) will be the worst of both worlds, as like you said, those algorithms will be face with un-anticipated inputs. The problem is, while cars we can automate way (given enough time), pedestrians we can't, unless we're willing to expend even more money in re-engineering our infrastructure to segregate road traffic entirely, keeping the contact surface to as minimal as possible -- only every home, every workplace, every commercial establishment, every rest stop, and every place vehicular maintenance is done. While this goal tends to appear in 'next city' brainstorming, it's easy to see why it's a challenge.
- matheist 10y agoOh, I thought your "big ask" comment was about the technology. Sure, you'd have to segregate away pedestrians, but there's already precedent for that with many roads (no pedestrians on the highways!), and aren't we dreaming about making new cities anyway, rather than retrofitting old ones? "No pedestrians allowed on the lanes which are painted red, not that you'd want to cross that small spiked fence anyway. Just use the overhead walkway like everyone else."