3 ms·
I'd assume any kind of "ETAnet" would be trained on everyone's driving patterns, not individuals.
by cameronbrown 7y ago
I'd assume any kind of "ETAnet" would be trained on everyone's driving patterns, not individuals.
- sib 7y agoI don't know, but I always assumed that a simple optimization would be one that compared your typical driving speed to posted speed limits on open roads. That is, do you usually drive 5 or 10 miles (kilometers) per hour under or over the limit when you have the choice?
- cameronbrown 7y agoI suppose they could. I just think there's enough driving patterns that can be detected in aggregate, that a network could detect your typical style much better with everyone's data, rather than training just from you.
- brewdad 7y agoThere must be some training involved on the user level. My route to the freeway can take 1 of 2 parallel roads. The bigger road is theoretically a minute faster but often enough the left turn lane backs up for multiple light cycles. The smaller road is quite consistent in it's timing. A minute slower if the lights work out and a minute faster if you would have to wait. I prefer the consistent route and now Google almost never tries to send me the "faster" way.