4 ms·
> Show me one that can handle a snow squall or heavy rain or missing signs and lane markers and construction zones, and I might think it’s possible this decade.
by alpaca128 6y ago
> Show me one that can handle a snow squall or heavy rain or missing signs and lane markers and construction zones, and I might think it’s possible this decade.
And the next step would be correct recognition and behaviour
based on the current location. Different countries use different
colors and styles for road markings. Some use yellow lines in
the middle, others orange in the case of redirection around
construction sites. Then there are the cases where the lines are
simply incorrect or out of date(which famously lead to the fatal
Tesla autopilot crash).
It is impressive how far the technology has come, but it's an
area with countless edge cases and unpredictable situations
coupled with high stakes if something goes wrong. I can see
self-driving trucks being adopted slowly on specific routes, but
a full switch not labeled as "pilot project" will still take a
while I think.
- wenc 6y agoI imagine for well-plied, fixed routes (interstates between two warehouses) there could be training runs (with a driver) to check edge cases along routes before going to prod. There could also be periodic retraining with drivers, as well as offline model retraining with the data that the trucks will be collecting continuously. This simplifies the problem significantly (even though it doesn't solve it completely -- there's always the 1% off case). I don't think one would just plop a general system on any route without human-supervised customization -- at least we're not there yet.