4 ms·
> A hardcoded 1 second delay during a potential emergency situation. Horrifying. As a controls engineer in the automotive industry, I can tell you that a 1-se
by unbalancedevh 7y ago
> A hardcoded 1 second delay during a potential emergency situation. Horrifying.
As a controls engineer in the automotive industry, I can tell you that a 1-second delay for safety-critical systems is not atypical.
The expectation is that the normal software avoids unsafe operation. Bounding "safe operation" is difficult, so if an excursion is detected, there's essentially a debounce period (up to 1 second) to let the normal software correct itself before override measures are taken.
This helps prevent occasional random glitches or temporary edge cases from resulting in a system over-reaction, like applying the brakes or removing torque unnecessarily, that would annoy the driver and potentially cause unsafe operation themselves.
Obviously there are still gaps with that approach. But there is supposed to be a driver in charge; and the intent is to prevent run-away unsafe behavior. It essentially boils down to due-diligence during development.
- nolok 7y agoTo counter balance your point : the original Volvo emergency braking system in the car saw the crash and wanted to brake 1,3 seconds before it happened. So Volvo engineers didn't think at all like you do / say. Their system was 0,1 second faster than Uber at detecting it, 1,1 seconds faster if you factor in Uber active suppression, and it would have braked 2,1 seconds sooner than the Uber did. Why didn't it? Because Uber deactivated it's braking ability.