For anyone wondering if the speed of light is a concern, a good mnemonic is that a foot is roughly equal to a nanosecond. So a millisecond is long enough (by a factor of ~50000) that camera placement doesn’t matter.
I used to work with a company who insisted on timestamping their sensor data with nanosecond accuracy. Mind you they were not working on anything high speed, or explosive in nature. Just robots moving around at best with highway speeds.
I asked them if they care so much about the accuracy will they also adjust their data for time-of-flight delays. Since by the time the light bouncing from an object hits your lidar/camera meters away what you are sensing is several nanoseconds the past.
To top it off they stored the nanoseconds elapsed since the unix epoch as a float double. So floating point weirdness immediately destroyed all the precision they worked so hard on.
Well, camera placement matters in the obvious way: it has to be on the finish "line". Where "line" is really a plane extending vertically from the line drawn on the surface.
I vaguely remember an article about a bike race photo finish that showed that this is not a given and reverse engineered where the camera was pointing (based on background) and who the likely winner was.
But in the end, if I am not mistaken the UCI rules state that the photo finish line is the final arbiter, not the paint on the ground. I don't think this is a great injustice - it's a sport and the rules are arbitrary to a great extent (e.g. male riders 180 cm to 189.9 cm tall must use a smaller bike than those 190 cm and taller in time trials - which sucks for those 189.9 cm tall).
As Grace Hopper so elegantly pointed out. She gave me one of her many nanoseconds, which were lengths of scrap telephone wire cut to a little under a foot long.
I do appreciate that lightspeed is "about a billion feet per second" and it's also a number that's known exactly, since the foot is defined in terms of the speed of light in a vacuum.
krisoft|1 year ago
I used to work with a company who insisted on timestamping their sensor data with nanosecond accuracy. Mind you they were not working on anything high speed, or explosive in nature. Just robots moving around at best with highway speeds.
I asked them if they care so much about the accuracy will they also adjust their data for time-of-flight delays. Since by the time the light bouncing from an object hits your lidar/camera meters away what you are sensing is several nanoseconds the past.
To top it off they stored the nanoseconds elapsed since the unix epoch as a float double. So floating point weirdness immediately destroyed all the precision they worked so hard on.
2OEH8eoCRo0|1 year ago
utensil4778|1 year ago
planede|1 year ago
I vaguely remember an article about a bike race photo finish that showed that this is not a given and reverse engineered where the camera was pointing (based on background) and who the likely winner was.
edit: found it https://www.tglyn.ch/blog/amstel_gold/
Luc|1 year ago
But in the end, if I am not mistaken the UCI rules state that the photo finish line is the final arbiter, not the paint on the ground. I don't think this is a great injustice - it's a sport and the rules are arbitrary to a great extent (e.g. male riders 180 cm to 189.9 cm tall must use a smaller bike than those 190 cm and taller in time trials - which sucks for those 189.9 cm tall).
satiated_grue|1 year ago
https://americanhistory.si.edu/collections/nmah_692464
jimbobthrowawy|1 year ago