I was curious if the sensor would pick up other things like trees or other cyclist, but it seems like they accounted for that:<p>> We then log a sensor events [sic] if the majority of cells in the sensor frame agree to the same value within a threshold parameter [...]. This ensures that sensor events are only logged when large objects like cars block the sensor’s field-of-view , i.e., one or more small objects like branches or distance pedestrians in the sensor’s field-of-view will not trigger this condition. While there is no guarantee that this approach strictly identifies cars, we empirically saw during testing that passing cyclists and pedestrians rarely satisfied this condition at the typical passing distance due to the wide field-of-view of the VL53L8.<p>Also interesting that it's quite cheap to build:<p>> The whole system can cost less than $25 [...]<p>From the paper
<a href="https://dl.acm.org/doi/10.1145/3706598.3713325" rel="nofollow">https://dl.acm.org/doi/10.1145/3706598.3713325</a>