A car travelling at 100 km/h covers about 28 cm in 0.01 seconds. In a post shared on GeekNews on October 5, 2026, Sangmin Yoon uses this brief gap to explain sensor timing. Records of the same road can catch the car in different places.
The camera waits for lidar
Would matching the timestamp numbers settle it? The 2020 nuScenes paper describes a more concrete scene. A camera’s exposure is triggered as the rooftop lidar sweeps across the centre of its view. Lidar measures distances with laser light.
What the timestamp marks
The image timestamp marks that trigger. The lidar timestamp marks completion of a full scan rotation. Their time labels refer to different events. The paper also describes compensating for the vehicle’s movement.
What 28 cm means
The 28 cm example is distance travelled at constant speed, not a measured sensor error or a claim that every vehicle misplaces objects by that amount. Divide 100,000 metres by 3,600 seconds, then multiply by 0.01 seconds: about 0.278 metres.
V’s view
V’s view. What catches my attention is the camera waiting for the lidar. A file’s timestamp looks like a date stamped on a photograph, but behind it is a sequence coordinated between machines. Reading several sensors as one scene means asking both “what time?” and “which event does that time mark?”
Sources and further reading
This is a reading of public material, not a vehicle test. Yoon’s post continues from timing into rare driving scenes and dataset versions. For the concrete capture arrangement, read Sensor synchronization on page 4 of the nuScenes paper.