Видео

Thanks for your comments! yes this is hard to comprehend. In the more recent videos, we have slightly better annotation, like this one: ruclips.net/video/YcCC0g1WgEI/видео.htmlsi=AqnViqpIJnTIbL2j

Radar can detect the object speed in a single frame. A moving truck is quite different than a billboard. Additional, we can also classify them by the detected object height.

Thank you!

note that this is accumulated point cloud. If you browse the channel, we have a lot of single frame, real time point clouds

you mean extrinsic calibration between different sensors? We do have compensation in the software when overlaying the point clouds between Lidar and Radar, it wasn't perfect but should be within a couple degrees.

@@AltosRadar Something similar, yes. What I mean is that for example for a corner sensor, let"s say rear-right, you might want to mount the sensor at 45 deg in azimuth, but since there are certain mounting tolerances maybe you mount the sensor at 45.5 deg. Do you have a way to detect that 0.5 deg error in an online calibration (while driving) based on radar reflections?

Radar could possibly do online calibration for intrinsic parameters; I cannot think of ways sensor do extrinsic calibration on the fly, there's no feedback @@alexandrubobaru