Ego motion is not considered about ab3dmot

[xinshuoweng]

Great to see such a discussion here! All are correct and great! I just want to share one update here that the ego motion compensation feature has been added last year, see:

AB3DMOT/AB3DMOT_libs/model.py

Line 433 in 8c17f16

trks = self.ego_motion_compensation(frame, trks)

[Po-Jen]

Interesting question, I want to share some thoughts for discussion.

Since this work uses constant velocity model, it's actually good if the ego vehicle and the target vehicle are moving at the same speed. Because in that case, the target vehicle's velocity are constantly 0, which fits perfectly to the constant velocity assumption for Kalman Filter prediction.

Sure there's no doubt that not considering ego motion is not perfect, but I think we should bear in mind that this work aims at providing a simple and efficient baseline method for benchmarking in future research. To me, it's already surprising to see AB3DMOT has similar performance as other SOTA methods.

[Sonne-Zhu]

Hi thanks for the project

I have an question to ask regarding ego motion of vehicle.

from your paper its clear that baseline aglorithm will not consider the ego motion of vehicle ( i.e motion of the car on which we place lidar )

if it so, then you how will you justify for that? for example , if the ego vehicle and target vehicle both are moving parallel to each with constant speed. in this scenario, target vehicle will be detected as static object with zero velocity

I'm confused about this, in the 2020iros paper, isn't the ego's motion obtained from GPS/IMU? The coordinates of the detected objects will be converted to the world system coordinate system through the ego pose, and then associated with the tracking results. If the ego vehicle and target vehicle both are moving parallel to each with constant speed. Isn’t the real speed of the object equal to the speed of the ego vehicle?