Closed Freeecode closed 4 months ago
Tkanks for sharing the great work! I have two problem about the NTU4DRadLM dataset.
Firstly, the timestamp of each radar point is missing, which is necessary to improve the accuracy of localization, and I found the proposed method, 4DRadarSLAM, also consider the motion distrotion. In addition, I also found the MSC RAD4R dataset(https://mscrad4r.github.io/) also not provide the timestamp of each radar point. Thereofore, I want to know wheather the timestamp of each radar point is provided by by the 4D radar sensor, or the post-process ignore the timestamp of each radar point when converting the raw data to the sensor_msgs/PointCloud2.
Secondly, I am curious about the velocity of each Radar point, I think the veclocity of each radar point, provided by the NTU4DRadLM dataset, is the relative radial velocity, is right? And the data, called radar_enhanced_pcl, has been filtered according to the absolute (i.e. ego motion compensated) radial velocity of the point, and only keep static points?
Looking forward to your reply! Best wishes!
I checked with Oculii customer service, they didn't output the timestamp of each radar point, so it cannot be obtained.
Second, yes it is relative radial velocity. radar_enhanced_pcl is the raw output by Oculii driver (the manufacturer output this with their special algorithm), it is still relative radial velocity (not absolute velocity).
I get it! Thanks for your reply!
Tkanks for sharing the great work! I have two problem about the NTU4DRadLM dataset.
Firstly, the timestamp of each radar point is missing, which is necessary to improve the accuracy of localization, and I found the proposed method, 4DRadarSLAM, also consider the motion distrotion. In addition, I also found the MSC RAD4R dataset(https://mscrad4r.github.io/) also not provide the timestamp of each radar point. Thereofore, I want to know wheather the timestamp of each radar point is provided by by the 4D radar sensor, or the post-process ignore the timestamp of each radar point when converting the raw data to the sensor_msgs/PointCloud2.
Secondly, I am curious about the velocity of each Radar point, I think the veclocity of each radar point, provided by the NTU4DRadLM dataset, is the relative radial velocity, is right? And the data, called radar_enhanced_pcl, has been filtered according to the absolute (i.e. ego motion compensated) radial velocity of the point, and only keep static points?
Looking forward to your reply! Best wishes!