motional / nuplan-devkit

The devkit of the nuPlan dataset.
https://www.nuplan.org
Other
673 stars 129 forks source link

Why absent of latitudinal jerk of ego in the closed-loop simulation metrics? #392

Open ghai2023 opened 1 month ago

ghai2023 commented 1 month ago

Hi, I am trying to do closed-loop simulation (the nonreactive one firstly so far) with nuPlan.

Among the low level metrics by defaults, _ego_jerkstatistics and _ego_lon_jerkstatistics are utilized in the file of _simulation_closed_loop_nonreactiveagents.yaml (path: \nuplan-devkit\nuplan\planning\script\config\common\simulation_metric\simulation_closed_loop_nonreactive_agents.yaml). Naturally, I'd like to ask, why did the metric of _ego_latjerk not show up in this yaml file? I do see the computing in the python file of _ego_latjerk.py and the yaml file of _ego_lat_jerkstatistics.yaml.

Of course, I understand that the information is included inside of _ego_jerkstatistics and _ego_lon_jerkstatistics. Sometimes, it did not go in this way, like the pair of ego_lat_acceleration_statistics and ego_lon_acceleration_statistics...

Thanks for any help that you can provide.

ghai2023 commented 1 month ago

@holger-motional @patk-motional