About physicsClientId, how to train a model parallelly??

[JiayunjieJYJ]

I find pybullet is very slow when I run use getCameraImage() in both GUI and DIRECT mode. So how to train a common RL algorithm in multi-physicsClient???

It means the data of different physicsClient can be collected into one common data collection at the same time, is there any demo about it ???

Thanks!!!