Comment by ArashEdalat

we have action chunking in robotics for the same reason so VLAs can analyze faster and come to actions, the latency and inference are both slow so many steps need to come together for this to happen correctly