Astribot SmoothRL keeps S1 moving during async inference
Loading video
Loading videoAstribot's SmoothRL trains S1 with online RL during async inference, so the humanoid keeps moving while the model computes the next action chunk.
Category: humanoid
Author: @XRoboHub
Date: 2026-09-03T00:00:00
Duration: 119.002s
Reference: https://arxiv.org/abs/2608.29768
