Locomotion policy trained in just 78 seconds in custom simulator

yacineMTB · x · 2026-08-30

The author adds that the locomotion policy mentioned in the previous post — trained in the custom simulator dingsim and transferable to MuJoCo — took only 78 seconds to train, highlighting the simulator's throughput and sim2real potential.

Related event: RL Locomotion Policy Trained in 78 Seconds on Custom Simulator(2 posts)→

Original post →

More from Embodied

Embodied channel →