FlashNeRD: Performance-First Contact-Rich Neural Robot Dynamics

Key points

• Neural Robot Dynamics (NeRD) pursues this by keeping collision detection analytical and replacing a simulator's numerical dynamics for the robot with a learned model.1
• Experiments across five robots show faster and more accurate dynamics, faster policy learning, and faster inference-time planning.1
• This speedup extends to policy learning, where PPO trains an ANYmal locomotion policy in 34 seconds, $3.8\times$ faster than the analytical simulator and $2.7\times$ faster than an optimized NeRD.1
• Cube-reorientation policies trained with FlashNeRD complete within 2% of the simulator-trained policy's target count.1

Sources (1)

• [1]FlashNeRD: Performance-First Contact-Rich Neural Robot Dynamics
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 09:23 PM

  Compared with analytical physics, learned dynamics models promise robot simulation that is faster, inherently differentiable, and easily adaptable to real data.

  Neural Robot Dynamics (NeRD) pursues this by keeping collision detection analytical and replacing a simulator's numerical dynamics for the robot with a learned model.

Extractive summary: sentences quoted from the sources.