FlashNeRD: Performance-First Contact-Rich Neural Robot Dynamics

Mohammadmehdi Ataei, Mohammad Amin Nabian, Jie Xu, Miles Macklin, Yashraj Narang, Eric Heiden, Octi Zhang, Yunrong Guo, Ruben D'Sa, Oliver Hennigh, Gavriel State, Sanjay Choudhry · N/A · 2026

Summary

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

Abstract Summary

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. Three limitations remain. NeRD offers little speedup over the simulator it learned from, accepts contact only at predefined points, and has no two-way coupling with objects it manipulates. FlashNeRD removes all three with a parallel streaming architecture that makes each prediction faster and more accurate, an encoder that accepts contacts wherever they occur, and two-way coupling with objects simulated by analytical solvers. Experiments across five robots show faster and more accurate dynamics, faster policy learning, and faster inference-time planning. Across three robots, FlashNeRD's dynamics model is up to $55\times$ faster than an optimized NeRD and more accurate over long rollouts. 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. With DIAL-MPC, a sampling-based MPC method, the robot climbs all six test platforms where fixed-contact NeRD manages one, at approximately half the analytical simulator's planning time. Cube-reorientation policies trained with FlashNeRD complete within 2% of the simulator-trained policy's target count.

Key Points

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.
Three limitations remain
NeRD offers little speedup over the simulator it learned from, accepts contact only at predefined points, and has no two-way coupling with objects it manipulates.
FlashNeRD removes all three with a parallel streaming architecture that makes each prediction faster and more accurate, an encoder that accepts contacts wherever they occur, and two-way coupling with objects simulated by analytical solvers.

Method

FlashNeRD has three components, one per limitation above, all built on the shared transition-learning problem of Sec. III-A . The first is a transformer that evaluates attention and feedforward branches in parallel within each block and caches past keys and values, so each rollout step encodes only the newest frame (Secs. III-B and III-C ).

The second is a contact encoder that accepts contacts wherever the collision detector finds them (Sec. III-D ).

Method — FlashNeRD: Performance-First Contact-Rich Neural Robot Dynamics — Fig. 4: Model-only throughput and prediction accuracy on CartPole, Franka, and G1. Each spoke shows reciprocal rollout NRMSE $A_{H}=1/E_{H}$ at $H=1,15,100,200$ or throughput $Q_{B}$ at batch sizes 16, 16,384, and 512,000. Values are on a logarithmic radial scale normalized by the better model on each spoke, so a larger radius is better. Rim labels give the FlashNeRD-to-Optimized-NeRD ratio, colored by the favored method.

Results

NeRD has already shown zero-shot sim-to-real transfer and fine-tuning on real data [ 4 ] , so the experiments below focus on what FlashNeRD adds. Section IV-A measures the transition model alone, Secs. IV-B and IV-C place it inside policy learning and planning, and Sec. IV-D couples it to an analytically simulated object.

We show that FlashNeRD increases model-only inference throughput while improving prediction accuracy over recursive rollouts, where predictions become inputs and errors compound.

Results — FlashNeRD: Performance-First Contact-Rich Neural Robot Dynamics — Fig. 5: PPO training reward versus accumulated training time (sum of logged iteration durations) for ANYmal-C velocity tracking with 16,384 environments over 300 iterations, with fixed contact for all learned models. Diamonds mark final policies evaluated in MJWarp with mean actions and the training reward, showing faster training reaches comparable quality.

Sources and Demonstrations

Figures are reproduced from Ataei et al. See the full paper for experimental details and the project page, when the paper links one, for demonstrations and videos.