Batched Differentiable Rigid Body Dynamics in PyTorch for GPU-Accelerated Robot Learning
This work significantly improves the throughput of rigid-body dynamics computation for large-scale, GPU-accelerated robot learning, benefiting researchers and practitioners in robotics and reinforcement learning.
This paper addresses the bottleneck of CPU-bound dynamics libraries in GPU-accelerated robot learning by introducing BARD, a PyTorch implementation of rigid-body dynamics algorithms. BARD achieves up to 64x higher throughput for Forward Kinematics and 63x for Jacobians compared to Pinocchio at batch size 4096, and is 8.5x faster than Pinocchio and 2.0x faster than ADAM in an Isaac Lab AMP training pipeline.
As robot control shifts toward large-scale reinforcement learning with in-loop dynamics computation, the community's reliance on CPU-bound libraries such as Pinocchio creates a throughput bottleneck in GPU-based training pipelines. We present BARD (Batched Articulated Rigid-body Dynamics), a self-contained PyTorch implementation of Featherstone's rigid-body dynamics algorithms, optimized for batched GPU evaluation and automatic differentiation. Three design choices make this efficient: a tiered lazy-evaluation cache that avoids redundant tree traversals, matmul-free joint transforms via pre-computed Rodrigues constants, and level-parallel propagation that reduces sequential operations to tree-depth batched steps. On five robot models (7-23 DOFs), BARD matches Pinocchio numerically while reaching up to 64x higher throughput for Forward Kinematics and 63x for Jacobians at batch size 4096 on an NVIDIA H200. We validate differentiability through gradient-based system identification on a 7-DOF manipulator, recovering link masses to 1.24% mean error under 5% torque noise, and integrate BARD into an Isaac Lab AMP training pipeline for an 11-DOF spined quadruped with 4096 parallel environments, where it is 8.5x faster than Pinocchio and 2.0x faster than ADAM for in-loop dynamics. BARD is open-sourced at: https://github.com/YueWang996/bard-pytorch-dynamics.