Kamino: GPU-based Massively Parallel Simulation of Multi-Body Systems with Challenging Topologies

TL;DR

Kamino is a GPU-based physics engine that enables high-fidelity, massively parallel simulation of complex multi-body systems with challenging topologies, supporting data-driven methods like reinforcement learning.

Contribution

It introduces native support for strongly coupled kinematic loops in GPU simulations, facilitating accurate and efficient modeling of complex mechanical systems.

Findings

Simulated 4096 environments in parallel on a single GPU.

Successfully trained RL policy for a complex biped with multiple kinematic loops.

Achieved high-fidelity contact dynamics without approximation.

Abstract

We present Kamino, a GPU-based physics solver for massively parallel simulations of heterogeneous highly-coupled mechanical systems. Implemented in Python using NVIDIA Warp and integrated into the Newton framework, it enables the application of data-driven methods, such as large-scale reinforcement learning, to complex robotic systems that exhibit strongly coupled kinematic and dynamic constraints such as kinematic loops. The latter are often circumvented by practitioners; approximating the system topology as a kinematic tree and incorporating explicit loop-closure constraints or so-called mimic joints. Kamino aims at alleviating this burden by natively supporting these types of coupling. This capability facilitates high-throughput parallelized simulations that capture the true nature of mechanical systems that exploit closed kinematic chains for mechanical advantage. Moreover, Kamino supports heterogeneous worlds, allowing for batched simulation of structurally diverse robots on a single GPU. At its core lies a state-of-the-art constrained optimization algorithm that computes constraint forces by solving the constrained rigid multi-body forward dynamics transcribed as a nonlinear complementarity problem. This leads to high-fidelity simulations that can resolve contact dynamics without resorting to approximate models that simplify and/or convexify the problem. We demonstrate RL policy training on DR Legs, a biped with six nested kinematic loops, generating a feasible walking policy while simulating 4096 parallel environments on a single GPU.