Skip to content

Bundled Contact Gradients: Stabilizing Differentiable Simulation for Deployable Dynamic Tasks

Sep 2026 · arXiv.org · Vol abs/2609.30951 · 1 citation · 44 references
Computer Science

TL;DR

Bundled Contact Gradients is proposed, a contact-local randomized smoothing framework for differentiable policy learning that evaluates a local bundle of randomized perturbation rollouts around the stiff contact configuration and aggregates their gradient signal thereby reducing gradient variance.

Abstract

Differentiable simulation provides analytic gradients of robot dynamics, enabling fast and sample-efficient first-order policy optimization. However, obtaining smooth and informative gradients through rigid-body contact typically requires softened contact models, often at the expense of physical fidelity and thereby limiting learned policies largely to simulation. This trade-off becomes particularly consequential for dynamic humanoid motions, where accurate contact dynamics are critical for transferring policies to the real world. Increasing contact stiffness in rigid-body simulation improves the fidelity of interactions, but also makes the dynamics increasingly sensitive to small state perturbations, producing high-variance gradients that can destabilize first-order policy learning. To address this, we propose \emph{Bundled Contact Gradients (BCG)}, a contact-local randomized smoothing framework for differentiable policy learning. When stiff contact is detected, our method evaluates a local bundle of randomized perturbation rollouts around the stiff contact configuration and aggregates their gradient signal thereby reducing gradient variance. We demonstrate the effectiveness of our method by successfully training and transferring dynamic motions zero-shot onto a real-world Unitree G1 humanoid platform. Videos and supplementary information can be found at https://bundledcontactgradients.github.io/

View source

Similar papers

#machine learning Open access Sep 2026

Ostrich: Taking Large Strides Through Stiff Contact in Differentiable Dynamics

Ostrich is presented, a GPU-accelerated rigid-body simulator that resolves hard contacts and friction with non-smooth Newton iteration at large timesteps, and differentiates the converged residual via the implicit function theorem, reusing the forward Schur complement to compute the adjoint at O(1) memory per timestep.

A. Kučera, K. Zimmermann · 2 citations
Aug 2026

Adaptive and Smooth Model Predictive Path Integral Control for Highly Dynamic Legged Locomotion

An enhanced model predictive path integral control framework that coordinates adaptive nominal-sequence update and perturbation generation within the standard control loop is proposed, which improves tracking accuracy, disturbance recovery, and contact-force smoothness while preserving the standard rollout-and-reweight...

Xian-Tao Sun, Sijie Li, Bin Lan et al. · 0 citations
Open access Sep 2026

Assisted exploration of difficult transitions in humanoid motion tracking

Humanoid motion tracking provides a scalable interface for transferring kinematic human motion to whole-body robot control, but highly dynamic and contact-rich references remain difficult because underactuation and discontinuous contacts can restrict the closed-loop state distributions explored during training. Externa...

Chao-Jie Fu, Cheng-Kai Su, Lei Jiang et al. · 0 citations
Preprint Oct 2026

Learning a Resolution-Consistent Jacobian Field for Bio-Inspired Rigid-Soft Finger

Bio-inspired tendon-driven rigid-soft coupled dexterous fingers exhibit strong nonlinearity and configuration-dependent sensitivity, making accurate modeling challenging. In discrete-time control, Jacobian-based kinematic algorithms typically rely on point-wise local linear approximations, which makes their performance...

Tian-You Liang, Hai-Sen Zeng, Shan-Jun Chen et al. · 0 citations
Preprint Sep 2026

Contact-Implicit Stein Projected ADMM for Discovery of Diverse Contact-Rich Manipulation Strategies

Contact-implicit trajectory optimization formulates contact-rich manipulation as a single constrained program; however, that single program run collapses onto one local optimum out of many equally valid contact modes, grasps, or push directions. As a consequence, the resulting manipulation strategy is reluctant to chan...

Hrishikesh Sathyanarayan, Christian Hughes, Ian Abraham · 0 citations
Preprint Sep 2026

History-Conditioned Flow Matching for Probabilistic Dynamics of Tendon-Driven Continuum Robots

Deterministic dynamics modeling of tendon-driven continuum robots remains challenging owing to uncertainties in material behavior, tendon transmission, friction, and contact. Measured joint configurations and nominal tendon commands do not fully characterize these internal mechanical factors, leaving uncertainty in the...

Hang-Tao Yang, Ting-Cong Liu, Jun-Jie Xiong et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.