aiqbench analysis
cuPIQP is a GPU-based solver for batched, differentiable quadratic programs, aimed at training robot policies end-to-end through model-based controllers. Its reported gains (up to 18.8x over Moreau, 4.6x faster training over QPAX) are vendor-style benchmarks from the authors, so they need independent replication across hardware and problem types. For ML readers, the practical question is whether differentiable optimization layers can become routine in GPU training pipelines.
CuPIQP: A Batched and Differentiable QP Solver on GPUs for Robot Learning and Control
arXiv:2610.41021v1 Announce Type: cross Abstract: Quadratic programs (QPs) are fundamental building blocks of many robotic tasks such as inverse kinematics, whole-body control, and model predictive control. Extending these methods to large-scale robot learning and control requires solving many QPs concurrently and, for end-to-end learning, differentiating through their solutions. Efficient batched optimization is therefore essential to bringing established model-based methods into GPU-based robotics pipelines. We present \textsc{cuPIQP}, a batched and differentiable QP solver on GPUs based on a proximal interior-point method, with unified support for dense, sparse, and multistage problems. The solver exploits parallelism across independent QPs and their internal structure while preserving per-problem numerical updates and termination decisions. Benchmarks with up to 16384 QPs per batch demonstrate solve-time speedups of up to $18.8\times$ over \textsc{Moreau} and $4.26\times$ over eight-core \textsc{PIQP}. In an end-to-end learning task, \textsc{cuPIQP} accelerates training by $4.6\times$ over \textsc{QPAX} and $7.8\times$ over \textsc{Moreau}. By reducing the computational cost, \textsc{cuPIQP} provides efficient, reusable infrastructure for integrating model-based optimization into large-scale robot learning and control.
Abstract:Quadratic programs (QPs) are fundamental building blocks of many robotic tasks such as inverse kinematics, whole-body control, and model predictive control. Extending these methods to large-scale robot learning and control requires solving many QPs concurrently and, for end-to-end learning, differentiating through their solutions. Efficient batched optimization is therefore essential to bringing established model-based methods into GPU-based robotics pipelines. We present \textsc{cuPIQP}, a batched and differentiable QP solver on GPUs based on a proximal interior-point method, with unified support for dense, sparse, and multistage problems. The solver exploits parallelism across independent QPs and their internal structure while preserving per-problem numerical updates and termination decisions. Benchmarks with up to 16384 QPs per batch demonstrate solve-time speedups of up to $18.8\times$ over \textsc{Moreau} and $4.26\times$ over eight-core \textsc{PIQP}. In an end-to-end learning task, \textsc{cuPIQP} accelerates training by $4.6\times$ over \textsc{QPAX} and $7.8\times$ over \textsc{Moreau}. By reducing the computational cost, \textsc{cuPIQP} provides efficient, reusable infrastructure for integrating model-based optimization into large-scale robot learning and control.
Optimization and Control (math.OC); Robotics (cs.RO)
arXiv:2610.41021 [math.OC]
(or arXiv:2610.41021v1 [math.OC] for this version)
https://doi.org/10.48550/arXiv.2610.41021 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 258 words
Related coverage
An Adaptive Gradient Clipping and Noise Injection Mechanism for Differentially Private Federated Learning
arXiv · Aug 19, 2026
DiffDreamer: Coupling World Models with Differentiable Simulation for Agile Quadrotor Control
arXiv · Oct 23, 2026
Finite Element-Based Material Learning via Automatic Differentiation: Learning constitutive neural network models from full-field deformation data
arXiv · Jun 6, 2026
Learn-Then-Differentiate Gradient Estimation
arXiv · Oct 1, 2026