aiqbench analysis
This paper introduces Q-SVMPC, a novel framework that integrates Reinforcement Learning guidance with Stein Variational Gradient Descent to enhance Model Predictive Control in robotics. By casting trajectory optimization as posterior inference, the method effectively mitigates mode collapse and preserves solution diversity while maintaining training stability across navigation and manipulation tasks. The approach represents a significant step forward in combining probabilistic inference with learning-based control for complex dynamic systems.
Q-Guided Stein Variational Model Predictive Control via RL-informed Policy Prior
arXiv:2507.06625v4 Announce Type: replace-cross Abstract: Model Predictive Control (MPC) enables reliable trajectory optimization under dynamics constraints, but often depends on accurate dynamics models and carefully hand-designed cost functions. Recent learning-based MPC methods aim to reduce these modeling and cost-design burdens by learning dynamics, priors, or value-related guidance signals. Yet many existing approaches still rely on deterministic gradient-based solvers (e.g., differentiable MPC) or parametric sampling-based updates (e.g., CEM/MPPI), which can lead to mode collapse and convergence to a single dominant solution. We propose Q-SVMPC, a Q-guided Stein variational MPC method with an RL-informed policy prior, which casts learning-based MPC as trajectory-level posterior inference and refines trajectory particles via SVGD under learned soft Q-value guidance to explicitly preserve diverse solutions. Experiments on navigation, robotic manipulation, and a real-world fruit-picking task show competitive learning efficiency, strong final performance, and training stability compared with MPC, model-free RL, and learning-based MPC baselines.
Abstract:Model Predictive Control (MPC) enables reliable trajectory optimization under dynamics constraints, but often depends on accurate dynamics models and carefully hand-designed cost functions. Recent learning-based MPC methods aim to reduce these modeling and cost-design burdens by learning dynamics, priors, or value-related guidance signals. Yet many existing approaches still rely on deterministic gradient-based solvers (e.g., differentiable MPC) or parametric sampling-based updates (e.g., CEM/MPPI), which can lead to mode collapse and convergence to a single dominant solution. We propose Q-SVMPC, a Q-guided Stein variational MPC method with an RL-informed policy prior, which casts learning-based MPC as trajectory-level posterior inference and refines trajectory particles via SVGD under learned soft Q-value guidance to explicitly preserve diverse solutions. Experiments on navigation, robotic manipulation, and a real-world fruit-picking task show competitive learning efficiency, strong final performance, and training stability compared with MPC, model-free RL, and learning-based MPC baselines.
Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2507.06625 [cs.RO]
(or arXiv:2507.06625v4 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2507.06625 Focus to learn more arXiv-issued DOI via DataCite
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 · 239 words
Related coverage
News
Quantum Computing Report · Jul 25, 2026
The Machines that Make the Machines
NVIDIA · Oct 7, 2026
AgentGFM: A Graph Foundation Model with Node-Agent Information-Flow Control
arXiv · Jul 31, 2026
AHA-WAM:Asynchronous Horizon-Adaptive World-Action Modeling with Observation-Guided Context Routing
arXiv · Jun 9, 2026