aiqbench analysis
This research advances safe reinforcement learning by enabling end-to-end training of high-dimensional controllers using Control Barrier Functions (CBFs) without the computational bottlenecks that previously limited systems to low dimensions. By integrating Jacobian-Free Backpropagation with operator splitting, the authors theoretically and empirically demonstrate scalability in multi-agent nonlinear control problems exceeding 1200 state variables while maintaining hard safety guarantees. This methodological breakthrough is significant for deploying robust AI agents in complex physical environments where failure is not an option.
End-to-End Learning of Safe Optimal Feedback Control in High Dimensions with Control Barrier Function Layers
arXiv:2607.20674v1 Announce Type: new Abstract: We consider the problem of learning high-dimensional semi-global feedback controllers under hard safety constraints enforced by control barrier functions (CBFs). Incorporating CBFs into end-to-end policy training requires embedding a quadratic-program-based safety filter as an optimization layer, but computational and differentiation bottlenecks have largely restricted prior approaches to low-dimensional systems, typically with at most 16 state dimensions. We address this limitation by combining operator splitting with the recently developed Jacobian-Free Backpropagation (JFB) method to enable scalable end-to-end training while preserving hard safety guarantees through the CBF safety filter. We justify this training methodology theoretically using nonsmooth analysis techniques and demonstrate its effectiveness on high-dimensional multi-agent nonlinear control problems with state and control dimensions up to 1200 and 400, respectively.
Abstract:We consider the problem of learning high-dimensional semi-global feedback controllers under hard safety constraints enforced by control barrier functions (CBFs). Incorporating CBFs into end-to-end policy training requires embedding a quadratic-program-based safety filter as an optimization layer, but computational and differentiation bottlenecks have largely restricted prior approaches to low-dimensional systems, typically with at most 16 state dimensions. We address this limitation by combining operator splitting with the recently developed Jacobian-Free Backpropagation (JFB) method to enable scalable end-to-end training while preserving hard safety guarantees through the CBF safety filter. We justify this training methodology theoretically using nonsmooth analysis techniques and demonstrate its effectiveness on high-dimensional multi-agent nonlinear control problems with state and control dimensions up to 1200 and 400, respectively.
Machine Learning (cs.LG); Systems and Control (eess.SY); Optimization and Control (math.OC)
arXiv:2607.20674 [cs.LG]
(or arXiv:2607.20674v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2607.20674 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 · 220 words
