Abstract:We introduce a compositional data-driven methodology for designing fully-decentralized safety controllers applicable to large-scale interconnected networks, encompassing subsystems with unknown mathematical models. Our compositional scheme leverages the interconnection topology and breaks down the network analysis into the examination of distinct subsystems. This is accompanied by utilizing a concept of control storage certificates (CSCs) to capture joint dissipativity-type properties among subsystems. These CSCs are instrumental in a compositional derivation of a control barrier certificate (CBC) specialized for the interconnected network, thereby ensuring its safety. In our data-driven scheme, we gather solely one input-output trajectory from each unknown subsystem within a specified time frame. By fulfilling a specific rank condition, this process facilitates the construction of a CSC for each subsystem. Following this, by adhering to compositional dissipativity reasoning, we compose CSCs derived from data and build a CBC for the unknown network, ensuring its safety over an infinite time horizon, while providing correctness guarantees. We demonstrate that our compositional data-driven approach significantly enhances the design of a CBC and its safety controller across the interconnected network. This advancement is achieved by reducing the computational complexity from a polynomial growth in relation to network dimension, when using sum-of-squares (SOS) optimization, to a linear scale based on the number of subsystems. We additionally demonstrate that the dissipativity-type compositionality condition can benefit from the structure of interconnection topology and potentially be fulfilled regardless of the number of subsystems. We apply our data-driven findings to a variety of benchmarks, involving physical networks with unknown models and diverse interconnection topologies.

MDP Abstractions from Data: Large-Scale Stochastic Networks

Relational State-Space Model for Stochastic Multi-Object Systems

Compositional Abstractions of Interconnected Discrete-Time Stochastic Control Systems

A Stability-Based Abstraction Framework for Reach-Avoid Control of Stochastic Dynamical Systems with Unknown Noise Distributions

Data-driven Interval MDP for Robust Control Synthesis

Data-Driven Control of Large-Scale Networks with Formal Guarantees: A Small-Gain Free Approach

Symbolic Abstractions with Guarantees: A Data-Driven Divide-and-Conquer Strategy

Sufficient Markov Decision Processes with Alternating Deep Neural Networks

Large Deviations of a Network of Neurons with Dynamic Sparse Random Connections

From Data to Control: A Formal Compositional Framework for Large-Scale Interconnected Networks

Distributed Markov Chains

Abstraction-Refinement for Hierarchical Probabilistic Models

Formal Abstraction of General Stochastic Systems via Noise Partitioning

Compositional Controller Synthesis for Interconnected Stochastic Systems with Markovian Switching

Compositional Reasoning for Weighted Markov Decision Processes

A GRU-based Mixture Density Network for Data-Driven Dynamic Stochastic Programming

Stochastic reaction networks within interacting compartments

Unifying Safety Approaches for Stochastic Systems: From Barrier Functions to Uncertain Abstractions via Dynamic Programming

Data-driven Construction of Finite Abstractions for Interconnected Systems: A Compositional Approach

Robust Anytime Learning of Markov Decision Processes

Scalable control synthesis for stochastic systems via structural IMDP abstractions