/rescue

algorithms for resilient consensus and coordination of multi-agent systems with intermittent communication

Primary LanguageJupyter NotebookMIT LicenseMIT

rescue

An observer-based algorithm for resilient consensus and cooperation (formation) over time-varying networks, subject to intermittent connections and a worst-case number of non-cooperative (adversarial) agents.

The resilience of an N-agent time-varying network $\mathcal{G}_t$ against a worst-case set of adversarial agents is lower bounded by the graph (network) algebraic connectivity in an integral sense that is defined as ($\mu$, $T$)-Persistency-of-Excitation connectivity $\frac{1}{T} \int_{t-T}^{T} ( L(G_{\tau}) + \frac{1^{}_N 1^T_N}{N} ) d\tau \geq \mu I_{N-1}$, where $T > t \geq 0$ and $ L(G_{t})$ is the Laplacian matrix.

Installation

git clone https://github.com/r-bahrami/rescue
pip install -r requirements.txt
# use exp1.ipynb or exp2.ipynb to reproduce the results

Example