1: \begin{abstract}

2: In this paper, we develop a novel dynamic distributed optimal safe consensus protocol to achieve simultaneously safety requirements and output optimal consensus. Specifically, we construct a distributed projection optimization algorithm with an expanding constraint set in the decision-making layer, while we propose a reference-tracking safety controller to ensure that each agent's output remains within a shrinking safety set in the control layer. We also establish the convergence and safety analysis of the closed-loop system using the small-gain theorem and time-varying control barrier function (CBF) theory, respectively.

3: Besides, unlike previous works on distributed optimal consensus, our approach does not require prior knowledge of the cost function the local objective or gradient function and adopts a mild assumption on the dynamics of multi-agent systems (MASs) by using the transmission zeros condition.

4:  \end{abstract}

5: