Optimization-Free Safety and Task Execution for High-Order Multi-Agent Systems via STL-Guided Auxiliary Reference Governor

This paper presents an optimization-free control framework for high-order multi-agent systems that ensures both safety and task execution under Signal Temporal Logic specifications. The proposed approach integrates STL predicates into a dynamic auxiliary reference governor structure, enabling the transformation of high-relative-degree safety constraints into tractable first-order control barrier function formulations. A dynamic safety margin and navigation field are designed to guide each agent toward its STL goals while respecting actuation limits and avoiding dynamic obstacles. Unlike conventional high-order CBF or quadratic programs-based controllers, our method yields closed-form control laws. This approach effectively overcomes the computational bottlenecks associated with online optimization, thereby significantly improving scalability and ensuring real-time performance for high-order multi-agent systems. The framework supports decentralized execution, allowing agents to operate independently with guaranteed safety and task satisfaction in dynamic, shared environments. Simulation results validate the effectiveness and efficiency of the proposed control scheme in comparison to existing methods.