ALARMS

Algorithmic and Logic-Driven Approaches for Repair Mechanisms of Symbolic AI

Advances in autonomous systems, multi-robot coordination, distributed computing, and AI-driven decision-making are transforming society and technology. Symbolic AI and Multi-Agent Systems (MAS) are central to modeling, reasoning, and controlling these complex systems. In safety-critical domains such as healthcare, cybersecurity, and advanced manufacturing, ensuring correct system behavior is paramount. Yet current design and adjustment methods are often ad hoc, manually intensive, and lack systematic mechanisms for identifying and repairing errors.

This project develops a formal, algorithmic framework for automated repair mechanisms in symbolic AI and MAS. By integrating advances in algorithmic reasoning, logic-driven approaches, and formal verification, ALARMS enables the detection of specification violations, miscoordination, or sub-optimal behavior, and the synthesis of corrective adaptations. This approach provides self-correcting, resilient mechanisms capable of maintaining correctness and robustness in dynamic and uncertain environments.

The project’s outcomes include formal frameworks, computational tools, and methodologies that facilitate rigorous specification repair, verification, and adaptive control. These contributions will advance the theory and practice of MAS and symbolic AI, impacting applications in planning, robotics, distributed protocols, and other intelligent systems that rely on formally verified and dynamically adaptable behaviors.