RESiLIENT: A Neural-Symbolic Resilient Threat-Response Framework for Large-Scale Hierarchical Swarms Article Swipe
YOU?
·
· 2025
· Open Access
·
· DOI: https://doi.org/10.1007/978-981-95-1050-4_1
· OA: W4415406024
Achieving resilience has become an emerging challenge for large-scale swarm autonomy, entailing both the identification of unforeseen events and the recovery of affected systems from such events. The complexity of modern autonomous systems introduces new system vulnerabilities to adversarial attacks on both physical- and cyber-systems. These vulnerabilities, coupled with the intricate interconnection of large-scale swarm systems, make swarm-level resiliency difficult to obtain. A simple failure in one local system can cascade to others, leading to catastrophe. In this poster, we highlight how neural-symbolic concepts, combining the best of machine learning and control theory in a unified framework, can enhance the resilience of hierarchical swarm operations. Through several illustrative examples, we showcase the advantages of neural-symbolic approaches in two broad categories: proactive and reactive strategies for resilient hierarchical swarms. With strong interpretability, our approaches collectively achieve resilient planning, continuous evolution, and swift re-organization against unknown threats and anomalies.
