Resilient and Adaptive Autonomy Using Multi-Agent Reasoning

Abstract

The ability to predict catastrophic system events is important for computers and autonomous systems overall, in order to employ mitigations or at least minimize damage. Predicting these “tipping points” using ma-chine learning is one way to forecast when a system will destabilize and the time before its expected failure. Frequently, the response time between detection and implementing a mitigation or shutdown needs to be at machine speeds vs. human-response speeds. We created and tested an algorithmic “toolkit” for different scales of complex systems, ranging from a small nonlinear analog system to a large power grid system. All of the algorithms can run without GPU enhancements, with the most lightweight nonlinear algorithm exhibiting zero-shot learning. We provided a second set of experiments with collaborative agents, which may allow zero-shot algorithms to be used in critical components to extend resilience for complex autonomous systems.