Explainable AI-Enhanced Supervisory Control for Robust Multi-Agent Robotic Systems

Explainability enables human oversight of autonomous systems: The framework uses an "explainable predictor" that maps mission requirements to control parameters and expected performance outcomes (energy use, tracking error), allowing human operators to understand and audit why the system makes specific decisions in real-time rather than treating it as a "black box."

Formal verification improves safety in multi-agent robotics: The system uses "timed-automata" for mode switching, which provides mathematically auditable transitions between different control states—this formal approach creates a verifiable safety record that can be reviewed, unlike traditional control systems that lack this transparency.

The same AI framework works across very different robotic domains: The approach was successfully validated on both spacecraft formation flying and underwater vehicles despite their completely different physics (gravity vs. hydrodynamic drag), suggesting this explainable AI method could be broadly applicable to various multi-agent robotic systems rather than being domain-specific.

Significant real-world performance improvements with measurable trade-offs: In spacecraft tests, the system achieved 21.7% lower tracking error and 81.4% lower energy consumption compared to traditional controllers, while the explainable predictor allows operators to see and understand these performance trade-offs before deployment.

Transparency supports accountability in autonomous systems: By combining formal verification (auditable mode switching) with explainable AI predictions, the framework creates a system where decisions can be traced and justified—critical for responsible deployment of multi-agent robots in safety-critical applications.