Baofeng Sun

Pairing multi-agent RL as a residual layer over physics-based model predictive control eliminates the out-of-distribution brittleness that typically prevents purely data-driven policies from operating in safety-critical physical systems.