Uncertainty-Aware and Temporally Regulated Expert Advice in Reinforcement Learning for Autonomous Driving

In three linesReinforcement learning framework for autonomous driving using uncertainty-aware expert advice with adaptive thresholds. Epistemic and aleatoric uncertainty trigger expert intervention; commitment-cooldown strategy prevents long-term dependence. CARLA experiments: +5-7% success rate vs IQN baseline.

Summary generated by Claude — human-verified