SPRIG: Stackelberg Perception-Reinforcement Learning with Internal Game DynamicsFernando MartinezJuntao Chenet al.2025AAAI 2025