Abstract: Rural transportation comes with unique technical challenges for autonomous vehicles. These challenges arise from the high-speed movement of other vehicles, curvy roads, roadside vegetation, and intersections with nonstandard design. Such conditions, including curvy roads and vegetation, limit visibility and create blind spots, making perception and decision-making more difficult. To this end, we developed an autonomous vehicle (AV) testbed designed to advance safe, reliable, and accessible mobility in rural environments, demonstrating the potential of autonomous driving to improve transportation and quality of life for these communities. Also, we have developed a learning-based approach that enables AVs to make adaptive, risk-aware decisions during intersection navigation and handling challenging scenarios such as unprotected left turn. Our method prioritizes safety by reasoning over past observations to mitigate partial observability, accounting for limited visibility and uncertainties in oncoming traffic, and supporting real-time operation. The developed method is tested and validated experimentally.