Autonomous vehicles have increasingly become a trend in transportation owing to the development of deep learning technologies. However, autonomous vehicle systems are mainly applied in specific domains due to the significant variations in traffic patterns and unexpected situations on the road. Consequently, fully autonomous driving faces numerous challenges. We developed a decision-making system for autonomous vehicles to provide information by analyzing various scenarios during construction. The route from Keelung to Badouzi was selected for the development. To address the issue of lane occupation in construction zones, traffic cones are placed as guiding lane markers. An algorithm that utilizes construction signage recognition was created to identify navigable roads for the lane recognition of autonomous vehicles. Considering the situation with narrowing construction zones and sudden lane changes by vehicles ahead, we developed a leading vehicle trajectory tracking algorithm, too. This algorithm tracks the trajectory of the leading vehicle to correspond to sudden lane changes. The construction signage recognition and vehicle trajectory tracking algorithms significantly enhance the decision-making capabilities of autonomous vehicles in construction zones.