With the advancement of autonomous driving technology, especially the growing adoption of SAE Level 3 and above systems, drivers are transitioning from active controllers to supervisors who must take over in emergencies. For hearing-impaired drivers in a fatigued state, conventional voice alert systems often fail to provide timely and effective warnings, increasing safety risks. This study proposes an adaptive in-vehicle interface that combines visual and tactile feedback to address these challenges. Experiments were conducted to evaluate response accuracy, reaction time, and cognitive load under varying levels of driver fatigue. The findings show that the integration of visual and tactile cues significantly improves takeover efficiency and reduces mental strain in fatigued drivers. These results highlight the potential of multimodal designs in enhancing the safety and driving experience for hearing-impaired individuals. By providing practical strategies and evidence-based insights, this research contributes to the development of more inclusive and effective interaction designs for future autonomous driving systems.