Overview
Level 3 is the only stage where handovers from autonomous to manual driving occur. To address the safety risks inherent in these transitions, I designed a comprehensive AI-driven guide that delivers multi-sensory alerts—combining AI conversation, visual cues, HVAC, adjustments, and audio—across 6 critical handover scenarios. Internal validation demonstrated a 72% reduction in driver anxiety, a 2.7-second faster reaction time, an 87% improvement in safety, and a user satisfaction score of 4.8 out of 5.
Tools/Tech: Figma, Framer, Photoshop, etc.
Category
User experience
UI design
Visual design
Background
I was shocked when I saw a news showing both the driver and the passenger asleep in a moving Tesla. It was surprising to see how easily people can fall asleep, even though Tesla’s system is only a Level 2.5 driver-assist technology that still requires full attention. I also learned that the alert used to wake the driver is just a simple “beep” sound. But Level 3 autonomous driving poses an even greater risk: the car can fully control itself under certain conditions, which may lead drivers to trust it too much. If the driver falls into a deeper sleep or becomes distracted, the transition back to manual control can be extremely dangerous. That’s why I decided to create a multisensory AI handover guide—to provide a safer and more intuitive driving experience for both the driver and the passenger.
Video source: NBC News
Key issue
Easy to fall asleep
Lacks context-aware guidance
No emergency response
Challenge
Only Level 3 autonomous vehicles require the driver to retake control during a handover, a moment that often involves delayed reaction, stress, and significant safety risks. However, existing systems lack real-time, intuitive support to guide drivers through this critical transition.
Objective
The goal was to design a comprehensive UX solution for Level 3 control transitions by identifying key scenarios based on real driving data, and applying LLM-powered AI guidance with multisensory feedback to reduce driver anxiety and improve response time.
Result
The system, optimized for six real-world scenarios, reduced driver anxiety by 72%, improved reaction time by 2.7 seconds, increased safety by 87%, and achieved a user satisfaction score of 4.8/5 in internal testing.
