Developed an LLM-based multimodal AI guide that assists the transition between autonomous and manual driving across various Level 3 modes, reducing driver anxiety by 72% and improving safety by 87%.
Client
Independent project
Role
Product Designer
Period
8 weeks, 2022
Overview
After watching a video of a moving Tesla where both the driver and passengers were asleep, I realized how dangerous it is to overtrust semi-autonomous systems. To address the specific safety risks that occur during Level 3 handovers when control returns to the human driver, I designed a multimodal AI guidance system. The system combines contextual voice prompts, visual cues, HVAC adjustments, and audio alerts across six major driving modes. Internal testing showed a 72 percent reduction in driver anxiety, a 2.7 second faster response time, an 87 percent improvement in safety, and a user satisfaction score of 4.8 out of 5.
Client
Preview the final image
Genty is a context-aware AI assistant designed to enable seamless communication across devices from smartphones to vehicles for Level 3 autonomous driving.
Powered by an LLM specialized for autonomous systems, Genty engages in natural conversations with the driver and provides smooth, empathetic, and clear guidance during handover moments and potential risk situations, adapting to each driving mode.
Challenge
Design an AI assistant that adapts across multiple Level 3 driving modes and guides drivers through autonomous-to-manual transitions with clarity, empathy, and real-time awareness.
Objective
Create an AI guidance experience that understands each driving mode and communicates with empathy, helping drivers stay calm and aware during handover moments through multimodal feedback such as voice, visuals, and subtle environmental cues.
Result
Developed a functional prototype powered by an LLM-based engine that adapts to multiple driving modes, reducing driver anxiety during takeovers by 72 percent, improving safety by 87 percent, and increasing user satisfaction to 4.8 out of 5.
8 weeks, 2023
Final Videos
Grounded in secondary research, I designed a context-aware and adaptive AI system that responds to 5 common handover modes and one critical emergency event during autonomous driving.
Rest mode / Sleep mode / Video mode / Meal mode / Makeup mode / Emergency mode
Hello
Any questions?
Let's connect and build something meaningful together