Correlation Between Reaction Time, Multi-Modal Feedback and Take-Over Requests for Level 3 Automated Vehicles

Area of Science:

• Human-robot interaction
• Automated vehicle safety
• Human factors engineering

Background:

• The transition towards fully automated vehicles (AVs) necessitates effective human-robot take-over requests (TORs).
• Failure in TORs can compromise driver situational awareness (SA) and lead to safety risks.
• Understanding the impact of feedback modalities on control transfer is crucial for AV safety.

Purpose of the Study:

• To investigate the correlation between reaction time, multimodal feedback, and successful transfer of control in AVs.
• To evaluate the effectiveness of different feedback modalities (audio, visual, haptic) during TORs.
• To determine optimal reaction time windows for safe control transitions.

Main Methods:

• Experiments conducted in a full-sized driving simulator with 19 human participants.
• Drivers engaged in a secondary task while the vehicle was in self-driving mode.
• Seven feedback modalities for TORs were tested, including novel soft robotic seat haptics, with a 7-second window for regaining SA.

Main Results:

• Reaction times below 2.6 seconds or above 6 seconds led to unsuccessful control transfers.
• Haptic feedback, when combined with existing audio and visual cues, resulted in faster and safer control transfers.
• Multimodal feedback enhances driver's ability to regain situational awareness and react to incidents.

Conclusions:

• Specific reaction time windows are critical for successful TORs in AVs.
• Haptic feedback offers a significant advantage in improving the timeliness and safety of control transfers.
• Integrating advanced feedback mechanisms is essential for mitigating risks associated with AV automation level transitions.