How to Interact with a Fully Autonomous Vehicle: Naturalistic Ways for Drivers to Intervene in the Vehicle System...

Area of Science:

• Human-computer interaction
• Autonomous systems
• Automotive engineering

Background:

• Autonomous vehicle technology enables drivers to engage in non-driving related tasks (NDRTs).
• Driver-vehicle interaction requires input modalities that align with in-vehicle activities and situational context.
• Previous research overlooked the impact of physical and cognitive load during NDRTs on input acceptance.

Purpose of the Study:

• Investigate naturalistic interactions in fully autonomous vehicles during NDRTs.
• Evaluate driver acceptance of various input modalities across different intervention scenarios and NDRT engagement levels.
• Develop a decision-making model for natural input preferences in autonomous vehicles.

Main Methods:

• Online methodology with 360 participants evaluating seven natural input interactions (touch, voice, hand gesture, combinations).
• Four NDRTs with varying physical and cognitive engagement levels were presented.
• Six common intervention scenarios were tested, resulting in 24 distinct cases.

Main Results:

• Non-driving related tasks (NDRTs) exert a greater influence on driver input interaction than intervention scenarios.
• Physical load variations have a more significant impact on input selection compared to cognitive load variations.
• Driver preferences for natural inputs were analyzed across different NDRT and intervention conditions.

Conclusions:

• Driver engagement in NDRTs is a primary factor influencing input modality selection in autonomous vehicles.
• Physical demands of NDRTs are more critical than cognitive demands in determining preferred input methods.
• A decision-making model can guide User Experience designers in creating intuitive interfaces for autonomous vehicle drivers.