Driving Like an Experienced, Well-Behaved Human Driver: Designing and Evaluating a Spatial Social Agent for Autonomous Vehicles Through Co-Design and Trust Measurement
Author(s)
Yang, Huaizhuo
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
To promote people’s acceptance of autonomous vehicles (AV), designers have been trying to calibrate user trust in AV through novel human-AV interaction designs. In this thesis, we explore what kind of anthropomorphic AV interface designs may better enhance user trust in AV systems. More specifically, we propose the novel concept of “an AV interface that is perceived as an experienced, well-behaved human driver” since people tend to place more trust in experienced, well-behaved human drivers than in autonomous driving systems. We explored this design concept through two sessions of co-design workshops with 12 industrial designers across all levels. We employed “Annotated Portfolio” to annotate the graphical design materials collected in the workshop and analyzed these annotations together with the transcripts generated from the workshop through Reflexive Thematic Analysis. The results suggest novel design insights and guidelines for an AV interface that plays the social role of an experienced, well-behaved human driver.
The design insights from this analysis informed the development of a spatial social agent prototype, which was implemented and tested within a virtual reality (VR) simulation to assess its effectiveness in fostering user trust in AV systems. The final prototype (EWDA) integrates key anthropomorphic elements that emulate the behaviors and communication patterns of an experienced, well-behaved human driver, creating a more relatable and reassuring AV interaction experience. To evaluate its impact, user testing with human-vehicle perception and human-vehicle trust measurement was conducted, examining whether the proposed design was perceived as a well-behaved, experienced human driver and significantly enhances user trust levels. The findings indicate that incorporating anthropomorphic and social interaction features into AV systems can effectively boost user trust and acceptance, highlighting the potential of the proposed anthropomorphic AV design strategies in addressing trust calibration challenges.
Sponsor
Date
2025-05-01
Extent
Resource Type
Text
Resource Subtype
Thesis