Effects of Semantic Segmentation Visualization on Trust, Situation Awareness, and Cognitive Load in Highly Automated Vehicles

要旨

Autonomous vehicles could improve mobility, safety, and inclusion in traffic. While this technology seems within reach, its successful introduction depends on the intended user's acceptance. A substantial factor for this acceptance is trust in the autonomous vehicle's capabilities. Visualizing internal information processed by an autonomous vehicle could calibrate this trust by enabling the perception of the vehicle's detection capabilities (and its failures) while only inducing a low cognitive load. Additionally, the simultaneously raised situation awareness could benefit potential take-overs. We report the results of two comparative online studies on visualizing semantic segmentation information for the human user of autonomous vehicles. Effects on trust, cognitive load, and situation awareness were measured using a simulation (N = 32) and state-of-the-art panoptic segmentation on a pre-recorded real-world video (N = 41). Results show that the visualization using Augmented Reality increases situation awareness while remaining low cognitive load.

著者
Mark Colley
Ulm University, Ulm, Germany
Benjamin Eder
Ulm University, Ulm, Germany
Jan Ole Rixen
Institute of Media Informatics, Ulm, Germany
Enrico Rukzio
University of Ulm, Ulm, Germany
DOI

10.1145/3411764.3445351

論文URL

https://doi.org/10.1145/3411764.3445351

動画

会議: CHI 2021

The ACM CHI Conference on Human Factors in Computing Systems (https://chi2021.acm.org/)

セッション: Human-AI, Automation, Vehicles & Drones / Trust & Explainability

[A] Paper Room 15, 2021-05-13 17:00:00~2021-05-13 19:00:00 / [B] Paper Room 15, 2021-05-14 01:00:00~2021-05-14 03:00:00 / [C] Paper Room 15, 2021-05-14 09:00:00~2021-05-14 11:00:00
Paper Room 15
12 件の発表
2021-05-13 17:00:00
2021-05-13 19:00:00
日本語まとめ
読み込み中…