Presentation
Predict and Model Worker Trust for Automated Vehicles in Manufacturing Plants
SessionLBR1: Late-Breaking Results
DescriptionAutomated Guided Vehicles (AGVs) have been deployed in numerous industrial settings, streamlining the transportation of parts for manufacturing processes. However, despite their prevalence, these vehicles typically operate separately from human workers due to a lack of trust issue. Therefore, to reach their full potential, AGVs must be assimilated into manufacturing landscapes with humans, where they can sense and respond to workers' trust. This study examined workers’ trust and behavior when interacting with AGVs. We considered three within-subject factors: AGV deceleration rate (0.1 or 0.7 m/s²), AGV approaching direction (N, NE, E, SE, S, SW, W, and NW), and user's expected crossing path (diagonal across an intersection or straight across). Then, a human subject study was conducted to collect behavioral data and self-reported trust levels from 16 participants acting as workers in a virtual reality-based manufacturing plant. For data collection, a VR headset, omnidirectional treadmill, and hand controllers were used to track eye gaze, participant movements, and hand positions, respectively. The data was then processed and analyzed to identify key factors that affect workers' trust. The results showed that an increase in users' deviation from their normal walking speed and expected crossing path corresponds to lower trust levels. Furthermore, an Analysis of Variance indicated that AGV approaching direction and expected crossing path significantly impacted the worker's expectation of AGV behavior, which can be strongly correlated to trust changes. Finally, we built and evaluated several machine learning models to predict workers' trust. As a future direction, we aim to design trust-aware AGVs.
Event Type
Late Breaking Results
TimeTuesday, September 10th3:20pm - 3:30pm MST
LocationFLW Salon I