Exploring Ethical Decision-Making and Acceptance in Autonomous Driving

Problem

One of the most critical challenges in the deployment of autonomous vehicles (AVs) lies in the ethical and cognitive dimensions of decision-making—particularly in high-stakes traffic situations. Project EEDA (Ethical and Experiential Decision-making in Autonomous Driving) investigated how humans perceive, trust, and respond to the behavior of self-driving cars during such critical moments. Beyond evaluating participants’ attitudes toward an AV’s moral or safety-driven decisions, the study also compared their responses when they were required to take over control or drive manually under identical conditions.

---

Approach

This project expanded upon the previous Westdrive LoopAR framework to incorporate multiple levels of driving autonomy:

• Manual driving, where participants maintained full control
• Semi-autonomous driving, requiring occasional user intervention
• Fully autonomous driving, with complete vehicle control
• Taxi-riding mode, where participants were passive passengers

The VR-based driving simulator enabled realistic interactions in a controlled yet immersive environment, integrating eye tracking, steering dynamics, and real-time vehicle telemetry. Participants experienced a series of traffic events designed to probe their ethical intuitions, situational awareness, and trust dynamics, followed by an in-depth Autonomous Vehicle Acceptance Model (AVAM) questionnaire assessing acceptance, perceived safety, trust, and anxiety.

A mixed-methods approach was used:

• Objective measures: gaze and fixation patterns, attention metrics, and reaction times
• Subjective measures: attitudes, trust levels, and emotional responses captured via questionnaires

---

Results

The findings revealed a nuanced interplay between user control, trust, and anxiety across autonomy levels:

• Semi-autonomous conditions increased trust and intention to use AVs due to user involvement in decision-making.
• However, these conditions also elicited higher anxiety, reflecting tension between engagement and comfort.
• Fully autonomous driving improved perceived safety and reduced anxiety but did not significantly raise long-term adoption intent.
• Eye-tracking data indicated that engagement—especially gaze focus and vigilance—correlated strongly with trust and perceived control.

Together, these results underline the cognitive trade-offs between user control and automation, suggesting that optimal AV design must carefully balance transparency, user participation, and comfort to foster sustainable trust and acceptance.

---

My Role

• Configured and monitored the VR and hardware setup, ensuring accurate sensor calibration and system reliability
• Provided technical expertise for experimental design and data synchronization across driving modes
• Assisted with data processing and analysis, focusing on gaze and engagement metrics

---

References

• Keshava, A., Nosrat Nezami, F., Schüler, T., König, P. (2024). Trust, Anxiety, and Engagement in Semi-Autonomous Driving: Balancing Interaction and Automation in Virtual Reality. Frontiers in Human Neuroscience, 18, 1048572.
• Nosrat Nezami, F., et al. (2023). Westdrive LoopAR: A Virtual Reality Platform for Ethical and Cognitive Studies in Autonomous Driving. IEEE Transactions on Human-Machine Systems, 53(4), 456–469.

---

Keywords: Autonomous vehicles, ethical decision-making, human–automation interaction, virtual reality, trust, user acceptance, active inference, cognitive modeling