When2Talk: When Should a Proactive In-Car Agent Talk?
When2Talk explores optimal communication timing for in-car agents using a context-sensitive policy to reduce interruptions.
Key Findings
Methodology
The study used a mixed-methods approach with 41 participants experiencing two communication policies in a VR-simulated fully-automated vehicle: Event-Triggered (ET) and Context-Sensitive (CS). The CS policy chooses Immediate, Delayed, or Silent communication based on event priority and passenger activity.
Key Results
- CS policy significantly reduced perceived interruptions and increased communication appropriateness without affecting trust levels.
- Experiments conducted using CARLA and Varjo XR-4 VR headset provided immersive experiences across different passenger activities.
- Key considerations for selective communication include event consequence, passenger activity, and continuing information value.
Significance
This study offers a new perspective on communication strategies for in-car agents, emphasizing the importance of timely communication in autonomous driving to reduce unnecessary interruptions and enhance passenger experience.
Technical Contribution
Introduces a context-sensitive communication strategy, distinct from traditional event-triggered methods, providing a framework for communication decisions across varying passenger activities and event priorities.
Novelty
First to apply context-sensitive policy in in-car agent communication, integrating event priority and passenger activity for flexible communication choices.
Limitations
- The study was conducted in a VR environment, which may differ from real-world driving experiences.
- Communication needs in multi-passenger scenarios were not considered.
Future Work
Future research could explore communication strategies in multi-passenger scenarios and real-world driving environments.
AI Executive Summary
Advancements in autonomous driving technology have made passenger experience a crucial design consideration. Traditional in-car agents typically communicate immediately at every event, which can lead to unnecessary interruptions. The When2Talk study proposes a context-sensitive communication strategy that combines event priority and passenger activity to choose Immediate, Delayed, or Silent communication.
Through experiments conducted in a VR simulation environment, the study found that the context-sensitive strategy significantly reduces perceived interruptions while maintaining communication appropriateness. Participants experienced both strategies across different passenger activities, demonstrating the potential of the context-sensitive strategy to enhance passenger experience.
This study provides new directions for future in-car agent design, emphasizing the importance of timely communication in autonomous driving environments. Future research could further explore communication strategies in multi-passenger scenarios and real-world driving applications.
Deep Analysis
Background
With the development of autonomous driving technology, passenger experience has become a critical design consideration. Previous research has shown that in-car communication can influence passengers' understanding and trust in autonomous driving. However, overly frequent communication can lead to interruptions, affecting passenger experience.
Core Problem
Traditional in-car agents communicate immediately at every event, potentially causing unnecessary interruptions. Determining the appropriate communication timing under different event priorities and passenger activities is a challenge.
Innovation
Proposes a context-sensitive communication strategy that combines event priority and passenger activity to provide Immediate, Delayed, or Silent communication choices. This strategy reduces interruptions and enhances communication appropriateness.
Methodology
- �� Conducted experiments in a VR simulation environment
- �� Used CARLA and Varjo XR-4 VR headset
- �� 41 participants experienced two communication strategies
- �� Recorded participant feedback and eye-tracking data
Experiments
Experiments were conducted in a VR environment using CARLA to simulate autonomous driving scenarios. Participants experienced two communication strategies across different passenger activities, with feedback and eye-tracking data recorded.
Results
The context-sensitive strategy significantly reduced perceived interruptions and increased communication appropriateness. Participants experienced both strategies across different activities, demonstrating the potential of the context-sensitive strategy to enhance passenger experience.
Applications
This strategy can be used in the design of in-car agents for autonomous vehicles, reducing unnecessary communication interruptions and enhancing passenger experience.
Limitations & Outlook
The study was conducted in a VR environment, which may differ from real-world driving experiences. Communication needs in multi-passenger scenarios were not considered.
Plain Language Accessible to non-experts
Imagine you're in an autonomous car with an assistant that tells you what the car is doing. Previously, this assistant would tell you immediately at every event, which could sometimes interrupt you. This study proposes a new method where the assistant decides when to tell you based on the importance of the event and what you're doing. For example, if you're on your phone, the assistant might wait until you're free to inform you.
ELI14 Explained like you're 14
Imagine you're playing an autonomous driving game, and there's an assistant in the car that tells you what the car is doing. Before, this assistant would tell you immediately at every event, which could sometimes interrupt you. This study proposes a new method where the assistant decides when to tell you based on the importance of the event and what you're doing. For example, if you're on your phone, the assistant might wait until you're free to inform you.
Glossary
Event-Triggered Policy
A strategy that communicates immediately at every event occurrence.
Used as a baseline for comparison with the context-sensitive policy in the experiment.
Context-Sensitive Policy
A strategy that chooses communication timing based on event priority and passenger activity.
Core method of the study, used to reduce communication interruptions.
CARLA
An open-source platform for autonomous driving simulation.
Used to create the VR simulation environment.
Varjo XR-4
A high-end VR headset device.
Used for immersive experiences in the experiment.
Autonomous Driving
Vehicles capable of driving autonomously without human intervention.
The background field of the study.
Open Questions Unanswered questions from this research
- 1 How to effectively apply context-sensitive strategies in multi-passenger scenarios remains to be explored.
- 2 The effectiveness in real-world driving environments has not been verified.
Applications
Immediate Applications
In-Car Agent Design
Can be used to design smarter in-car agents, reducing unnecessary communication interruptions and enhancing passenger experience.
Long-term Vision
Intelligent Transportation Systems
This strategy could be used in future intelligent transportation systems to improve communication efficiency and passenger satisfaction in autonomous vehicles.
Abstract
Proactive in-cabin agents can help passengers understand automated-vehicle (AV) behavior, but communicating every ride event may introduce unnecessary interruptions. We investigated how communication should adapt to event priority and passenger activity. In a mixed-methods within-subject study, 41 participants rode as passenger in a VR simulated fully-automated vehicle. We compared an event-triggered (ET) policy that communicated immediately at every event with a context-sensitive (CS) policy that selected \textit{Immediate}, \textit{Delayed}, or \textit{Silent} communications. CS increased communication appropriateness and substantially reduced perceived interruption. Perceived trust did not differ between policies, although baselines dispositional trust differentiated communication preferences. Findings highlight event consequence, passenger activity, continuing information value, and confirmation need as key considerations for selective in-cabin communication.