Incorporating Explanations into Human-Machine Interfaces for Trust and Situation Awareness in Autonomous Vehicles.

Autonomous vehicles often make complex decisions via machine learning-basedpredictive models applied to collected sensor data. While this combination ofmethods provides a foundation for real-time actions, self-driving behaviorprimarily remains opaque to end users. In this sense, explainability ofreal-time decisions is a crucial and natural requirement for building trust inautonomous vehicles. Moreover, as autonomous vehicles still cause serioustraffic accidents for various reasons, timely conveyance of upcoming hazards toroad users can help improve scene understanding and prevent potential risks.Hence, there is also a need to supply autonomous vehicles with user-friendlyinterfaces for effective human-machine teaming. Motivated by this problem, westudy the role of explainable AI and human-machine interface jointly inbuilding trust in vehicle autonomy. We first present a broad context of theexplanatory human-machine systems with the "3W1H" (what, whom, when, how)approach. Based on these findings, we present a situation awareness frameworkfor calibrating users' trust in self-driving behavior. Finally, we perform anexperiment on our framework, conduct a user study on it, and validate theempirical findings with hypothesis testing.