中国科学院行为科学重点实验室知识库

In traditional traffic environments, human drivers can communicate with other drivers, cyclists, and pedestrians through gestures, facial expressions, etc., to convey their intentions. However, most of the current autonomous vehicles cannot effectively communicate with other road users with an autonomous driving model. The critical problem is that other road users need help understanding the intention of the autonomous vehicle. Thus, autonomous vehicles need to communicate with the outside world in addition to the ability to detect other road users and make relevant maneuvers to avoid potential conflicts. The external human-machine interface (eHMI) communicates with other road users outside the vehicle. It is believed to effectively resolve the conflict between fully autonomous vehicles and other road users. As pedestrians need to understand the intentions of the vehicles quickly and accurately, usability is of great importance for an eHMI. The current study sought to thoroughly analyze the existing literature on evaluation methods and establish a theory-driven evaluation system. We developed our evaluative framework based on the Situational Awareness model of pedestrians. In making a safe crossing decision, the pedestrians need to form a correct situation awareness in which they must form accurate perception, proper understanding, and timely prediction of vehicle behaviors [1–3]. Based on this model, a well-designed eHMI must facilitate the perception, understanding, and prediction processes. In this way, safety, perceptibility, intelligibility, and adaptability are required, and the four major usability domains have been theoretically established.

In traditional traffic environments, human drivers can communicate with other drivers, cyclists, and pedestrians through gestures, facial expressions, etc., to convey their intentions. However, most of the current autonomous vehicles cannot effectively communicate with other road users with an autonomous driving model. The critical problem is that other road users need help understanding the intention of the autonomous vehicle. Thus, autonomous vehicles need to communicate with the outside world in addition to the ability to detect other road users and make relevant maneuvers to avoid potential conflicts. The external human-machine interface (eHMI) communicates with other road users outside the vehicle. It is believed to effectively resolve the conflict between fully autonomous vehicles and other road users. As pedestrians need to understand the intentions of the vehicles quickly and accurately, usability is of great importance for an eHMI. The current study sought to thoroughly analyze the existing literature on evaluation methods and establish a theory-driven evaluation system. We developed our evaluative framework based on the Situational Awareness model of pedestrians. In making a safe crossing decision, the pedestrians need to form a correct situation awareness in which they must form accurate perception, proper understanding, and timely prediction of vehicle behaviors [1–3]. Based on this model, a well-designed eHMI must facilitate the perception, understanding, and prediction processes. In this way, safety, perceptibility, intelligibility, and adaptability are required, and the four major usability domains have been theoretically established.