A Method to Automatic Measuring Riding Comfort of Autonomous Vehicles: Based on Passenger Subjective Rating and Vehicle Parameters

Abstract

As a milestone product of the AI era, the autonomous vehicle has attracted tremendous attention from the whole society. When autonomous vehicles (AV) provide transportation services as passenger vehicles in the future, a comfortable riding experience will be the fundamental element of usability. In such a case, it is necessary to establish an objective and sound evaluation system to evaluate the comfort level of autonomous vehicles. We hereby develop the comfort level model of autonomous vehicles with the following three steps: (a) Explore subjective evaluation indicators: Invite passengers to test autonomous vehicles and collect their ratings of the comfort level; (b) Establish the subjective comfort evaluation model: classify the evaluation indicators, continuously collect the evaluation data of the comfort level from the passengers during the testing process, and then use the structural modelling method to form a subjective evaluation model of the comfort level; (c) Develop the automatic scoring tool: collect subjective and objective data through data collection apps, form a calculation function with machine learning algorithm that fits the subjective and objective data, and develop an automatic scoring tool based on it. This precisely developed evaluation system and the empirical data-based scoring tool can be used to guide technological development, optimize algorithms, and improve strategies within the AV corporate. On the other hand, it can help to unify evaluation standard for AV industry, improving the experience of autonomous vehicle rides.