Safety Virtual-Reality Evaluation as a Service for Intelligent Electronic AVs: An Edge-Cloud Consumer-Customized Approach

Abstract

Recently, with the rapid development of intelligent autonomous vehicles (AVs), the safety of AVs become a critical concern for public consumers. Traditional safety evaluations for AVs focus on discovering unsafe cases during testing, however, they still leave several challenges especially in providing a safeguard of AVs acceptable to public consumers. First, existing schemes are unable to provide flexible public safety services for AVs, especially for customized safety services based on the demands of various consumers and traffic scenarios. Second, existing schemes cannot evaluate all traffic scenarios during testing such as future unknown scenarios and the scenarios with the degraded performance of the AVs. To address the aforementioned challenges, we propose a consumer-customized safety evaluation as a service for AVs, CSEaaS, which is a virtual-reality approach to provide scenario-coverable intelligent evaluating for AVs based on edge-cloud cooperation. Specifically, we design customizable safety indicators for AVs to satisfy the safety demands of consumers and construct an intelligent evaluating scheme based on virtual reality to test unknown traffic scenarios under real AV performance during traveling. Simulation results demonstrate the feasibility and effectiveness of CSEaaS. To the best of our knowledge, this is the first work to propose a customized evaluation-as-a-service for AVs.