Closed field testing of Connected and Automated Vehicles (CAV) is an essential
pillar for verifying the functionality and performance of CAV and promoting its
large-scale deployment. Recently, many closed test fields in the world have been
newly built or rebuilt for testing CAV such as M-City in USA and AstaZero in
Sweden. However, few construction methods, standards, and specifications of
closed test fields for testing CAV were reported with details. We propose the
construction practice of the Closed Connected and Automated Vehicles Test Field
(CAVTest) of Chang’an University in China. The CAVTest has a multilayer
architecture, from down to the top, which consists of an application scenario
layer, a perception and a physical communication layer, a network link layer,
and a management service layer, respectively. The modularization design of
CAVTest ensures the system compatibility with different functional equipments,
the accuracy of the test record, the scientific nature of data analysis, and the
supporting conditions for scenario reproduction and factor analysis. Besides, at
least 22 types of autonomous vehicle test scenarios have been built in CAVTest,
and a digital twin system has been established. Two case studies for testing CAV
are carried out, which are the AEB test for pedestrians at signalized
intersections and the vehicle-to-vehicle (V2V) communication test under static
and dynamic conditions, respectively. The results showed that CAVTest can fully
conduct closed field tests for the safety and communication performance of
CAV.