Intelligent Co-Simulation Framework Combined with Real Test Fields for Collaborative Driving

The simulation test for autonomous driving has become the predominant method due to its cost-effectiveness, high efficiency, and safety. The simulated traffic flow environment plays a crucial role in verifying and evaluating autonomous driving systems, effectively revealing the challenges they are confronted with in complex traffic scenarios. However, the lack of fidelity in pure simulation tests may impact the confidence of the results. To address this issue, an autonomous driving test framework is proposed in this paper, which is based on the digital twin technology that is combined with the real test fields and the corresponding virtual digital twin. It utilizes multifunctional real-world test fields as the testing scene, integrates virtual traffic objects into the system, and maps real vehicles into virtual road environments during testing. This framework offers greater flexibility in scene configuration, extensive coverage, and enhanced safety during testing processes. Additionally, it constructs human-driven vehicles (HDVs) and connected and autonomous vehicles (CAVs) models with different driving styles to simulate mixed traffic flows that align with Chinese drivers' behaviors on a large scale. Various traffic flow test scenarios are constructed to evaluate different autonomous driving systems' performance, achieve collaborative driving functions, and visualize the process, validating the effectiveness of the developed test framework.