Complexity in modern vehicles consists of an increasingly large multitude of components that operate together. While functional verification of individual components is important, it is also important to test systems of interacting components within a driving environment, both from a functional perspective and from a driver perspective. One proven way for testing is vehicle simulators and in the reported work the main goals have been to increase flexibility and scalability by introducing a distributed driving simulator platform. As an example, consider a workflow where a developer can easily go from a desktop simulation to an intermediate driving simulator and further to a high-fidelity driving simulator with Hardware-In-the-Loop systems close to a complete vehicle. To support this, a distributed simulation architecture was designed and implemented, based on user needs defined in a previous project which divides a driving simulator environment into four major entities with well-defined interfaces. These entities are Session Control, Environment Simulator, Driving Simulator, and Vehicle simulator. High Level Architecture (HLA) Evolved, an IEEE (Institute of Electrical and Electronics Engineers) standard, was chosen as the standard for communication. HLA Evolved is based on a publish-subscribe architecture, and is commonly used for distributed simulations. The entities and the communication topology are described in detail in the report. The evaluation of the distributed simulation architecture focused on flexibility and scalability, and on timing performance. Results show that the implemented distributed simulation architecture increased flexibility and scalability, compared to the non-modified architecture, as several distributed set-ups were tested successfully. However, distributed simulation has an inherent communication latency due to packaging and sending of data between entities. The communication latency was estimated to be one millisecond. This latency effect needs to be considered for a distributed simulation, especially if the communication between the Driving Simulator and the Vehicle Simulator is sensitive to such delays. During evaluations of the distributed simulation architecture, the Driving Simulator and the Vehicle Simulator were always located at one site in a low latency configuration. Thus, further investigations of the resulting communication delays between a Driving Simulator and a Vehicle Simulator at different physical locations are necessary. Two demonstrations, open for anyone with an interest in the distributed simulation architecture, were performed, one in Gothenburg and one in Linköping. During the demonstrations, the distributed simulation architecture was shown where participants could freely test a distributed simulation set-up including the VCC Hardware-In-The-Loop simulator and three of VTI’s moving. (Author/publisher)
Abstract