The transition to software-defined vehicles (SDVs) represents a development paradigm in which vehicle functions are continuously expanded and modified through software updates, and verification activities are therefore moving toward automated regression verification frameworks. This talk interprets an SDV as a distributed real-time software system, and, from the perspective of SDV prototyping, organizes a ROS2-based simulation-in-the-loop (SIL) approach to propose a testing pipeline that links simulation with a low-risk physical platform. ROS2 enables easy composition and replacement of perception-planning-control functions, and its DDS-based communication and QoS policies allow reliability and latency constraints to be specified explicitly. From an implementation standpoint, Jetson AI proposes a flow that executes scenarios in a gazebo-based SIL environment, collects logs and derives quantitative metrics, and then carries the same ROS2 interfaces over to a low-risk physical platform to assess reproducibility of the results.