Dataset availability is a fundamental requirement for advancing Automated Driving Functions (ADFs) through ever more reliable machine learning systems. The huge costs related to the development of real-world datasets have spurred the utilization of simulations environments to collect data from virtual reality scenarios. Such synthetic data can be used to pre-train neural models, that may in turn be fine-tuned on a smaller amount of real-world data. We present the development of a lightweight simulation environment, specifically targeting data collection for Driver Maneuver Intention Recognition (DMIR) systems. We created a realistic highway driving scenario to capture Lane Change (LC) intentions. The simulation setup offers a driving experience through a comprehensive system that closely replicates real-world conditions, including a wide screen, steering wheel, and pedals for the vehicle control. Procedural traffic generation was implemented to maintain computational efficiency while ensuring realism. Preliminary tests with four participants demonstrated the simulator's effectiveness, yielding an average of six LCs per minute. User feedback highlighted areas for improvement, such as perceived road width and traffic realism, indicating potential enhancements like additional cameras for a broader field of view. Our findings underscore the viability of simulation environments in extensively and cost-effectively generating data usable to advance AD technologies.