The advance in interactive/cooperative robotics and integration of robots into populated environments puts emphasis on robot-robot and robot-environment (incl. human) interactions. Sensing and sensory data interpretation play ever increasing role because sensing becomes the primary way of collecting information about the environment. In this application domain the environment is unstructured, which results in demanding data interpretation and control algorithms. Similarly, the control and decision-making algorithms inevitably become more complex in order to be able to cope with the unstructured and dynamic environment. Consequently a sophisticated evaluation/test environment is required, which provides full control of the circumstances, reproducibility and flexible mix of real and virtual components. The paper presents a multi-agent based distributed HIL simulation environment (MARS), which assures scalable real-time performance and introduces an extension with high fidelity sensor simulation functionality.