Today, behaviour models are incorporated to perform a sequence of tasks, involving motion planning, object recognition, localisation, manipulation and collaboration, and potentially even learning. Testing models in isolation is insufficient; entire missions that integrate several models should be thoroughly validated before deployment. This paper provides two new contributions: first, to profit from the development of the simulator as an extension to Model-View-Controller (MVC) where the view can be optionally incorporated. Second, to use the simulator with the stripped-GUI to massively scale up the testing the integration of models of behaviour that fulfil missions performed under the paradigm of continuous integration. We explore the challenging aspects of this testing context and illustrate it with a case study where software models of behaviour are parameterized.