Trajectory analysis can be useful to understand pedestrian behaviour in indoor environments. However, collecting real data can be difficult and time consuming, and many publicly available data sets and trajectory generation methods focus on outdoor scenarios. We propose a trajectory generation method which takes into account the various restrictions of indoor environments, while only requiring a set of points  of interest and a floor plan picture as input. After computing the Delaunay triangulation of the points, we produce a base trajectory by simulating a walk through the resulting graph. Then the base trajectory is enhanced via crowd simulation rules, increasing the realism of the produced data.