In this master's thesis, a method for detecting and preventing collisions between a train and a tunnel is presented. The approach combines point cloud processing with parametric modelling of wagon motion. The tunnel is divided into horizontal layers, the walls are approximated using B-splines, and wagon motion is simulated along control points by determining the wagon's position based on its wheelbase. To detect collisions, six critical points are sampled at each height, and their distances to the tunnel geometry are calculated. An optimisation procedure determines the largest permissible wagon model. The system supports fitting various cargo shapes and provides interactive 3D visualisation. The main contributions are a collision detector that accounts for tunnel geometry, a procedure for deriving the largest safe wagon model, and two tools: one for cargo fitting and one for visualisation.
|
