Simulation-based collision detection for CNC machining using sensorbased image recognition

In milling processes, collisions lead to cost-intensive machine damages and long-term maintenance downtimes. Collision tests are essential in CAD/CAM planning when it comes to small batch sizes e.g. for aerospace parts. Currently, experimental tests are carried out based on the nominal clamping situation and CAD data. Clamping errors, probing errors and particularly incorrect tool lengths are thereby not considered. For this reason, a concept for a sensor-based collision detection system is presented. The integration of an automatic image recognition in combination with a material removal simulation enables an accuracy of 96.87 % in collision detection for three defined reference workpieces.