Process optimization and control of a patching plant for shuttering panels

The goal of this work is the automation and optimization of the patching process of wood defects for shuttering panels. In this patching process, the raw panels go through an optical defect scanner that determines position and shape of the wood defects. Then, the panels are processed at a patching robot. To maximize the throughput of the plant, several research tasks in the field of process optimization and control have to be performed. The process optimization begins with the computation of the patch arrangement, such that each wood defect is covered by the minimum number of unisize, circular patches. The proposed patch placement algorithm is based on the concept of Hexagonally Closest Packaging. The computed patch locations have to be processed in the time optimal sequence. This problem is similar to the well-known travelling salesman problem. Two solution strategies are proposed, a classical Ant Colony Algorithm and a Local Search Receding Horizon Algorithm. To conclude the optimization process, the robot motion between each consecutive pair of patch locations needs to be executed in a time optimal way. To this end, a real-time capable Bang-Bang Trajectory Generator accounting for arbitrary initial velocity and acceleration of the patching robot is proposed. Next to process optimization, real-time control of the patching robot is discussed. The main challenge is the development of a control strategy that is able to position the panel in a fast and precise manner despite undefined relative motion between the panel and the patching robot. To overcome this issue a strategy for sensor data fusion called Trajectory Updating is proposed. The optimization algorithms as well as the control strategies are implemented and extensively tested at a prototype patching plant.