This work focuses on automating and optimizing the patching process for wood defects in shuttering panels. Initially, raw panels are scanned optically to identify the position and shape of defects, followed by processing with a patching robot. To enhance plant throughput, several research tasks in process optimization and control are undertaken. The optimization starts with calculating the patch arrangement to cover each defect with the fewest unisize, circular patches, utilizing a Hexagonally Closest Packing algorithm. The sequence for processing these patch locations is determined in a time-optimal manner, akin to the traveling salesman problem. Two solution strategies are introduced: a classical Ant Colony Algorithm and a Local Search Receding Horizon Algorithm. Furthermore, the robot's motion between patch locations is optimized using a real-time Bang-Bang Trajectory Generator, accommodating varying initial velocities and accelerations. Real-time control of the patching robot is also addressed, focusing on developing a strategy that ensures fast and precise panel positioning despite undefined relative motion. A sensor data fusion method called Trajectory Updating is proposed to tackle this challenge. The optimization algorithms and control strategies have been implemented and rigorously tested at a prototype patching plant.
