The master's thesis addresses the development of a robotic system for automated scoring of bread dough on a moving conveyor belt in an industrial bakery environment. The objective of the thesis was to develop a solution that enables precise and time-efficient scoring despite the variable positions of the products and the continuous motion of the conveyor. Special emphasis was placed on the synchronization of multiple robots without the use of conventional reflective sensors, as well as on achieving the required cycle time. The work addresses challenges in the modern food industry, where automation contributes to increased productivity, consistent quality, and reduced dependence on manual labor.
The developed scoring system includes data acquisition using a laser scanner, point cloud processing, localization of individual dough pieces, and cut planning. The point clouds are segmented into individual loaves based on the known structure of the production process. For each dough piece, a geometric analysis is performed using the PCA method, which is used to determine the orientation and bounding volume of the loaf. Based on this, the positions of the cuts are calculated, adapted to both the shape and the local height of the dough. An important part of the methodology is the algorithm for grouping cuts into time-synchronized rows, enabling efficient and safe execution of scoring with multiple robots. Virtual cuts are used for synchronization, ensuring an equal number of operations per robot and consequently preventing potential collisions.
The results show that the system reliably detects loaves, correctly plans cuts, and successfully organizes them even in the presence of irregularities such as offsets, missing or merged dough pieces. The average processing time per loaf is approximately 37 ms. An important time-related factor is the execution of cuts by the robot, where three scenarios with different scoring durations (0.18 s to 0.73 s per cut) were tested. System analysis demonstrated that, with appropriate configuration and trajectory planning, the required time performance for industrial application can be achieved. The cuts were executed precisely, with uniform depth adapted to the local geometry of the dough.
Automated scoring of bread dough using robotic systems is therefore feasible and effective. The developed solution allows flexible adaptation to different product types and system configurations (number of scanners and robots). Despite the achieved results, potential for further improvements remains, particularly in communication, trajectory optimization, and error handling.
|
