Simulation of a Scaled Assembly Process with Collaboration of a Robotic Arm and Monitoring through a Vision System for Quality Control

The research develops a scaled model of semi-automatic assembly using a cobot and vision system, evaluating quality, process representativeness, and advantages and limitation of this simulation. Recent developments include semi-automatic assembly with cobot and vision systems, enabling realtime anomaly detection, improving quality, traceability, and efficiency in industrial processes. A scale semi-automatic assembly model with cobot's vision demonstrate integration in a modern educational manufacturing environment, enhancing efficiency, precision, and applicability in real industrial processes.

Our protocol combines cobots and vision in a scale model, allowing practical educational evaluation of efficiency, precision, and consistency, surpassing the limitations of traditional simulation or manual practice. Our laboratory will focus on optimizing cobot human collaboration and developing a vision system with a neural network to address lighting issues, improving defect detection, and scalability to industrial processes. To begin, organize all the necessary components for assembly on the replenishment tray, aligning them according to the designated layout.

Enter the programming sequence into the interface. Then wait for the collaborative robot to initiate the assembly sequence by picking up the bottom part of the box and transferring it to the assembly point. Allow the robot to pick up the worm and position it into the designated slot within the assembly.

Then the robot picks up the worm gear and assembles it on top of the box. Once the robotic subassembly is completed, wait as the robotic arm transfers it to the manual assembly area for further processing by an operator. In the manual assembly area, have the operator pick up the subassembly and continue the build by following the designated assembly order.

Upon completion of the manual assembly, place the fully assembled part vertically on the tray, ensuring that the worm is oriented toward the back. Once secured, have the collaborative robot place the product near the sensor on the conveyor for camera inspection. For worm shape evaluation, after selecting the inspection tool, register a reference image.

Click on the reference image icon in the top right corner. Select register image and click execute to capture the image. For worm gear parameter configuration, select the pattern region option to adjust the detection area.

Choose the polygon shape, outline the perimeter of the part, and click OK to confirm. To detect colors, select the pattern region option to refine the area around the worm gear. Choose the circle shape, mark the worm gear's perimeter, and click OK to apply the changes.

Then, select the mask region option to exclude unwanted areas from analysis. Choose the rectangle shape, outline the red edge of the part, and click OK to confirm. Now, enable the software interface from the computer and activate the switch to run mode.

Then select the utility icon, click on the statistics option, and choose the preferred graph type, such as a trend graph or histogram, to support data-based quality analysis by the new process manager. The shape histogram showed a normal distribution centered slightly above the nominal value, indicating the process was under statistical control, although most parts were closer to the upper specification limit. The process capability indices revealed a strong alignment with the lower specification limit, but a much lower capacity near the upper limit, leading to a low overall process capability.

The control chart showed that initial readings were unstable due to measurement system adjustments, followed by a mid-stage outlier likely caused by a defective part, and ended with a stable trend within upper control limits. The color histogram revealed that measurements clustered near the tolerance limits, suggesting only marginal compliance with specifications, and the presence of two skewed distributions indicated process instability. Capability analysis for color showed the process was centered, as CPU and CPL values were similar, but high variability reduced overall capability to a CPK of 0.539.

The control chart for color showed extreme instability with wide variation and frequent control failures throughout the production cycle.