AMS Solutions developed a machine vision inspection system to detect wood separator sticks that may jam lumber-stacking equipment at sawmills. The vision system uses Cognex In-Sight vision tools for quality control on the lumber processing line and Cognex vision software to display and review the inspection results. The system ensures uninterrupted operation of mechanical lumber stackers.
After lumber products are cut to size in a sawmill, they are stacked into packages before being sent to a kiln drying process that heats the wood to a specified temperature in order to remove moisture and eliminate any insect infestation. The lumber is stacked in tiers and each tier is separated by small boards called sticks, a specialty wood product used to maintain spacing during the kiln drying process. Depending on the specific requirements of the mill, sticks come in many sizes but are typically 4 to 5 feet (122-152cm) long, 1.5 to 2.5 inches (3.8-6.4cm) wide and 0.75 inches (1.9cm) thick. The sticks are continually cycled through the sawmill, kiln and planermill and can be reused many times before they need to be replaced.
Automatic stick placing
Before automatic stick placing systems were developed, people would manually place the sticks between every tier as the package was stacked and dispose of broken sticks as required. With automatic stick placing systems, the sticks are loaded into hoppers located along the length of the packages and are automatically dropped at the required locations. These hoppers are automatically refilled using an overhead pusher system with diverters that route sticks into the hoppers. To feed the overhead pusher system, the sticks are first singulated by an unscrambler system and then transported by a lugged chain which maintains each stick in its own lug space until it is handed off to the overhead pusher chain. While being transported in the lug chain section, the sticks are inspected and rejected if required.
Automated stick placing machines used in mechanical lumber stacking operations require detection of sticks that may jam the stacking machine. The system must reliably reject sticks that are broken or are not the correct size to fit in the machine in order to maintain continuous operation. There are a number of methods currently used to detect these defective sticks. Some older systems rely on manual inspection to reject sticks before they enter the machine. Manual systems are costly, pose a safety concern and are not nearly as reliable as automated systems. Other systems use a series of “photo-eyes” to automatically measure stick characteristics at a few locations. These are not reliable defect detection systems because they gather limited information.
Profile scanning option
One proposed solution to the problem of detecting defective sticks utilizes lineal scanning technology. In a typical configuration, multiple scanners would be positioned around the sticks on the lugged conveyor. Each scanner would capture a 3D image of a portion of the stick and the multiple images would then be combined in a computer to provide a complete 3D profile of the stick. The cost of this approach is easily justified in log and board optimizer applications but isn’t cost effective for stick inspection. As a result of the high cost, most sawmills chose alternate less effective stick inspection methods.
Vision system selection
“Generally, today’s vision systems are divided into two groups: PC-based and self-contained smart camera vision systems,” said Richard Vetter, Vice-President of AMS Solutions. “We decided to go with a vision system because they are generally easier to configure, validate, and maintain.” Because they are solid state, smart camera vision systems intrinsically provide a more stable platform than Microsoft® Windows®-based alternatives, requiring less maintenance. Being configurable rather than programmable, smart camera vision systems also make it easier to accommodate future changes.
With this plan in place, AMS Solutions developed the IdealSTICK machine vision inspection system to cost-effectively evaluate the geometric defects of the entire stick at rates up to 300 LPM. The IdealSTICK system has been successfully deployed at multiple locations.
“We selected the Cognex In-Sight vision system because it provides 1600 by 1200 pixel resolution which is sufficient to inspect the entire top surface of the stick to approximately 0.050” (1.3mm) accuracy and because its high speed processor can easily complete the inspection operation within the time requirement,” Vetter said. “We also like the ruggedness including its rugged die-cast aluminum case, sealed M12 connectors, and protective lens cover that provides IP67 and IP68 rated protection against dust and moisture.”
Vision system operation
Utilizing a series of lasers and plan view image analysis on the Cognex In-Sight platform, IdealSTICK is able to evaluate sticks for thickness, width, length, bow, crook, twist, breakage and overall shape. The IdealSTICK system is calibrated by simply capturing an image of a stick with the desired dimensions. The operator stops the conveyor, places this “ideal” stick in position to be viewed by the vision system and pushes the calibrate button. From this point on, the vision system compares each stick to the one used for calibration and rejects sticks whose dimensions vary too much from the ideal stick. The vision system program was developed using the Cognex In-Sight Explorer Spreadsheet interface. This interface makes it possible to program a vision application by dragging and dropping vision tools into a matrix without ever writing a line of code.
A photo-eye detects when a stick is in position below the camera and triggers the camera to take the picture. AMS programmers used vision tools available in the Cognex programming software to analyze the stick and generate a Pass/Fail result. The result of the analysis including a picture of the inspected stick is displayed in real-time using an industrial touchscreen PC mounted on the scanner frame and running the Cognex vision software. The software was used to create a dynamic real-time user interface that can be used not only to monitor the current piece being analyzed but also to review the previously rejected sticks. The review mechanism allows the operator to easily see why the system rejected previously inspected sticks.
“The first IdealSTICK systems are performing very well,” Vetter concluded. “The customers are very happy with the inspection systems and our expectation is that an increasing number of sawmills will recognize the advantages of machine vision-based stick inspection including the relatively low cost, high accuracy and the ability to function reliably in the sawmill environment.”