AUTOMATIC PHYSICAL AND AESTHETIC PATCHING OF WOOD-BASED PANELS SAVES WOODSTOCK RESSOURCES

Joerg Eberhardt, R. Massen, Marc Oliver Kuehn

Baumer Inspection GmbH, Konstanz, Germany www.baumerinspection.com

Summary
Wood-based panels with a natural wood top layer such as veneered panels, plywood furniture or multilayer mixed-material panels must comply not only to stringent physical quality standards, but still more to very high aesthetic criteria, as the latter are decisive for the purchasing by the customer. The traditional way of manual visual inspection and of sorting out the quality panels with a too low visual quality suffers from serious drawbacks: too slow, too subjective and too costly in terms of loss of a precious natural material. We present a technology for the automatic inspection and the automatic patching of panels which combines camera-based detection of flaws, automatic patching by NC-controlled putty or dowel injectors and, as an outlook, automatic camouflage of the repaired defects by ink-jet based local post-decoration with matching colours and wood grain graphics.

INTRODUCTION

Natural wood surfaces always show up local defects such as holes due to bad knots, resin pockets, open joints etc., defects which cannot be tolerated in the final products: multilayer table-tops, parquet, solid wood boards, plywood, shuttering boards, gluelam, furniture etc.. Many companies producing panels with a natural wood top layer therefore use a manual repair strategy: up to 12 workers are placed on both sides of the conveyor belt to perform a visual inspection of every panel; they locate and visually grade every defect and then repair it manually using hand-hold drilling equipment and dowels or putty to fill up the cleared defect zone (see Fig. 1) Needless to say that the result of this purely physical repair strategy is far from being ideal: a) the required workforce is important and expensive, the working conditions strenuous and prone to generate skeleton and joint health problems b) there is an important waste of material when patching with fast drying and expensive two-component putty c) manual patching is only economical for a rather small number of defects, typ. < 10 defects per panel face d) the result of the manually patched panel is not documented; its final quality grade is set by the worker, introducing important subjectivity in the final grading and pricing of the patched panels e) even if the physical properties of a repaired defects may be acceptable (the surface is even and well closed), the patched defects remain highly visible (Fig. 2)

Fig. 1 The manual patching of wood-based panels requires a large number of workers, rather healthcritical working conditions and substantial repair time. A too large number of defects ( typically > 10 defects) swiftly leads to a panel not worth anymore to be patched

We report on a new technology for a fully automatic and fast patching of large multi-layer wood panels using advanced multi-sensorial image processing and sophisticated numerically controlled XYZ manipulators. This technology is able to patch by inserting specially shaped dowels or by injection of 1- and 2-component putty. Due to its degree of automation, patching can be economically be applied to panels with a large number of defects, allowing both a better use of our precious wood stock and the use of lower wood qualities for the production of nevertheless high quality panels. We finally discuss our research project to patch panels not only for physical defects, but to use advanced colour image processing and local ink-jet decoration to make the patched defects invisible to the critical eye of the customer, thus restoring an almost perfect first grade quality from lower grade wood.

Figure 2 The state-of-the-art patching by manually inserting dowels or by injecting putty into local defects performs a physical local reparation of a defect, but not an aesthetical one: the repairs are highly visible .

2. AUTOMATIC PATCHING

Baumer Inspection GmbH, Germany, the world leader in the automatic visual inspection of panels for the flooring and furniture industry, has taken a new generic approach to the task of repairing wood-based panels, a technology which is able to work with any type of natural material (wood, natural stone etc.). We have developed together with the Austrian manufacturer of numerically controlled production equipment FILL GmbH (www.fill.co.at) a highly modular set of imaging devices and numerically controlled XYZ manipulators to automate any type of panel repair technology (insertion of dowels, injection of 1- or 2component putty). The system performs in its current version a classical physical patching, i.e. it generates from a panel with an even very large number of defects a patched panel with a nicely closed surface. We call this technology generic because it is based on a highly modular architecture comprising different camera/illumination systems, image processing algorithms able to detect defects, repair strategies stored in a knowledge base and high-speed/high precision numerically controlled XYZ axes with tools for clearing defects and patching them, either via dowel insertion or by putty injection.

Figure 3

Layout of the automatic patching system

Every patched panel is fully documented for automatic grading and sorting into batches for different products (furniture, building etc.).This generic system is prepared for being updated with a further mode which will produce patches which match to the local aesthetics of the panel; it will then produce a final panel where the patches are hardly visible to the human eye (see chapter 4).

As pictured in Fig.3, the panels are first scanned with a multi-sensorial high-resolution multiple-camera scanner while moving on a conveyor belt. This advanced two-side scanner first optically defines a panel-oriented carthesian XYZ coordinate system with the [Xo,Yo, Zo] origin positioned at one panel corners; it then detects, grades and locates these defects within that panel-based coordinate system, computes a repair strategy using a knowledge base of repairing rules and then transmits these data to the XYZ working-tool manipulator for carrying out the repair steps. An arbitrary number of independently operating XYZ tool holders can be addressed simultaneously to distribute the repair tasks over multiple NC tools and thus to increase the patching speed. Typically we will use two manipulators in parallel to patch an entire panel in about 90 seconds.

Figure 4

The panels are scanned by the multi-sensorial ColourBrain scanner (right top) before entering the complex automatic patching plant with several NC patching tools operating in parallel (center)

As so often in machine vision technology, a task which at the first glance appears rather straightforward becomes more than complex when it comes up to operate reliably in a rough production environment, dealing with panels with an enormous variety of wood species, with high contaminations by dust, wood glue, severe panel non-flatness etc.. A defect to patch must be located with an absolute accuracy of 1 mm on a 5000 mm long panel, despite severe panel non-flatness. Our system takes care of the 3-dim. distortions and measures defect positions mapped to this non-even landscape including the void volumes to be filled with putty. It generates the relevant data set to drive the patching tool controller: positions, contours, type of defect, grade, patching strategy to achieve a low visibility of the patched areas etc.. The generated repair strategies are optimised in view of maximum repair speed with a minimum of NC tools. Many of the implemented methods are protected by patents of Baumer (both granted and pending).

4. PHYSICAL PATCHING AND AESTHETIC RECONSTRUCTION ?

Fig.2 clearly shows that even the best patching of a panel, be it by doweling or by injection of putty, is not able to reconstruct a local natural wood appearance. Patching workers often choose among a small number of different coloured putties to make the patched defect as poorly visible as possible. But even a carefully patched panel never reaches the visual quality grade (and price) of a defect-free panel. The same is true for defects patched with a dowel: they remain very visible because of the local mis-match of wood grain. This aesthetical down-grading is a serious economical loss which is today taken as unavoidable by the wood industry. It is of course also an ecological loss, as lower quality wood species with a large number of knots, resin pockets, cracks etc. cannot be used for highquality panels for the furniture, the flooring and the interiors industry. We have therefore put forward the question, if it would not be both economically and ecologically challenging to develop an automatic patching technology which is able to make the patched defects almost invisible by using a local aesthetic reconstruction of the original wood appearance /1/. By imaging the regions around the defects with well calibrated colour cameras and by using machine-learning principles to automatically learn the natural wood structure in the neighbourhood of a patch, we are able to re-synthesize the lost original wood grain and colours. With an equally carefully calibrated ink-jet printer head, we could then decorate the patch with this computed local dcor to make it almost invisible to the human eye. Fig. 5 shows (in a computer simulation) an early result of this research project which confirms the validity of our approach and which promises stunning results.

Figure 5 Simulation of the automatic physical and aesthetic patching of local defects in panels: the missing wood grain and wood colours are automatically reconstructed based on the pictural information around the defect (Baumer patents pending).

CONCLUSIONS

The fully automatic patching of wood-based panels is economically faisible even for woods with a large number of defects using a new sophisticated technology which combines advanced camera technology, pattern recognition, machine learning and NC tooling. Our concept of automatic local aesthetic reconstruction of the patched defects furthermore leads to a technology which makes the patched defects almost invisible to the human eye . This is a green technology as well, as it allows for a far better use of wood including the exploitation of fast growing species and woods with a large number of defects, still enabling the production of (patched) panels for demanding applications in the flooring, furniture and building industry.

REFERENCES
[1] MASSEN R et al. (2009): Automatic Visual Monitoring of the Physical and of the Aesthetic Quality of bare and decorated Panels in the Production Line. 10th European Panel Products Symposium, org. Bangor University, Wales,UK.

[2] Massen R (2010): Saving Resources by Advanced Vision-Based Automatic Patching of Wood-based Panels. 1rst Int. Conference on Processing Technologies for the Forest and Bio-based Products Industry. Octobre 2010, Campus Kuchl, Fachhochschule Salzburg, Austria