Artificial Intelligence (AI) is Already at Work in the Forest Industry

by | Apr 25, 2024 | 2024, Logging & Sawmilling Journal, March/April 2024, New Technology, Sawmill

First, it is important to note that Artificial Intelligence (AI) systems working today in Canada’s wood product manufacturing facilities—to monitor and optimize lumber production—are not ‘AI Chatbot’, which is so much in the news these days.

In the forestry sector, human programmers are using high quality data supplied by sawmills to train their own proprietary AI systems to execute specific actions, recognize problems, provide optimal solutions, and implement solutions with minimal human intervention.

These systems are machine centre-specific, meaning that a sawmill could have one Comact-branded AI system from BID Group optimizing their primary breakdown, one system from USNR optimizing their edger, a system from Carbotech optimizing their trimmer, and another one from Autolog optimizing lumber grading.

But we aren’t too far removed from the fully interactive HAL 9000 computer that became famous in the 1968 science fiction movie, 2001 A Space Odyssey. What was science fiction is now science fact.

“This is just the beginning,” says Diego dos Santos, Applied Artificial Intelligence Manager at the BID Group. “I see the wood products manufacturing sector being completely transformed and autonomous over the next 10 years, running at its best efficiency with the least operators possible, deploying smart systems using AI across the entire production line and providing operators with data to facilitate their decision making.”

That sentiment over AI’s growing influence within the sawmilling sector is shared by Scott Norton, Senior Vice-President for Automation Solutions at equipment manufacturer and installer, USNR.

“We are still in the process of determining where and when to apply an AI solution,” he says, “so I expect that 10 years from now, we will see AI-based solutions that no one imagined today.”

For those companies still not engaged or on the fence about installing AI-based systems, there soon may be no alternative for the highest quality equipment.

“I think in 10 years from now, a big portion of the sawmills, if not all of them, will be equipped with AI-powered systems if they want to stay in business and keep up with the new technologies,” says Eric Michaud, Vice-President of Sales and Marketing at the Carbotech Group, which also owns Autolog.

“Most of the manufacturers will use it on a permanent basis on their optimization systems and eventually on their PLC or automation program,” he adds.

AI systems currently working in the wood product manufacturing sector are not meant to replace workers, but to allow existing employees to work more efficiently. It can also help companies faced with the challenge of chronic labour shortages to plug a hard-to-fill employee position, as long as there is an AI system able to carry out that same task.

In some cases, it could literally save some sawmills from closing in more remote areas, where workers are particularly hard to find. And it could also help to attract young, more tech-savvy employees to the forestry sector.

In today’s increasingly automated world, the introduction of AI-enabled systems, including robotics, primarily aims to enhance precision, productivity, and safety, particularly in labour-intensive environments. While these innovations fill roles often deemed less satisfying due to their repetitive natures, the transition may lead to job displacement in rare cases where current employees find it challenging to adapt to new skills and tasks.

Since becoming part of the product offerings of several industry suppliers, there is no doubt that AI systems have delivered results that are transforming the lumber industry. That’s because properly trained AI-enabled decision-making systems have proven to be more accurate, reliable, significantly faster and more dependable than human beings in such functions as log management, lumber production, dimension and species identification and sorting, waste reduction, kiln drying, grading and planing.

“In the context of production lines going faster than ever, it would be impossible for humans to see and analyze all the defects that a grading system with vision and AI can do,” says Michaud. “They are more precise and more consistent.”

This is just the tip of the iceberg. It can also create a safer work environment, where a camera and an AI system can monitor production in dangerous environments, detect problems, and stop production or solve those problems without human interaction, for example, if a log or board is out of position, and even dangerously out of position, on a sorter, merchandizer or trimmer line.

It is not too far-fetched to think about one AI system capable of operating an entire sawmill with minimal human interaction. In fact, the BID Group is working on a system right now that can analyze the health of an entire sawmill, identify bottlenecks, and recommend changes in a fraction of the amount of time compared to current human-based, quality control methods. It also offers an AI system that can analyze data from two machine centres that operate AI programs from two different manufacturers, for customers to plan and problem solve specific issues related to its manufacturing process.

USNR is also working in the same space.

“All of our lumber grading products now include DNN (Deep Neural Networks) for defect classification—well over 100 systems,” Norton says. “These tools are superior to traditional algorithm-based solutions. We are now investigating other AI technologies to evaluate mill-wide data trends and solve other problems.”

What everyone agrees upon is that the whole AI landscape insofar as applications within the sawmilling sector is evolving.

“In some areas, our industry is quite advanced in the usage of AI technologies,” says Norton, “but in others, we are just getting started. Our industry is quite advanced at using AI for lumber grading, but there are a lot of advances to come in data mining to improve operational flow.”

“AI is getting powerful so rapidly, the potential of it is extraordinary and scary at the same time because of its insane quick progression,” says Michaud. “AI is becoming so powerful, we don’t see the end of this ascension—so who knows where we’re going with that?”

There are many specific AI system applications already available. Here are some examples:

• Improving the debarking process through wood species classification to tailor debarker settings and employing residual bark detection for closed-loop feedback adjustments;

• Better products sorting for enhanced drying and grading process through wood species classification of boards;

• Optimal auto grading process through natural and manufacturing wood defects detection and classification;

• Enhancing quality control by identifying the cutting tool that has sawn the boards when there needs to be an adjustment because boards are off-size;

• Better process monitoring to identify production anomalies and issues that could generate long downtime periods, break mechanical equipment, or possibly injure an employee;

• And, management of robotic functions.

So, AI in lumber production is here and will only become more prevalent and advanced. For some companies, the question is where to start with the minimal amount of risk and smallest capital investment to learn more about it and evaluate its performance.

There are several companies today that can provide a one-stop solution to install these types of systems—all sawmill owners need to do is provide the financing. According to dos Santos, it’s a common misconception that transitioning to AI-enabled systems is smoother and more cost-effective for already modernized sawmills with existing optimizations. Even less modern sawmills can successfully integrate AI technologies, potentially leveraging unique approaches and customizations to bridge technology gaps.

One of the least expensive systems for a company to get a feel for AI is a monitoring system, he adds, for use on one part of the production line, like the wave feeder, to monitor the log gap leading into the primary breakdown unit in an effort to maximize sawing productivity, because it often only requires one computer and one camera. Should a client appreciate and understand the results, they can then focus on bottlenecks in their manufacturing process and tackle one challenge using AI-enabled systems, evaluating training requirements and the return on investment (ROI).

A common concern is how quickly current AI systems could become obsolete. It’s important to understand that customers have the option to retain their existing version, which continues to perform effectively, while upgrades are available for those seeking enhanced capabilities. The biggest cost will be the initial hardware installation.

“The first client that bought AI from us in 2018, they still haven’t had to update any computers,” says dos Santos. “The decision to upgrade depends on whether the system meets your needs or if you identify areas for enhancement. In most cases, we simply update the software. Depending on the computer’s capabilities, we might add more power, such as extra memory or GPU cards to expand functionality. Our goal at BID Group is always to fully leverage and enhance the current hardware before considering more significant upgrades.”

Norton says—and others agree—that in the near term, equipment manufacturers will focus on AI-based purpose-built solutions once a company identifies a bottleneck, emphasizing that AI-based solutions are only the latest of many tools that companies like USNR can bring to the table. At present, it still hasn’t become an all-or-nothing proposition for companies.

“The traditional approach to capital expenditure is still a good approach—using vendors like USNR to help identify areas for improvement and the ROI that these technologies can achieve,” he says. “At USNR, we see AI as the newest, but still one of many tools that we can apply to a problem.”

 

Computer-based lumber grading systems delivered the first taste of what AI-enabled systems could achieve

Artificial intelligence-enabled monitoring and optimization systems, currently sweeping the Canadian forest sector, have become more prevalent over the past decade.

The first hint that AI was knocking at the door of the forest industry was when computer-based lumber grading systems using machine learning entered the market from companies like Comact and Autolog around 2006.

Today, AI systems have advanced beyond machine learning to become deep neuro networks. AI deep learning-powered products started to become available through companies like the BID Group and USNR in a big way around 2018. Today, the BID Group has about 20 separate and specific optimizing and monitoring AI system applications from the mill infeed to the outfeed, and capable of performing specific tasks according to a client’s needs. Across a multitude of BID Group systems in the market, over 250 are currently operational with AI-enhanced features.

“With a deep neuro network, a computer will use data (inputted initially by a programmer for training) to learn to perform a specific task, and if there are errors and the AI system is alerted to those errors with better data, the AI system will find ways to fix the errors on its own. It will develop its own algorithm, and the user only needs to provide good data to make the corrections,” says Diego dos Santos, Applied Artificial Intelligence Manager at the BID Group.

The critical factor for these systems to work properly right from day one is for wood product manufacturing facilities to provide very high quality data to train the AI system as well as for correcting errors should they crop up down the road, dos Santos adds. Since the supplier incorporates AI into their equipment and proprietary programs, there’s no need for customers to have internal AI expertise or knowledge. The idea is to use customer’s data to tailor the AI application to the customer’s needs, as there is no one-size-fits-all in wood processing.

“As the old saying goes, ‘garbage in, garbage out,’” says dos Santos. “The system’s accuracy depends on the quality of the data provided, which is supervised by humans. Once a system is consistent and providing good results, it will not try to continue learning and develop bad habits by itself. We tell it to continue learning only with new data that we provide.”

USNR has also been a pioneer developing AI-based systems.

“USNR has been installing AI systems for about 10 years,” says Scott Norton, Senior Vice-President for Automation Solutions at USNR. “Virtually every sawmill or planer mill in North America with automated lumber grading or defect recognition has AI installed today. In 10 years, it has changed from almost no one providing an AI-based solution to everyone supplying an AI-based solution. Our customers have benefitted from higher accuracy.”

Tony Kryzanowski

Author

0 Comments

Submit a Comment

Your email address will not be published. Required fields are marked *

Clearing for Canada’s Energy Future

Clearing for Canada’s Energy Future

B.C.’s DENT Construction has been involved with some of the biggest construction jobs in B.C. in recent years, including clearing work for one of the largest construction projects in Canadian history—the Trans Mountain Pipeline expansion project.

Upgraded Technology for Truro

Upgraded Technology for Truro

An upgrade to the J.D. Irving sawmill in Truro, Nova Scotia is delivering increased recovery thanks to enhanced scanning and positioning technology.

The Edge: Partial harvest systems to meet specific management objectives within boreal mixedwood and pure softwood stands

The Edge: Partial harvest systems to meet specific management objectives within boreal mixedwood and pure softwood stands

The Canadian Wood Fibre Centre (CWFC) technical development group of the Canadian Forest Service (CFS) and their partners have a history of developing innovative partial harvest systems that address issues of forest managers in the Prairies. Here are a few options available to industry to address the current and anticipated consequences of a changing climate, as well as to help maintain a sustainable fibre supply as Canada continues to transition to managed forests.