PROCESSWEST Magazine Online
News

Technology Profile: Implementing automation in Canada's forestry industry


July 27, 2015  


Print this page

A new generation of workers are ushering in an unexpected future for the forestry industry that has less to do with the robotics commonplace in automobile factories and more to do with advanced versions of harvesting machines such as feller-bunchers and harvesters.

FPInnovations forest operations research manager Jean-Francois Gingras says concerns about potential labour shortages and the realization that the industry must accommodate a new generation of workers with a different set of expectations are chiefly responsible for forcing the development of new technologies. “It’s important to point out that the technological changes that are being discussed within the industry will be incremental,” says Gingras, who is involved in putting together an FPInnovations “Robotics in the Forest” workshop in Montreal this spring.

“We won’t be launching the space shuttle, but rather, it will be gradual automation of repetitive functions where machines will self-learn as they work their way through their tasks.” Gingras notes advanced types of technology such as machines with capability for self-navigation or locating specific trees are still years away. Instead, he points to John Deere's Intelligent Boom Control (IBC) introduced to the market last year as n example of progressive technology that already exists.

"It’s what we call intuitive forwarder technology, and we think it’s the first step towards automation,” says Gingras A forwarder is a vehicle that picks up felled trees from the forest floor using a boom arm and hydraulic grapple, places the logs into a rack or “bunk” directly behind the operator’s cab and transports the logs to a roadside landing where they are loaded onto a truck for further transport to be processed. In older boom mechanisms, the operator controlled each of the individual boom joint movements independently.

John Deere’s optional IBC technology now allows the operator to directly control the movement of the boom tip. One joystick moves the boom tip in the horizontal plane, while a second guides it vertically, resulting in faster cycle times and more precise control. Added sensors actively dampen and protect boom structures for longer wear life. The significance of the development of IBC is successful operation of the boom is attainable on a more consistent basis regardless of the experience level of the operator, says Richard Lawler, director of engineering for John Deere's construction and forestry division in Dubuque, Iowa,

“By integrating position feedback into the equipment, the operator is more easily able to control the machine without necessarily being aware of the technology that is being utilized in the background,” says Lawler, who leads the development of all forest equipment, globally, for John Deere. “It’s similar to the way automakers have been able to reduce the variation in braking distances through technology. A driver who is forced to stop a vehicle equipped with ABS quickly is unaware the ABS system takes over when that happens. The driver is also probably not aware the ABS system is wildly complex with many components, including sensors and microcomputers, all working seamlessly behind the scenes, reducing the variations in operator induced output.

“The challenge for us in the forestry equipment sector is much the same. We’re focusing on technology that is somewhat invisible to the operator while reducing the variation in the operator-induced productivity.” Through the use of feedback systems, the recently launched IBC on forwarders reduces the variation in the operation of the boom regardless of the experience possessed by the operator, adds Lawler. Just like the automotive ABS system, the IBC is integrated into the vehicle so the operator isn’t necessarily aware of what is happening in the background. In studies of the IBC system, John Deere has found a 10 to 20 per cent improvement in productivity depending on various factors.

The feedback systems work similar to techniques used for automated welding or pick-and-place robotics where a robot arm is imbedded with sensors to determine its position in space. If the position is known, microprocessors can then determine velocity and derive many other levels of information. “The IBC system now available is the first time we’ve gone to full position feedback systems which gives us a lot of future opportunity for increased automation,” says Lawler.

“We’re focused on developing the highest value for the customer, to improve productivity by reducing the impact caused by operator skill levels, while also improving the durability of the equipment.” Systems have allowed John Deere to provide immediate improvement in that area. “This isn’t the first time we have used feedback systems in our equipment,” he says.

“For instance, we developed an auto shift system on skidders where engine speed, engine load and other feedback loops inform the machine of the optimum gear it should be in, resulting in increased productivity and reduced fuel consumption.” And because hand-eye co-ordination is different for everyone, John Deere has engineered techniques such its Rapid Cycle System to automate the dump feature on tracked feller-bunchers to decrease the number of buttons the operator must push. It’s another one of the ways the company is trying to reduce the variability in productivity caused by the operator.

To track and measure the machine’s productivity and its mechanical condition, John Deere has also designed a telemetrics system called TimberLink, currently available in all markets worldwide. TimberLink works by combining a Machine Telematics Gateway (MTG) that includes cellular communication and GPS antennas. Machine CAN data is collected by the MTG and wirelessly transferred to a data server, where it’s made available via the JDLink website.

A satellite modem is available for areas where cellular signals are not available or reliable. Lawler says the current automated systems have been incorporated so the automation can be switched to manual mode for operators that have the skills for high productivity. Although, he points out that tracking by the telemetrics systems indicates experienced operators rarely turn off the systems because even they see the value in increased automation. Perhaps a bit surprisingly, Lawler says the incentive to implement this type of advanced technology has been the changing demographics of forestry operators. “Historically, there hasn’t been a high demand for technology from within the forestry industry,” he says. “Traditionally, the experience level of forestry equipment operators has been at a very high level.

However, as many experienced operators retire from the industry it is becoming increasingly difficult to attract new operators that are willing to work in remote areas, in a job that can be quite repetitive. This turnover is impacting the average total productivity of equipment and driving the need to make equipment more intuitive for less experienced operators.” Additionally, Lawler notes the integration of technology needs to be done in a way that doesn’t negatively impact the reliability and uptime of the equipment. “While similar sensing technologies have been successfully used in both the automotive and industrial sectors for many years, sourcing sensors and microprocessors that will survive in the harsh forestry environment, at a cost-effective price, has traditionally been a challenge,” says Lawler.

“More recently, being able to jointly develop and leverage robust technology with our Agricultural business has been a significant enabler to reliably bring this type of technology to the forestry industry at a price point that the industry can support.” The changing demographics are also a real and pressing matter for the end users of those products such as lumber mills who are relying on logging operators to supply feedstock. That’s a major concern for companies such as Montreal-based Resolute Forest Products Inc. The company, which owns or operates some 40 pulp and paper mills and wood products facilities in the United States, Canada and South Korea, produces newsprint, specialty papers, market pulp and wood products.

Forestry, like many sectors, is facing a shortage of skilled workers — in particular, operators of logging and mobile equipment, lumber, pulp and paper mills who are continually looking for electrical, mechanical, energy and automation specialists, says Mike Martel, vice-president of forest products operations for Resolute. The forest sector jobs today are more specific in terms of the technical knowledge and specialization required to run and maintain equipment which places increasing demands for advanced training and skills for operations to achieve their potential, he says.

“The future in my mind is that we not only need to figure out how to attract people to the business, but to determine how we can operate with fewer people, because of labour availability in many regions,” says Martel. Resolute has three new operations in Ontario that require fill 400 positions — half in the forestry operations and half in the mills. “At the new saw mill we’re building right now, we have replaced human positions with technology,” says Resolute.

“For instance, we’ll incorporate advanced camera systems to measure the flow of lumber through the mill instead of an individual monitoring it. That will allow us to run faster and produce more. “We also design our machine centres so that specialists far removed from the mill location can connect, monitor and make adjustments on the fly which lessens the demand to have very specific skills in each operation.” For forestry equipment operators, Martel feels the ability of technology to provide more information is essential for them to make faster and more informed decisions, in other words, becoming more efficient in operating and maintaining equipment.

This information is in the form of a digital display providing a precise location of the operator and equipment allowing the operator to optimize his work in a dense forest populated by streams, slopes and other landscape features that need to be protected. And, automation for forestry harvesting isn’t necessarily limited to forwarders or other transportation vehicles. Martel says the use of drones is very promising. “For example, we now have a logging contractor who uses a drone to help spot sensitive features and assist in planning road construction and harvesting operations.

Another supplier has acquired a drone to measure the size of wood chip piles,” says Martel. “It’s common for chip piles to contain 40,000 to 50,000 tonnes of material. The drone can quickly and accurately calculate the size of chip inventories compared to past methods.” But, Martel feels there are limits to the benefits of automation. Technology has great promise in terms of improving the safety, cost and efficiency of operations, but at the end of the day, a truck is still a truck. It has an engine and a payload and requires an experienced operator to run it. Payloads are improving with the use of technologies such as structural composites, which remove weight without compromising strength and are providing the capability for loading more product on the truck safely within the regulations.

The advantage of lab analysis

While the automotive sector already has an established track record in the use of robotics, and the mining sector has also moved significantly in that direction, the design and development of automation technology for the forest industry requires some very specific parameters, explains Simon Westerberg, a recent PhD graduate from the Department of Applied Physics and Electronics at Sweden’s Umeå University. Westerberg, who, along with colleagues, has published a number of papers describing interactive studies examining potential forestry technologies, says traditional industrial robotics is performed in known, structured and controlled environments. This means that robotic systems know what to expect (i.e. known objects at known locations), which makes tasks such as object recognition and localization much easier. In the forest, sensors and other equipment must handle large variation in temperature, weather and lighting conditions.

Hydraulic systems are also harder to model than systems driven by electrical motors, further complicated by changes in physical parameters caused by temperature variations. Some of these problems can also be found in automated construction and mining. One notable difference, though, may be that, in forestry machine automation, the focus is on efficiency (speed), while accuracy and precision in each sub-task is less relevant and the consequences of erroneous interpretation of sensor data may be more crucial in mining and construction scenarios. Westerberg says one of the objectives for the studies he and his colleagues carried out was to improve harvesting productivity through the use of automation when compared to the standard manual control.

“In current forest machines, the operator can be a bottleneck when it comes to manipulating machinery, such as the crane used in a forwarder,” he says. “Because of the complicated control interface, the operator does not use the full capacity of the machine. This can be, at least partly attributed to the complexity and non-intuitiveness of the current joystick-based control method.” One solution that has been suggested is crane-tip control, or Cartesian control (the use of three linear sliding joints that move much like wrist action), which allows the operator to control the crane tip using the 3-D Cartesian co-ordinates of the crane tip instead of the 4-D joint co-ordinates employed in the current double joystick method. Because the co-ordination of the joints is automated, there’s less joystick manipulation and a lower workload.

Westerberg believes one approach to automation that could resolve the operator bottleneck and be realizable in a shorter term is to allow autonomous execution of certain simple crane motions, such as the repetitive loading and unloading motions. Automated motions not only relax the workload of the human operator and introduce natural breaks with time for muscle recovery, but they can also be planned with the intention of optimizing some performance index.

“Our motion-planning method allows us to plan crane motions that are optimized,” he says. “This kind of optimization with many variables is more or less impossible to perform for a human mind but relatively easy for a computer given a correct model of the machine.”

Taking automation to the next level

How viable is a remote-controlled forwarder, one in which the operator controls the machine at a distance? Known as teleoperation, it’s another potential technology being examined by Westerberg and his colleagues. “There are many advantages to completely remove the cabin, such as lower weight, reduced machine production costs and new machine designs that are currently impossible because of current cabin placement,” Westerberg says.

“However, removing the cabin is only practically feasible when the sensor and remote visualization systems are robust and reliable enough to maintain a high productivity. In an earlier stage, where sensing capacity is limited, remote teleoperation may still be viable in order to improve the situation for the operator. A primary example would be during work in steep terrain, where there is a higher risk for an overturning vehicle, or otherwise high-risk locations. The operator could then later switch back to in-cabin control for higher productivity in lower-risk operations”.

Some teleoperation systems use streaming video as feedback, which can be considered a cheap and easy solution. Another option is to use a virtual environment to present a virtual dynamic model of the remote location. This would also be applicable where the view of the operator is limited in the cabin. In a situation such as this, a detailed representation of the local environment is added in the virtual environment and objects near the vehicle are sensed, recognized, classified, and dynamically added. This can help the operator to plan and perform operations, but where it is not reliable or fast enough (as in real-time) to be the only source for visual input. That is, the operator would use the monitor to select target locations, but also use direct view to verify target locations.

There are many different sensor technologies to enable this such as laser range finders, radar, sonar, lidar, 3-D cameras and regular 2-D cameras; each have different advantages and drawbacks regarding speed, resolution, field of view, invariance to weather and lighting conditions. Even with these available technologies, object recognition in an unknown and unstructured environment is in no way an easy problem to solve. “Many challenges are still left to address, and I don’t want to speculate too much about which technologies will solve these problems except to say that most likely a combination of different sensors would be required,” Westerberg says.

“Complete remote teleoperation will likely not be introduced directly. The way I can see it happen is, step by step, first as local teleoperation introduced to solve some specific problem, for smaller machines for early thinning or other operations, where an optimal machine design would require the exclusion of a cabin. “Once such teleoperated machines are being used, it is not unlikely that additional features, such as augmented reality feedback, would be added for better visual operation at a larger distance.

This technology would also be transferred to other types of machines where removing cabins also could be a long-term goal.” Westerberg says his collaboration with industry in producing various studies has convinced him that manufacturers are increasingly optimistic these types of technologies will eventually become reality. “In recent years, I feel that the attitudes have transitioned from a more passive monitoring of research results and trends to active development towards more intelligent machines,” he says.

“The fact that companies have started adding new sensors to their products and present new system architectures that allow more software-based intelligent control is a sign that this process has already started. Now I rather get the feeling that many companies do not only want to avoid being left behind, but instead want to lead the development.”

Ernest Granson is a Calgary-based writer and editor, and a contributor to PROCESSWest.