Can robots be incorporated into fish processing?

by Thomas Jensen
industrial robot

Automatic systems – Effortless, fast and efficient

This article was featured in Eurofish Magazine 2 2024.

Fish and seafood are delicate food products that require careful handling. The fishing industry has only minimally implemented automated systems, or robots, into seafood processing in comparison to other industries. However, given the increasing challenges of staff shortages, escalating costs, and supply fluctuations, the fish industry must now explore the possibility of utilising automated systems and think seriously about them.

Fish are a natural commodity that exists in a multitude of species and sizes, but unfortunately, they are prone to spoilage very quickly. This has a significant impact on processing and storage procedures, which in turn affects the quality and safety of aquatic products. In the realm of fish processing, speed is a crucial factor. However, this often presents challenges due to staff shortages or fluctuating landing data. Despite these challenges, the level of quality requirements by the customer continues to rise, placing immense pressure on processing companies. There is an urgent need for action in nearly all areas, including sorting, various processing steps, packaging, storage, and shipping of the products. Similar challenges exist in other industries, where ­innovative technologies are increasingly being utilised to replace human labour, maintain precision, and enhance competitiveness. At the core of this transformation are intelligent processing machines, particularly robots, whose performance is constantly ­improving thanks to image recognition systems, artificial intelligence and sensor technology. Where simple mechanics once prevailed, today, self-learning systems are often employed, enabling a partnership between human
and machine.

Due to the increasing global demand for industrial robots, an international competition has emerged for the best solutions. In 2022, European Union member states alone installed approximately 72,000 units, representing a 6% increase from the previous year. Historically, the automotive industry has been the primary consumer of industrial robots. Everyone is probably familiar with images of deserted production halls where robots accurately assemble or paint car bodies, seemingly all controlled by some magic hand. However, robot technology is now being utilised in other sectors such as metal, machine, plastics, and chemical industries. With the rapid advancement of robotics, these remarkable machines are taking on more and more tasks. Robots are more precise and versatile than humans, working tirelessly around the clock without fatigue or shaky hands. Furthermore, modern robots can be just as sensitive as human workers.

Science fiction becomes a daily reality

The emergence of science fiction as a daily reality is a phenomenon that has been long in the ­making. In 1921, the writer Karel Čapek introduced the concept of humanoid robots in his drama R.U.R. (Rossum’s Universal Robots), coining the term “robot” from the Czech word robota (slave labour). The play depicts an unscrupulous entrepreneur who seeks to replace his workers with artificial beings. Čapek‘s vision of ­human-like machines has continued to shape the public imagination to this day. Such robots in human form already exist in certain fields, such as the care of the elderly and catering, but they constitute only a small fraction of the spectrum. In fact, robots in everyday life often have a vastly different appearance, such that they are no longer immediately recognizable as such. For instance, mobile robots are now used to mow lawns and vacuum homes, defuse mines, plough fields, and transport materials in various industries as autonomous carriers. Not all robots have arms or resemble humans, but they are increasingly important in relieving humans of heavy, tiresome, boring, dirty, or dangerous work. Some even interact directly with humans, as in the case of “cobots” (a composite of “collaboration” and “robot”), which are designed to work closely with humans.

Despite the growing use of robots in various industries, the fish ­processing industry has yet to fully embrace this technology. Much of the work is still carried out ­manually, as this allows for better consideration of the individual characteristics of the fish. Even in areas where machines are already in use, they typically require manual feeding. Fully self-sufficient processing lines, in which humans only perform control and supervisory tasks, are exceedingly rare. However, there are certain aspects of fish and seafood processing that could benefit from automation. Some areas literally “cry out” for the use of intelligent machines. This is because some activities are monotonous and repetitive and the working conditions unpleasant and uninviting. Severely cold rooms, humid air, slippery floors and sharp cutting tools can even pose a health hazard to employees. The volatility of supply chains also causes problems. As fresh produce arrives or market demand suddenly picks up, action must be taken quickly, which makes working hours ­difficult to predict. Under these conditions, it is not easy to find suitable personnel, especially since the rise of labour costs. As wages rise, some producers have relocated labour-intensive processes to countries with cheaper production costs. However, this leads to an increasing number of protests in their respective countries over job losses.

The potential of robots is only selectively used

In the medium term, greater mechanisation and automation of fish processing is unavoidable. With fish species such as salmon, herring or mackerel, which are available in large quantities in approximately equal sizes, mechanical processing has long been common. This is more difficult with non-homogeneous catches and fish of different sizes, which first have to be laboriously sorted before they can be processed. This is currently still mostly done manually, but it could just as easily be done by robots. With intelligent image recognition systems, sorting objects of different types, textures, shapes and sizes according to defined criteria is no longer an unsolvable problem. Both robust products such as deep-frozen scallops and sensitive fresh fish and even delicate items such as breaded fish fingers can be handled quickly and gently using flexible end-of-arm tools (EOAT) with suitable sensors and adapted grippers. It should no longer be an unsolvable problem for intelligent systems, which can correctly identify objects and their position in boxes or on a conveyor belt and safely grip them, to be supplied to this precisely-targeted machine processing.

A worthwhile area of application for intelligent, self-learning systems and robots is the filleting of fish. Mechanical filleting machines today already cut very precisely and with high yields, but in the end, they often still work through routines that have been bluntly set. Individual differences between each animal are hardly taken into account. Experienced skilled workers still achieve an average of one to two percent higher yields when hand filleting. However, modern computer vision technology makes it possible to measure moving objects from a distance. With this support, for example, automatic beheading and filleting machines can be precisely adjusted to the shape and size of each individual animal. This fine adjustment takes place in fractions of a second in automated systems, which guarantees a very high working speed. The image processing software controls the knives of the filleting machine and ensures optimal cutting, which significantly improves the fillet yield. Such robots not only support fish processing, but also represent an entry into the complete automation of processing systems.


High requirements make it difficult to use

The utilization of intelligent systems and robots in fish processing remains a challenge due to the high demands of the industry. While the benefits of such technology are compelling, the ­inadequate supply and high acquisition costs of these systems, coupled with the rigorous requirements of fish processing, make it a difficult field of application. The harsh conditions of cold and damp rooms, exposure to salt, vinegar, and fish slime, pose a significant strain on the technology, particularly the sensitive electronics and sensors. The high hygiene standards within the food sector are added on top of this. Everything must be easy-to-clean and disinfect so as not to create a breeding ground for bacteria or other germs. This is a challenging task, especially with robot technology, which often features joints, cables, and hydraulic hoses. Nonetheless, viable solutions do exist. The Norwegian research company Sintef employs the six-axis robot VS-087 from Denso Robotics for fish filleting. The robot uses fine sensors to precisely measure and analyze each fish, allowing the computer to determine where the knife should be applied and where skin and bones remain. The robot also identifies and separates unsightly blood spots accurately. Marel‘s FleXicut platform utilizes X-ray technology to detect remaining bones in fillets, which are then removed cleanly using a fine, high-pressure water jet.

The path towards achieving fully automated processing chains for fish and seafood is still relatively lengthy, however, it is no longer an unattainable vision of the future. A plethora of ­potential applications are being developed in the areas of packaging and palletizing of fish products, although several delicate work steps must also be addressed. At the fish international trade fair in Bremen, K-Robotix and its partner team unveiled a packaging system for smoked fish that utilises industrial robots from Kawasaki and Toshiba. The first robot lines Styrofoam boxes with foil. The second robot places the fish in the prepared boxes, which are then stacked on pallets by the third robot. What appears to be a simple process in theory is an enormous challenge in practice. Smoked mackerel fillets are susceptible to breakage and must be handled with care to ensure they reach their packaging undamaged. The most challenging aspect of the system is the robot, whose gripper is controlled by an image processing system. The camera captures the contours of each individual fillet and calculates their focal points to position the gripper accurately. This is an exceedingly complex task, as even a peppercorn or a spot of fat on the grill can interfere with the image recognition software. Additionally, the gripper must not grasp too tightly. Consequently, the cost of controlling the robots is exceedingly high. Experts estimate that the cost of the necessary peripherals is approximately three times the price of the robot.

The feasibility of industrial applications has been demonstrated through various examples, such as the impressive results achieved by the IPL robot from Marel. This robot can be configured for numerous pick-and-place applications in fish processing, including the ­handling of both fresh and frozen fish or portions, which are taken off the conveyor and placed directly into trays or thermoformer bags. The robot‘s image processing system recognizes important features such as size, position, and orientation of the products and turns them in the correct direction if necessary. Marel also offers intelligent robotic solutions for packaging small-piece fish products that need to be combined to specified target weights, with the RoboBatcher series. The robot‘s sophisticated grippers ensure that even delicate fish products are not damaged during packaging. The state-of-the-art dosing software “batches” the product close to the target weight with minimal wastage.

Similarly, KUKA robots have been successfully employed in palletizing fish crates at Pakfish in Ruska, Poland since 2020. They are then frozen and serve as a buffer for other fish processors, especially canners. In order to meet the high demand, up to 80 tons of fish every day have to be packed in boxes weighing around 10 kg and stacked on pallets. The robots have relieved employees of physically demanding and tiring work, while also offering enormous potential for the retail trade, particularly in the area of packaged fillets in modified atmosphere packaging (MAP). These products are popular with consumers because they have a predictable calculated weight, can be personally selected by the customer without any service personnel, and can often even be prepared in the retail box. This has led to significant sales increases, especially since the possibilities in this area are far from exhausted. Just how important MAP products have become is demonstrated by the sales success of salmon in discount stores, where the customer-friendly 250 and 300 gram packs have become a popular item. This has noticeably boosted sales volumes of fresh salmon, even though MAP trays are said to account for less than 10 percent of the total market. With portioning and packaging robots, this profitable marketing channel could certainly be significantly expanded not only for salmon but also for other fish species.

Consequently, what applies to the packaging and palletizing of fish products also applies to the reverse process, unpacking and depalletizing. Robots can relieve humans of the burden of handling and emptying large fish boxes, cardboard boxes, and heavy insulating tubs. In some salmon processing companies, depalletizing robots are already in use, which can empty two cartons at the same time and supply two processing lines with their contents. Robots can also make a contribution to automating the process chains when it comes to infeed of raw materials.

Novel areas of application are continuously emerging

A promising domain for industrial robots is the cleaning and disinfection of processing plants. The stringent hygiene requirements in food processing companies render this task an essential but mostly unpopular permanent responsibility. Presently, machines and premises in fish processing are predominantly cleaned manually. The daily repetitive cleaning work is arduous, relatively costly, and poses a certain risk to staff and the ­environment due to the use of hazardous disinfectants and other chemicals. Despite professional routine and experience, human error occasionally creeps in, resulting in poor hygiene, endangering food safety, and ­necessitating costly, image-damaging product recalls for the producer. Such risks can be ­mitigated by employing robots. If the cleaning robots are appropriately “trained,” i.e., programmed for the spatial conditions and particularly critical corners, they execute all the specified tasks according to the best practices of industrial cleaning. Always with the same care, without tiring or getting distracted. Their use can even be environmentally friendly and material-saving because robots do not waste any water and precisely maintain the programmed concentrations of the cleaning agents.

Modern cleaning and disinfection robots should even be ­capable of reacting interactively to different bacterial contaminations. They use sensitive sensor technology to measure the concentration of germs on the surfaces of the machines and can rework the contaminated areas accordingly. Tests have ­demonstrated that when it comes to cleaning fish processing plants, robots perform as well as human cleaners with 15 years’ experience, as measured by the reduction in bacterial counts.

However, particularly high expectations rest on cobots, those collaborative robots that support people in many jobs and improve productivity. Cobots can be utilised for a variety of different applications, from pick-and-place jobs to feeding processing machines to packaging the finished products. Since cobots act “intelligently” and stop immediately when touched, they do not pose any danger to humans unlike conventional industrial robots, which enables direct cooperation. Innovative features such as machine vision should soon even enable cobots to carry out quality controls. All this demonstrates that robots also have a promising future in fish processing!

Manfred Klinkhard

You may also like