‘Compared to human labour, artificial intelligence (AI) yields a more consistent quality. Robots are particularly good at tasks that involve routine work,’ stated Jaan Kraav, the head of production at AS Tarmeko Spoon. The positive impact of AI can be seen both in tangible results and in numbers.
Tarmeko Spoon, which had been looking for a way to boost production efficiency and eliminate certain bottlenecks for some time, decided to participate in an AI development marathon organised by Tehnopol, which brought them a triumphant capital injection and an investment that changed the company’s life.
Out of more than 20 candidates, the jury initially selected 14 artificial intelligence projects for manufacturing companies, of which the top four were granted a total of 190,000 euros for investment and to use towards achieving the project goals. The winners included the joint project of Tarmeko Spoon, a veneer and plywood manufacturer, and Leanest, with whom the company had previously collaborated.
A visible improvement
The 65,000-euro grant enabled the company to deploy an AI-driven camera that is able to identify defects in veneer while still on the turning line, determine the quality of the material, and, based on this information, decide where the material needs to be cut. The AI heavily outperforms human labourers. The final results of the project should be revealed in a year or two.
The AI has the potential to speed up the operation of the veneer-turning line by up to 30%, as well as increase the quality of dried veneer by 3% compared to the system in use today. This would allow the company to cut one whole shift from their operations. And if demand should grow, it would not be a problem for the company either.
Going forward, the company is also planning to deploy AI for the sorting of finished products. This job is currently performed by a total of six people working in three shifts, who will be assigned to other duties. The deployment of AI and the increase in efficiency does not mean that any employees will be laid off, simply that their tasks and work processes will be adjusted and revised.
According to Kraav, human labour is irreplaceable. While AI can perform routine or high-precision tasks at speeds that are unattainable by humans, the whole process still needs to be managed and defined by a person. The company currently has three robots in operation, but is not planning to stop there. Tarmeko Spoon has managed to escape the effects of both the coronavirus crisis and the war in Ukraine so far. In fact their business is growing rather than shrinking.
If the AI project with which Tarmeko Spoon, in partnership with Leanest, won the AI development marathon organised by the Tehnopol science and business park and for which they were awarded funding is carried out in full, the company can expect an up to 15% increase in turnover as well as lower production costs thanks to a reduced need for timber and power. And these are not the only benefits of the deployment of the AI system, although, according to Jaan Kraav, it is still early to go into details about everything.
How does a manufacturing plant arrive at artificial intelligence?
In light of the above, other manufacturing enterprises might find themselves asking how you can determine whether investing in AI is reasonable or how you can get there. What are the risks? ‘Tarmeko Spoon had actually been looking for solutions to various problems for some time, but had not managed to find anything that would be a perfect fit,’ Kraav revealed. ‘Our good partner Leanest OÜ did offer one solution earlier on, but since it was on the expensive side, and implementation would have taken a very long time, the plan was dropped.’ Eventually, Leanest came up with an idea for utilising AI in a way that would be more affordable as well as cost-effective for Tarmeko Spoon.
‘The first project we started working on was a log-sorting line. An AI-driven camera would assess quality and would quickly identify all defective logs. A human operator might not notice every flaw across the full circumference of a log, but an AI camera certainly could.’
The original plan for the project presented at the AI marathon was actually to boost the efficiency and output of another production line. ‘We were planning to install a camera that would tell the line where it is best to cut the material, but found that this would be extremely expensive. So we set about looking for a different solution. Eventually, we arrived at what we ended up presenting at the AI marathon with Leanest: a way to avoid veneer being sent on for drying if it would later be rejected anyway during post-processing, as this involves a major commitment of time and resources. So, as you can see, we approached the bottleneck from a completely different angle. And it was the right call! Now we can determine the quality of the veneer right away instead of later on down the road, after committing resources unnecessarily.’
Meeting with AIRE
During the AI marathon that took place in winter, Tarmeko Spoon also met with representatives of AIRE, or AI & Robotics Estonia. AIRE is Estonia’s candidate for the European Digital Innovation Hubs Network, which is financing hubs across Europe in 2022–2025. AIRE can be considered an Estonian digital innovation hub that offers help to every Estonian company interested in AI or the latest technologies. Co-operation with researchers from the Tallinn University of Technology and the University of Tartu further boosts the credibility of the organisation.
It was AIRE that recommended a mentor to Tarmeko Spoon to help implement the AI solution and advise the company during the project. After all, AI in theory and AI in actual production processes, where there are a number of other factors to consider, are two totally different things.
The support person that was engaged for the project, who deals with a wide variety of AI solutions on a daily basis, is Tähve Lõpp – a man who has been active in IT for 22 years and who has been involved in the development of all manner of software solutions. For the past three years, he has served as the head of the Smart Manufacturing Unit at Ericsson Eesti, where several exciting innovations are underway, the details of which, according to Lõpp, are still too early to discuss. Most of them have to do with AI, of course. ‘The only hint I can give is that, one day, there might a digital twin of Ericsson’s Tallinn factory… It will take time, obviously, but it will happen eventually.’
AI cannot help where production processes are not in order
Tähve Lõpp disclosed that although he has developed various software solutions in Estonia and abroad both for private and public sector entities for ten years, his heart now belongs firmly to industry. Which is one reason why he wants to contribute to the deployment of AI at Tarmeko Spoon. ‘I admit that I have always been drawn to the manufacturing sector, perhaps because it’s oriented towards producing something concrete that we need in our daily lives. At Ericsson, for example, we manufacture mobile devices so people can communicate with each other.’
At the same time, it is no secret that in Estonia the manufacturing sector is lagging behind many other sectors when it comes to digitisation (maybe as much as ten years, in fact), but AI will eventually also make its way there. ‘Not today, not yet, but the time will come, and quite soon.’ According to Tähve, any company that wants to remain in business for another 10 or 15 years should definitely be thinking about AI, or else they will lose any advantages they currently may have before they realise it. The future belongs to AI.
‘The primary prerequisite is for production processes to be in order; they must be airtight. AI cannot fix flawed processes. When considering innovative solutions, you should first ascertain how useful it will be in the aggregate.’ For AI technology, the expert proposed the following rule of thumb: if it will get you the necessary changes in a year, make the investment right away; if in two years, take some more time to think about it. If it will take three years, however, you are better off ditching the plan. ‘Automation can be done in many different ways; the key is to determine where efficiency is most needed and what makes the most sense – i.e., where it will save you the greatest amount of time and money. At Ericsson, analysing the company’s enormous testing datasets by means of a machine learning system is such a point,’ Lõpp noted. Companies that are struggling with the decision as to whether or not to deploy AI can certainly seek assistance from AIRE, but there are also a number of other good IT experts and consultancy firms that can help with this in Estonia. ‘Just find them, asking for advice does not require making any investments,’ Lõpp said.
The projects of AS Tarmeko Spoon and Ericsson Eesti were supported under Tehnopol’s AI programme in co-operation with AI & Robotics Estonia (AIRE).