Twenty years ago the developers of the optimisation algorithm of AIDimension (previously Procomp) from Oulu ran into a problem. The calculation system developed by the company to optimise logistics and mobile work required so much processor power, that no reasonable server solution seemed sufficient to run the calculation fast enough. Mathematically, the size of the optimisation problems to be solved by the algorithm was huge: even with a small amount of data, there were more possible route options than there are atoms in the entire universe. Calculation of large optimisation problems could take up to a day or more.
So something had to be done. In AI Dimension’s product development, the issue was solved with a parallel computing method, which knew how to divide the calculation to be done simultaneously by two or more different processors. The method worked, and AI Dimension’s system began to produce functional route and schedule results efficiently and reliably. At the same time, the patent process was started.
And now the work has finally paid off.
“In July we received confirmation from the US patent authorities that our application has gone through,” says Timo Kaartinen, AI Dimension’s CEO. “The process took ten years with all the complaints and clarifications, so sometimes you thought it was over. In the past, we have already received approval in China, but in the EU the process is incomplete. Business-wise, the US patent is very significant right now.”
The demand for patents related to parallel computing is growing worldwide, especially due to the fast growth of artificial intelligence applications. “Artificial intelligence requires huge amounts of computing power, so surprisingly we are in the eye of the storm in a positive sense. For example, Nvidia is growing its parallel computing patent portfolio at accelerating speed,” says Kaartinen. “The ideas of the patent granted in our programs have been used for a long time. Now a completely new field of use is also opening up for the patent.”
Although AI Dimension’s method has its roots in the beginning of the last decade, in a technical sense the solution is not outdated at all. “On the contrary, we were ahead of our time,” says Timo Kaartinen. “Processor powers have increased during this time, but so has the amount of necessary calculations. Our solutions reduce the need for the use of super-efficient processors, when thanks to parallel computing even smaller ones can do the job.”
More information:
Director of Business Development Timo Kaartinen
AI Dimension (previously Procomp), a software company from Oulu, has been developing optimisation solutions for logistics and home care for over twenty years. US 11,694,129 B2, registered in July 2023, is the company’s first patent accepted in the United States. It deals with parallel computing methods related to route optimisation, logistics and production and shift planning on two or more processor platforms.