Photo analysis and volume indicators for optimization of the recycling value chain

Important results from the project

Image recognition: Purpose, recognize unwanted material. Aim, to detect well enough so we can replace manual picking. Compliance, we can show that detection is good, but must add automatic pick to be able to prove the solution. Volume measurement: Volume scan. Purpose, get a product to replace scales. Compliance, failed, aborted. Filling level container. Purpose to know when containers must be emptied to optimize transport. Completion, went well. The solution is with a customer. Contract proposals: Purpose, everyone should feel safe. Compliance, well, all parties were satisfied.

Expected long term effects

Image Recognition: Result, a product that can display / detect unwanted material on belt. The result is as expected. But to get full results, automatic pickup must be added. Volume measurement: Result, Volume scan with Lidar. Did not work. We expected a "measure" for volume produced. Filling level container. Result, a product that can be used by a customer for transportation optimization. Went as expected and are now in use. Contract proposals: Results, We have agreements that are all satisfied with. As expected.

Approach and implementation

The project has been running so that we have meetings regularly and that Stena has provided production facilities where tests have been carried out. This has been done in a good way but a major challenge was to not disturb the production when testing. For example. it is difficult to see if the image detection is correct when the belt is running at 1m / s and there is a large amount of material here. Without possibility to analyze afterwards and then "the correct answer" is already gone.