CNH & Robovision
One of the most important items on the roadmap of CNHi is the development of autonomous vehicles. In Zedelgem, the team is focusing on automating combines, balers and forage harvesters. A first step on that roadmap for combine harvesters has been made commercially available last year: IntelliSenseTM.
IntelliSenseTM is a feature that senses the internal state of the harvester and automates a huge part of the internal settings of the harvester. Crucial element in the IntelliSenseTM system is the smart camera that analyses grain sample cleanliness. The initial version of this camera made use of state of the art image processing techniques. Relatively fast the team reached the boundaries of this methodology: algorithms only capable in capturing limited complexity and algorithm development and tuning took huge amount of time.
In order to break those boundaries, the team discovered the advantages of artificial intelligence and contacted Robovision to execute a proof of concept. Feasibility on accuracy was proven fast, and most effort was spent on trying to embed the network on the current platform which is not 100% suited for running complex neural networks. Goals were met and during a follow up project we´ll investigate future proof hardware.