Carbon Detector housing for the required components
The implementation of an autonomous carbon/bath detector and seperator is unique anywhere in the world.
MHO Infrastructure had been
Components:
The seperation of the carbon and bath materials is currently complete manually.
The ability to use an automated process to seperate the components has the ability to ....
MHOI has implemented a carbon detector for line 1 at BSL in 2019 as a proof of concept about the feasibility of an automated system. This v1 implementation used an AI trained model and colour camera to distinguish between the two materials.
False Positive rate
This project required a significant amount of research and development to implement and improved on the v1 model. As part of the improvement it was decided to move away from using a convential camera and colour spectrum analysis and use a multispectrical camera. The Australian CSIRO were asked to conduct an extensive investigation into the light spectrum and intensity difference between the two materials. Through initial implementation of a new AI trained model on the spectrium data it was evident that the rate of false positives was still signnificantly high and a secondary source of truth would be needed.
To reduce the false positives of the belt detection a 3d laser line scanner was selected. This would be used to determine the material height.
The implementation of the carbon detector is unique anywhere in the world
The designed of the system required careful consideration of the selected components.
The construction and installation of the system