I have an application which requires I count the amount of inventory in a silo. The inventory i am counting are cigarette packs that are wrapped in cellophane. They are stacked in a metal silo where the packs fit neatly up against all 4 walls. I am taking the measurement from the roof of the silo looking down at the top most pack and calculating the distance. I am then using that distance to derive how many packs are left in the stack. the range I am working with is 4cm to 26cm.
My issue is the following. The distance reading I get is not stable. If I take a pack and flip it over I can get 2 different values. The difference in values can be so large that in some instances I can completely remove the pack and the range wont change. When i try to place a white paper on top of the pack the values change again however the the readings are far more stable. This issue does not effect all packs as some can be flipped and the ranges stay constant. I have also tried to spray paint the silo walls (which are metal) mat black and that made no difference. I tried to take the packs out of the silo and using a tripod measured them against a black background and still the issue persists. I have tried using the VL53L1 and VL53L0 sensor with Nucleo-x board running the test bench software. I played with some of the preset parameters and adjusted the ROI to the smallest zone and although it makes the issue slightly better it does not eliminate it.
Am i missing something here? is there a parameter that i can change that would help with this issue?
