Hi all! On Android ARPointCloud has confidenceValues available. However, what exactly those values mean? It seems like ARCore documentation "The normalized confidence of the point" is as vague as it can be. E.g. what exactly is confidenceValues.Value[v] == 0.1? Does it mean that this Feature Point has 10% chance of being a good one and 90% of being wrong/fake? It looks like the most reasonable explanation, but why "normalized" then? Or does it mean that this feature point is within 90% of some standard deviation (e.g. 1 meter - thus 0.1 confidence may be up to 90cm away from a real point)? Also internet search gives me some results for "normalized confidence" keyword, including "self-normalized confidence", however I couldn't find a clear definition and it seems that it is defined differently in different papers (seems like the bottom line would be "confidence divided by some 'ideal' confidence for the scenario").