Sign of uncertainties in Dimension3d

[ainar]

open-simulation-interface/osi_common.proto

Line 92 in a8fb7cc

// The dimensions are positive. Uncertainties are negative or positive.

On this line, the documentation let the uncertainties on Dimension3d values be negative. However an uncertainty, according to the RMSE formula, can only be positive. Is it an error or intentional?

[ghost]

Interesting finding! Needs to be checked

[ainar]

By going further, I notice that actually the problem can be bigger and is generic to every uncertainty.

It is difficult to characterize an error/tolerance on a measurement. The error/tolerance can depend on more factors, we need to define on which and how many.
For example, if a Dimension3d message characterize the dimension of a car, its tolerance depends on environmental conditions, the temperature of the car, its materials, etc. but the tolerance will not change regardless if you measure the width, the length or the height.

Depending on how precise you want OSI to be, these informations should be implemented in a specific new message for each kind of tolerance.

[jdsika]

@PhRosenberger the definition of uncertainties is also a major part in the ISO23150 and should be clarified in #367

[PhRosenberger]

@PhRosenberger the definition of uncertainties is also a major part in the ISO23150 and should be clarified in #367

I agree.

[jdsika]

Could you have a look at his @FlorianMueller87, please?

[FlorianMueller87]

I agree that uncertainties could be positive and negative. The ISO defines them with real value types, see e.g. A.1.23 of the ISO.
The ISO defines the error model in B.4.1 which includes trueness and precision, B.4.1.3.2.
For the harmonisation with ADI an error-model would also be necessary.

[yash-shah-asam]

@thomassedlmayer Could you please look at this in the sensor modeling group?