Oliver Harrald

Add multi-distance hypothesiser

2

The standard DistanceHypothesiser only allows a fixed measure to be used. If one has incoming detections of BearingRange type, and some of Cartesian type, clearly the distance should be calculated...

Currently, the model uses transpose of emission matrix. This is not correct. I've left this as an issue in the hope of a discussion. https://github.com/dstl/Stone-Soup/blob/14db74f6557a46ff3ddd8db12e2d23480ecdfee2/stonesoup/models/measurement/categorical.py#L63

While easy enough to understand in context of its branch, a standalone commit message can be difficult to understand the meaning of when merged to the main branch. This isn't...

Some data readers build a dictionary when being iterated through. For example, the `CSVGroundTruthReader` builds a `groundtruth_dict` so that it can keep track of path ids that it has come...

Tutorial covariances changed to squares

3

Not an essential change, but it may provide some clarity. Perhaps when defining covariance matrices / noise in the tutorials, we could use squares (eg. noise_covar = np.diag([2**2]) instead of...

Many metric component calculations are repeated (see SIAPMetric generator). Caching component values would make metric generation quicker.

#285 comments highlight the need to be considering some measurement models as something other than 'reversible' or 'irreversible' For example, range-rate models' inverse functions currently return approximations to state (cartesian)...