This is fairly rare, but can happen in some specific scenarios. For example, a gray cat on gray hardwood floor in a dim room may be difficult for Felik to reliably detect. There are, however, tweaks that can be made to the settings to still make Felik work well in this situation.
In other cases Felik may stop detecting your pet after they have stopped moving for a while. Felik has a number of algorithms to try to determine which objects are moving and which are stationary. For example, if you get up from a table and move a chair, Felik may for some time detect that as movement and avoid the chair, but will soon realize that the chair is part of the room. If your pet stops moving and lays down, eventually Felik may incorporate your pet into it's background model of the room. However, most pets will perform some movement if the laser gets too close or on them, in which case Felik will once again detect them and move the laser away.
We define failures to detect the actual pet as False Negatives.
Reducing False Negatives
If Felik is just not detecting your pets, the first step should be increasing the Sensitivity value. If this doesn't help, it may be that the pet and the floor are too similar in color, or the room is too dark. Try making the room brighter.
If your pet is the "lazy type" who likes to lay around during a play session, Felik may begin ignoring them. Note that if your pet starts moving, Felik will pick them up again. In our testing, if the laser gets too close to an animal, they tend to react by moving, which causes Felik to detect them and move the laser away.
As our classifiers (detection algorithms) improve, we expect to be able to better differentiate between an inanimate object and a pet. So in the future we hope to hone the algorithm to the point that it detects even non-moving targets if they resemble a living being.