Moisture sensor and/or defogger with bayesian improvements, and related methods
In certain example embodiments, moisture sensors, defoggers, etc., and/or related methods, are provided, More particularly, certain example embodiments relate to moisture sensors and/or defoggers that may be used in various applications such as, for example, refrigerator/freezer merchandisers, vehicle windows, building windows, etc. When condensation or moisture is detected, an appropriate action may be taken (e.g., actuating windshield wipers, turning on a defroster, triggering the heating of a merchandiser door or window, etc.). Bayesian approaches optionally, may be implemented in certain example embodiments in an attempt to improve moisture detection accuracy. For instance, models of various types of disturbances may be developed and, based on live data and a priori information known about the model, a probability of the model being accurate is calculated. If a threshold value is met, the model may be considered a match and, optionally, a corresponding appropriate action may be taken.
1. A method of classifying perturbations to an interface comprising a substrate supporting a conductive coating, the method comprising:
generating, using a first set of electrodes provided at a first location relative to the substrate, an electrical field;
measuring, using one or more second sets of electrodes, parameters of the electrical field at one or more second locations, each said second location being different from the first location; and
classifying a region of the interface into a specified state based on the measured parameters of the electrical field;
responsive to the specified state being indicative of a first perturbation type, disregarding the classification; and
responsive to the specified state being indicative of a second perturbation type different from the first perturbation type, sending an output signal to control electronics;
wherein the first and second perturbation types distinguish between electromagnetic interference (EMI), one or more kinds of moisture-related events, and one or more kinds of human interface interaction related events.
2. The method of claim 1 , wherein EMI is treated as a first perturbation type.
3. The method of claim 1 , wherein the one or more kinds of moisture-related events include a plurality of moisture-related events including icing, condensation, and other types of moisture.
4. The method of claim 3 , wherein the moisture-related events are treated as a second perturbation type.
5. The method of claim 1 , wherein the one or more kinds of human interface interaction related events include a plurality of human interface interaction related events including touching of the interface and hovering.
6. The method of claim 1 , wherein a human passing by the interface is treated as a first perturbation type.
7. The method of claim 6 , further comprising sending a control signal based on the accepted model.
8. The method of claim 7 , wherein control signals associated with the models are modifiable over time based on machine-learned parameters related to respective control signal target(s).
9. The method of claim 1 , wherein the classification distinguishes between propagating electromagnetic waves, moisture-related perturbations to the interface, and human interactions with the interface.
10. The method of claim 1 , wherein pre-stored references are models derived experimentally.
11. The method of claim 10 , wherein the models are adjustable over time based on continued interactions with the interface.
12. The method of claim 1 , further comprising:
determining whether the specified state is indicative of a perturbation non-event; and
responsive to a determination that the specified state is indicative of a perturbation non-event, disregarding the classification.
13. The method of claim 12 , wherein disregarding of the classification includes:
generating a new electrical field using the first set of electrodes;
measuring new parameters of the new electrical field at the one or more second locations, using one or more second sets of electrodes; and
classifying the region of the interface into a new specified state, based on the new parameters and the plurality of different pre-stored references.
14. A method of classifying perturbations to an interface comprising a substrate supporting a conductive coating, the method comprising:
(a) generating, using a first set of electrodes provided at a first location relative to the substrate, an electrical field;
(b) measuring, using one or more second sets of electrodes, parameters of the electrical field at one or more second locations, each said second location being different from the first location;
(c) classifying a region of the interface into a specified state based on the measured parameters of the electrical field and a plurality of different pre-stored references, the classifying being performed in connection with a machine learning model; and
(d) refining the different pre-stored references over time through multiple repetitions of (a) through (c);
responsive to the specified state being indicative of a first perturbation type, disregarding the classification; and
responsive to the specified state being indicative of a second perturbation type different from the first perturbation type, sending an output signal to control electronics;
wherein:
the specified state is indicative of a first perturbation type or a second perturbation type; and
the first and second perturbation types distinguish between electromagnetic interference (EMI), one or more kinds of moisture-related events, and one or more kinds of human interface interaction related events.
15. The method of claim 14 , wherein the machine learning model includes Bayesian analysis.
16. The method of claim 14 , wherein the pre-stored references are models, initially derived experimentally.
17. The method of claim 14 , wherein the pre-stored references are treated as background information for an inference engine.
18. The method of claim 5 , wherein the pre-stored references are models and the classification accepts one of the models.
19. The method of claim 14 , wherein the state is classifiable as being one of a plurality of different perturbation types, the different perturbation types including electromagnetic wave propagation and human interaction with the interface.