System and method for discerning human input on a sensing device
Systems and methods for detecting and classifying types of physical inputs on an input surface of a human machine interface (HMI) input structure are disclosed. In response to a physical input on the input surface, one or more sensor signals are received from respective sensors associated with the HMI input structure. One or more features are determined for each received sensor signal. Based on the one or more features for each sensor signal, a position on the input surface is determined by classifying the one or more sensor signals. The classification of the one or more sensor signals can be performed by one or more machine learning algorithms. Based on a classification of the physical input, an action associated with the determined location is executed.
1 . A method for detecting and classifying physical inputs on an input surface of an input structure, wherein the input structure the input surface and a plurality of sensors, and wherein the method comprises:
polling at least one of the plurality of sensors at a defined polling frequency;
updating a baseline level for each of the at least one of the plurality of sensors until the baseline level converges during a quiescence period when the polling of the at least one of the plurality of sensors at the defined polling frequency is occurring;
receiving a sensor signal generated by at least one of the plurality of sensors in response to a physical input on the input surface;
detecting an input event based on the physical input wherein detecting the input event based on the physical input comprises determining that a voltage level of the sensor signal from the at least one of the plurality of sensors is diverging from the updated baseline level from the quiescence period;
determining one or more features associated with the sensor signal and the at least one of the plurality of sensors;
determining a location of the physical input on the input surface by classifying the sensor signal using a machine learning algorithm based on the one or more features; and
executing an action associated with the determined location on the input surface.
2 . The method of claim 1 , wherein each sensor in the plurality of sensors is configured to measure changes to the input surface of the input structure or to detect movement on the input surface.
3 . The method of claim 2 , wherein each sensor in the plurality of sensors comprises at least one of a force sensor, an infrared sensor, a strain gauge, or an ultrasonic sensor.
4 . The method of claim 1 , further comprising filtering the sensor signal from the at least one of the plurality of sensors or to remove noise from the sensor signal.
5 . The method of claim 1 , further comprising filtering the sensor signal from the at least one of the plurality of sensors by estimating and removing drift.
6 . The method of claim 5 , wherein the drift is removed by at least one of Kalman filtering, deconvolution, or high pass filtering using a Butterworth technique.
7 . The method of claim 1 , wherein determining the sensor signal from the at least one of the plurality of sensors is diverging from the quiescence period comprises determining the sensor signal from the at least one of the plurality of sensors is diverging from the quiescence period based on an indication that the voltage level of the sensor signal associated with the one or more of the plurality of sensors has surpassed the baseline level by an activation threshold.
8 . The method of claim 7 , wherein the activation threshold is based on a baseline signal level that is updated at select times based on sensor signals received from the plurality of sensors.
9 . The method of claim 1 , wherein the one or more features are determined using a magnitude and a polarity of a difference between a signal level of the sensor signal and a baseline signal level for the sensor signal.
10 . The method of claim 1 , wherein the sensor signal is further classified based on a rate of change of the sensor signal during an active period.
11 . The method of claim 1 , wherein the machine learning algorithm is at least one of a Decision Tree induction, a Random forest, a gradient boosting tree, a Nearest Neighbor, or a Support Vector Machine.
12 . A system, comprising:
a processing device;
an input structure comprising an input surface and a plurality of sensors below the input surface; and
a memory operably connected to the processing device and storing processor-executable instructions, that when executed by the processing device, cause operations to be performed, the operations comprising:
polling at least one of the plurality of sensors at a defined polling frequency;
updating a baseline level for each of the at least one of the plurality of sensors until the baseline level converges during a quiescence period when the polling of the at least one of the plurality of sensors at the defined polling frequency is occurring;
receiving a sensor signal generated by the at least one of the plurality of sensors in response to a physical input on the input surface;
detecting an input event based on the physical input wherein detecting the input event based on the physical input comprises determining that a voltage level of the sensor signal from the at least one of the plurality of sensors is diverging from the updated baseline level from the quiescence period;
determining one or more features associated with the sensor signal and the at least one of the plurality of sensors;
determining a location of the physical input on the input surface by classifying the sensor signal using a machine learning algorithm based on the one or more features; and
executing an action associated with the determined location on the input surface.
13 . The system of claim 12 , wherein determining the location of the physical input on the surface by classifying the sensor signal based on the one or more features comprises determining the location of the physical input on the input surface by classifying the sensor signal using a machine learning algorithm based on the one or more features.
14 . The system of claim 12 , wherein determining the sensor signal from the at least one of the plurality of sensors is diverging from the quiescence period comprises determining the sensor signal from the at least one of the plurality of sensors is diverging from the quiescence period based on an indication that that the voltage level of the sensor signal associated with the one or more of the plurality of sensors has surpassed the baseline level by an activation threshold.
15 . The system of claim 14 , wherein the activation threshold is based on a baseline signal level that is updated at select times based on sensor signals received from the plurality of sensors.
16 . The system of claim 12 , wherein each sensor in the plurality of sensors comprises at least one of a force sensor, an infrared sensor, a strain gauge, or an ultrasonic sensor.
17 . The system of claim 12 , wherein the memory stores further processor-executable instructions for filtering the sensor signal from the at least one of the plurality of sensors or to remove noise from the sensor signal.
18 . The system of claim 12 , wherein the memory stores further processor-executable instructions for filtering the sensor signal from the at least one of the plurality of sensors by estimating and removing drift.
19 . The system of claim 18 , wherein the drift is removed by at least one of Kalman filtering, deconvolution, or high pass filtering using a Butterworth technique.