Abstract: In some implementations, techniques may include at a plurality of times while a user of the first portable device is moving with the first portable device: performing ranging at a respective position with a second device to determine a respective distance, thereby determining a plurality of respective distances, where the second device is stationary; obtaining raw measurements from a motion sensor of the first portable device. In addition, the device may include using the raw measurements at the plurality of times to determine relative positions at the plurality of times, the relative position from an initial position. The techniques may include estimating a second position of the second device that optimizes a loss function that includes differences of the respective distances at the relative positions and the actual distance between the relative positions and the second position.

Abstract: A device implementing a system for machine-learning based gesture recognition includes at least one processor configured to, receive, sensor data for a first window of time and additional sensor data for a second window of time overlapping the first window of time. The sensor data and the additional sensor data are provided as inputs to a machine learning model, the machine learning model having been trained to output a predicted gesture, predicted gesture start time, and predicted gesture end time based on the sensor data. A predicted gesture is determined based on an output from the machine learning model, and to perform, in response to determining the predicted gesture, a predetermined action on the device.

Abstract: Touch sensor panels/screens can include a first region having a plurality of touch electrodes and a second region without touch electrodes. In some examples, to improve touch sensing performance, a first algorithm or a second algorithm is applied to determine whether an object corresponding to the touch patch is in contact with the touch screen. Whether to apply the first algorithm or the second algorithm is optionally dependent on the location of the touch patch.

Abstract: Aspects of the subject technology provide improved gesture detection including collection of data from multiple sensors into a data structure that may be analyzed to estimate a plurality of gesture inferences, and then the plurality of gesture inferences may be integrated into a detected gesture for the period of time. Analysis of the data package may be performed by separate machine learning models, each model producing a corresponding gesture inference.

Abstract: In some examples, touch data can include noise. Machine learning techniques, such as gated recurrent units and convolutional neural networks can be used to mitigate noise present in touch data. In some examples, a gated recurrent unit stage and a convolutional neural network stage can be arranged in series, such as by providing the output of the gated recurrent unit as input to the convolutional neural network. The gated recurrent unit can remove noise caused by a first component of the electronic device and the convolutional neural network can remove noise caused by a second component of the electronic device, for example. Thus, together, the gated recurrent unit and the convolutional neural network can remove or substantially reduce the noise in the touch data.

Abstract: A device implementing a system for machine-learning based gesture recognition includes at least one processor configured to, receive, from a first sensor of the device, first sensor output of a first type, and receive, from a second sensor of the device, second sensor output of a second type that differs from the first type. The at least one processor is further configured to provide the first sensor output and the second sensor output as inputs to a machine learning model, the machine learning model having been trained to output a predicted gesture based on sensor output of the first type and sensor output of the second type. The at least one processor is further configured to determine the predicted gesture based on an output from the machine learning model, and to perform, in response to determining the predicted gesture, a predetermined action on the device.

Abstract: In some examples, touch data can include noise. Machine learning techniques, such as gated recurrent units and convolutional neural networks can be used to mitigate noise present in touch data. In some examples, a gated recurrent unit stage and a convolutional neural network stage can be arranged in series, such as by providing the output of the gated recurrent unit as input to the convolutional neural network. The gated recurrent unit can remove noise caused by a first component of the electronic device and the convolutional neural network can remove noise caused by a second component of the electronic device, for example. Thus, together, the gated recurrent unit and the convolutional neural network can remove or substantially reduce the noise in the touch data.

Abstract: Embodiments are disclosed for hold gesture recognition using machine learning (ML). In an embodiment, a method comprises: receiving sensor signals indicative of a hand gesture made by a user, the sensor data obtained from at least one sensor of a wearable device worn by the user; generating a first embedding of first features extracted from the sensor signals; predicting a first part of a hold gesture based on a first ML gesture classifier and the first embedding; generating a second embedding of second features extracted from the sensor signals; predicting a second part of the hold gesture based on a second ML gesture classifier and the second embedding; predicting a hold gesture based at least in part on outputs of the first and second ML gesture classifiers and a prediction policy; and performing an action on the wearable device or other device based on the predicted hold gesture.

Abstract: Touch sensor panels/screens can include a first region having a plurality of touch electrodes and a second region without touch electrodes. In some examples, to improve touch sensing performance, a first algorithm or a second algorithm is applied to determine whether an object corresponding to the touch patch is in contact with the touch screen. Whether to apply the first algorithm or the second algorithm is optionally dependent on the location of the touch patch.

Abstract: In some examples, an electronic device can use machine learning techniques, such as convolutional neural networks, to estimate the distance between a stylus tip and a touch sensitive surface (e.g., stylus z-height). A subset of stylus data sensed at electrodes closest to the location of the stylus at the touch sensitive surface including data having multiple phases and frequencies can be provided to the machine learning algorithm. The estimated stylus z-height can be compared to one or more thresholds to determine whether or not the stylus is in contact with the touch sensitive surface. In some examples, the electronic device can use machine learning techniques to estimate the (x, y) position and/or tilt and/or azimuth angles of the stylus tip at the touch sensitive surface based on a subset of stylus data.

Abstract: In some examples, touch data can include noise. The noise can be generated by a component of an electronic device that includes a touch screen. For example, one or more signals transmitted to the display circuitry of an electronic device can become capacitively coupled to the touch circuitry of the device and cause noise in the touch data. Machine learning techniques, such as gated recurrent units and/or convolutional neural networks can estimate and reduce or remove noise from touch data when provided data or information about the displayed image as input. In some examples, the algorithm includes one or more of a gated recurrent unit stage and a convolutional neural network stage. In some examples, a gated recurrent unit stage and a convolutional neural network stage can be arranged in series, such as by providing the output of the gated recurrent unit as input to the convolutional neural network.

Abstract: A device implementing a system for machine-learning based gesture recognition includes at least one processor configured to, receive, from a first sensor of the device, first sensor output of a first type, and receive, from a second sensor of the device, second sensor output of a second type that differs from the first type. The at least one processor is further configured to provide the first sensor output and the second sensor output as inputs to a machine learning model, the machine learning model having been trained to output a predicted gesture based on sensor output of the first type and sensor output of the second type. The at least one processor is further configured to determine the predicted gesture based on an output from the machine learning model, and to perform, in response to determining the predicted gesture, a predetermined action on the device.

Abstract: A device implementing a system for machine-learning based gesture recognition includes at least one processor configured to, receive, from a first sensor of the device, first sensor output of a first type, and receive, from a second sensor of the device, second sensor output of a second type that differs from the first type. The at least one processor is further configured to provide the first sensor output and the second sensor output as inputs to a machine learning model, the machine learning model having been trained to output a predicted gesture based on sensor output of the first type and sensor output of the second type. The at least one processor is further configured to determine the predicted gesture based on an output from the machine learning model, and to perform, in response to determining the predicted gesture, a predetermined action on the device.

Abstract: Cross-coupling correction techniques on a touch sensor panel can be improved using machine learning models (particularly for touch sensor panels with relatively low signal-to-noise ratio). In some examples, the machine learning model can be implemented using a neural network. The neural network can receive a touch image and perform cross-coupling correction to mitigate cross-talk due to routing traces of the touch sensor panel. Mitigating cross-talk can improve touch sensing accuracy, reduce jitter, and/or reduce false positive touch detection.

Abstract: A method for authenticating a user is disclosed. The method includes collecting, by a processor of an electronic device and while the electronic device is worn by a user, measurement data from a set of sensors of the electronic device. The method also includes providing, by the processor and to a machine-learning model, the collected measurement data from the set of sensors and previously collected sets of measurement data for a known user. The method also includes obtaining, by the processor, an indication of whether an extracted feature set is similar to one of a number of classified feature sets. At least one of the classified feature sets is classified as belonging to the known user and generated based on the previously collected sets of measurement data for the known user. The method also includes determining, by the processor, whether the user is the known user based on the obtained indication.

Abstract: In some examples, touch data can include noise. Machine learning techniques, such as gated recurrent units and convolutional neural networks can be used to mitigate noise present in touch data. In some examples, a gated recurrent unit stage and a convolutional neural network stage can be arranged in series, such as by providing the output of the gated recurrent unit as input to the convolutional neural network. The gated recurrent unit can remove noise caused by a first component of the electronic device and the convolutional neural network can remove noise caused by a second component of the electronic device, for example. Thus, together, the gated recurrent unit and the convolutional neural network can remove or substantially reduce the noise in the touch data.

Abstract: Embodiments are disclosed for a machine learning (ML) gesture recognition with a framework for adding user-customized gestures. In an embodiment, a method comprises: receiving sensor data indicative of a gesture made by a user, the sensor data obtained from at least one sensor of a wearable device worn on a limb of the user; generating a current encoding of features extracted from the sensor data using a machine learning model with the features as input; generating similarity metrics between the current encoding and each encoding in a set of previously generated encodings for gestures; generating similarity scores based on the similarity metrics; predicting the gesture made by the user based on the similarity scores; and performing an action on the wearable device or other device based on the predicted gesture.

Abstract: A device implementing a system for machine-learning based gesture recognition includes at least one processor configured to, receive, from a first sensor of the device, first sensor output of a first type, and receive, from a second sensor of the device, second sensor output of a second type that differs from the first type. The at least one processor is further configured to provide the first sensor output and the second sensor output as inputs to a machine learning model, the machine learning model having been trained to output a predicted gesture based on sensor output of the first type and sensor output of the second type. The at least one processor is further configured to determine the predicted gesture based on an output from the machine learning model, and to perform, in response to determining the predicted gesture, a predetermined action on the device.

Abstract: A device implementing a system for machine-learning based gesture recognition includes at least one processor configured to, receive, from a first sensor of the device, first sensor output of a first type, and receive, from a second sensor of the device, second sensor output of a second type that differs from the first type. The at least one processor is further configured to provide the first sensor output and the second sensor output as inputs to a machine learning model, the machine learning model having been trained to output a predicted gesture based on sensor output of the first type and sensor output of the second type. The at least one processor is further configured to determine the predicted gesture based on an output from the machine learning model, and to perform, in response to determining the predicted gesture, a predetermined action on the device.

Abstract: Finger detection and separation techniques on a multi-touch touch sensor panel can be improved using machine learning models (particularly for touch sensor panels with relatively low signal-to-noise ratio). In some examples, a machine learning model can be used to process an input patch to disambiguate whether the input patch corresponds to one contact or two contacts. In some examples, the machine learning model can be implemented using a neural network. The neural network can receive a sub-image including an input patch as an input, and can output a number of contacts. In some examples, the neural network can output one or more sub-image masks representing the one or more contacts.