Machine learning for personal care devices

A machine learning system with neural networks analyzes personal care device data to predict usage features and provide feedback, addressing the need for user feedback on device condition and recommending timely replacements, enhancing user experience.

KR103046933B1Active Publication Date: 2026-11-11EDGEWELL PERSONAL CARE BRANDS LLC
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
KR1020237002680
Authority / Receiving Office
KR · KR
Patent Type
Patents
Current Assignee / Owner
Priority Date
2020-06-24
Filing Date
2021-06-24
Publication Date
2026-11-11
Estimated Expiration
2041-06-24

AI Technical Summary

Technical Problem

Personal care devices, such as razors, have a limited lifespan and require user feedback on usage and condition to prevent poor performance and potential injuries, which existing systems fail to provide effectively.

Method used

Implementing a machine learning system with sensors and a neural network to analyze data from personal care devices, predicting usage features and providing feedback on device condition and recommending replacements.

Benefits of technology

Enhances user experience by accurately predicting device end-of-life and providing timely recommendations for maintenance or replacement, thereby preventing poor performance and injuries.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 112023008285451-PCT00001_ABST
    Figure 112023008285451-PCT00001_ABST
Patent Text Reader

Abstract

A system, device, and method for applying machine learning to provide feedback related to a personal care device are described. An embodiment of the system, device, and method may identify data from a device such as a razor comprising a plurality of sensors on the razor, predict one or more features such as shaving features based on the data using a machine learning model trained using data from the plurality of sensors, and provide shaving feedback based on one or more features such as shaving features. An embodiment of the system, device, and method may train a machine learning model, execute the trained machine learning model, capture and / or process data for the machine learning model, and provide feedback based on the output of the trained machine learning model, or provide a combination thereof.
Need to check novelty before this filing date? Find Prior Art

Description

Technology Field

[0001] Cross-reference regarding related applications

[0002] This application claims priority to International Application PCT / US2021 / 038856 filed on June 24, 2021, which claims priority to U.S. Provisional Application No. 63 / 043,451 filed on June 24, 2020, the contents of which are incorporated herein by reference in their entirety.

[0003] The following concerns machine learning in general for personal care, more specifically for personal care devices. Background Technology

[0004] Many people use various tools and devices for personal care and grooming. The effectiveness of these tools or devices depends on factors such as how they are used, whether they are in good condition, or whether they require repair or replacement (of some or all of their components). As such, these tools or devices often have a limited lifespan (or at least one of their components), and therefore their effectiveness may decrease as that lifespan nears its end.

[0005] If a user continues to use a product after it has exceeded its useful life or requires parts replacement or general rework, the user may have an unpleasant experience and potentially suffer injury. Therefore, systems and methods are needed to obtain and provide user feedback regarding the use and technology of these personal care and grooming products, as well as the condition of these tools and / or devices.

[0006] A non-limiting example of a personal care and grooming device or tool is the razor. Many people use razors for personal care and grooming. The effectiveness of a razor can vary depending on factors such as how the razor is used and whether the blade is sufficiently sharp. Consequently, razor blades have a limited lifespan, and their effectiveness may decrease towards the end of their service life.

[0007] If a user applies poor shaving techniques or if the razor is nearing the end of its lifespan, the user may have a poor user experience or suffer injuries. Therefore, there is a need in the industry for systems and methods to obtain and provide user feedback regarding shaving techniques and the condition of the razor.

[0008] Personal care devices may include those used for grooming or cleaning, such as razors, trimmers, derma planning tools, face and body brushes and scrubbers, water picks, dental cleaning tools, and toothbrushes. Personal care devices may include devices that require replacement because they have a limited lifespan, regardless of whether the replacement is for the entire personal care device or a component of the personal care device.

[0009] A method, apparatus, non-transient computer-readable medium, and system for machine learning for a personal care device are disclosed. An embodiment of the method, apparatus, non-transient computer-readable medium, and system identifies data from a personal care device comprising a plurality of sensors on the personal care device, predicts one or more usage features based on the data using a machine learning model trained using data from the plurality of sensors, and provides feedback based on one or more usage features.

[0010] The device may include a processor, a memory that communicates electronically with the processor, and instructions stored in the memory. The instructions enable the processor to identify data from a personal care device comprising a plurality of sensors, use the data from the plurality of sensors to predict one or more usage features based on the data, and provide feedback based on one or more usage features.

[0011] A non-transient computer-readable medium may store a code for personal care. In some examples, the code includes instructions executable by a processor to: identify data from a personal care device comprising multiple sensors; predict one or more usage features based on the data using a machine learning model trained using data from the multiple sensors; and provide feedback based on one or more usage features.

[0012] In some embodiments, a plurality of sensors include an inertial measurement unit (IMU), a capacitance-to-digital converter, a force sensor, a rinse sensor, a grip sensor, a temperature sensor, an accelerometer, a gyroscope, a magnetometer, a resistance contact, a piezo contact microphone, a Hall effect sensor, an external camera, a strain gauge, a load cell, a reed switch, a linear variable differential transformer, or any combination thereof.

[0013] In some embodiments, the machine learning model is trained using additional data for at least one additional sensor not included in the personal care device. In some examples, the at least one additional sensor includes a force sensor.

[0014] Some examples of methods, devices, non-transient computer-readable media, and systems may further include receiving data via a wireless communication connection to a personal care device. In some examples, a machine learning model is trained using survey data collected from a user, and the survey data includes usage quality information, device end-of-life information, or both.

[0015] In some embodiments, one or more features may include shaving features such as the number of shaving strokes, shaving quality information, razor end-of-life information, or any combination thereof. In some examples, the feedback may include shaving feedback such as instructions to improve shaving quality. In some examples, the shaving feedback includes razor replacement feedback.

[0016] A method, apparatus, non-transient computer-readable medium, and system for machine learning for a personal care device are described. An embodiment of the method, apparatus, non-transient computer-readable medium, and system can collect data from a plurality of sensors on a personal care device, identify one or more usage features based on the data, identify a subset of sensors, wherein at least one sensor is not included in the subset of sensors, and train a machine learning model to predict one or more features using data from the subset of sensors.

[0017] The device may include a processor, a memory that communicates electronically with the processor, and instructions stored in the memory. The instructions may be operated so that the processor collects data from a plurality of sensors on a personal care device, identifies one or more usage features based on the data, identifies a subset of sensors, wherein at least one sensor is not included in the subset of sensors, and trains a machine learning model to predict one or more features using data from the subset of sensors.

[0018] A non-transient computer-readable medium may store a code for personal care. In some examples, the code includes instructions executable by a processor to: collect data from a plurality of sensors on a personal care device, identify one or more usage features based on the data, identify a subset of sensors, wherein at least one sensor is not included in the subset of sensors, and train a machine learning model to predict one or more features using data from the subset of sensors.

[0019] In some examples, at least one sensor includes a force sensor. In some examples, one or more features may be a face feature including the number of strokes, stroke orientation, or both.

[0020] Some examples of the methods, devices, non-transient computer-readable media and systems described above may further include providing feedback based at least partially on a machine learning model trained using data from a subset of sensors.

[0021] Methods, devices, and systems for personal care devices are described. Embodiments of the methods, devices, and systems may include a hub and a personal care device. The personal care device may provide data directly to the hub, and the hub may provide data to a server. The data may be raw data, and the server may be located in the cloud. The server may implement one of the machine learning methods described above to generate feedback based on the data. The server may also implement one of the machine learning methods to generate replacement orders for components of the personal care device. The feedback may be provided to a computing device linked to the hub and / or the personal care device.

[0022] In some examples, data may be provided wirelessly from a personal care device to a hub. Data may be generated by at least one of a plurality of sensors placed in the personal care device. Feedback may be provided wirelessly to a computing device linked to the hub and / or the personal care device.

[0023] In some examples, the link between the hub and the server may be secure. In one embodiment, raw data may be transmitted from the hub to the server's dedicated IP address. Brief explanation of the drawing

[0024] FIG. 1 illustrates an example of a system for personal care and / or grooming according to an aspect of the present disclosure. FIG. 2 illustrates an example of a razor according to an aspect of the present disclosure. FIG. 3 illustrates an example of interactive razor hardware according to an aspect of the present disclosure. FIG. 4 illustrates the structure of an Internet of Things (IoT) razor system according to an exemplary embodiment of the concept of invention. FIG. 5 illustrates a hub according to an exemplary embodiment of the concept of the invention. FIG. 6 illustrates an example of a feedback server for a personal care system according to an aspect of the present disclosure. FIG. 7 illustrates an example of a process for feedback machine learning for a personal care device according to an aspect of the present disclosure. FIG. 8 illustrates an example of a sensor reduction server for a personal care system according to an aspect of the present disclosure. FIG. 9 illustrates a process for sensor reduction machine learning for a personal care device according to an aspect of the present disclosure. Specific details for implementing the invention

[0025] The present disclosure generally relates to personal care, health, hygiene, and grooming, and more specifically to using machine learning methods to obtain information related to personal care devices and provide feedback.

[0026] Non-limiting embodiments of the present disclosure use machine learning techniques to identify razor data, predict shaving characteristics based on the razor data, and provide shaving feedback. In some cases, the razor data may include life information for predicting cartridge end-of-life. Thus, in one exemplary embodiment, using razor data can predict the end-of-life of a cartridge. The razor data may include any type of signal indicating that the end user needs to replace the cartridge. Data-driven cartridge replacement recommendations can provide the end user with an overall better shaving experience. To this end, data procured from users and optionally specific end users can lead to cartridge end-of-life prediction. In some examples, sensor data from the razor or other personal care device may be collected and analyzed without using machine learning to predict feedback.

[0027] Embodiments of the present disclosure may use deep machine learning models. Such models may include interactive personal care or grooming tools, tools, or devices, such as interactive razors. Learning models may also be referred to as limited sensor post-processing software development. One embodiment may remove unnecessary sensors while retaining sufficient data to be used to predict useful information for the end user, such as the end of life of replaceable components like the end of life of a razor cartridge—additional feedback that the user may be interested in may also be obtained through data generated by the device, tool, or device. Furthermore, the simplification of device designs, such as razor designs, by removing at least some of the excluded sensors by training machine learning algorithms with fewer sensors can lower the manufacturing cost of devices such as razors and enable competitive and economical production.

[0028] An exemplary embodiment of the concept of the invention disclosed herein describes an Internet of Things (IoT) razor system capable of automatically providing a user with razor blade replacement and / or feedback based on user behavior. Briefly, the IoT razor system comprises a razor equipped with a plurality of sensors that generate raw data as the user shaves, and a short-range wireless communication device that outputs the raw data. The IoT razor system further comprises a hub configured to receive raw data from the razor and provide the raw data to a server via the cloud. The server then processes the raw data to, for example, predict the end of life of a razor cartridge and generate a replacement order that is automatically issued for a new razor cartridge based on the predicted shaving characteristics (e.g., predicted end of life, etc.).

[0029] Additionally, notifications based on predicted shaving characteristics may be provided to the user. Notifications may include, but are not limited to, diagnostic feedback (e.g., commands regarding alternative shaving procedures, current lifespan / usage level, etc.), recommendations for various consumer products (e.g., different types of cartridges based on predicted shaving characteristics, pre- or post-use skincare products such as shaving aids based on predicted shaving characteristics, etc.), replacement order confirmations (e.g., order status, shipping status, order confirmation or acceptance, etc.), messages (e.g., visual, auditory, tactile, etc.). In one embodiment, notifications may be delivered to the shaver user via an IoT device located in conjunction with any or all components of the IoT shaver system at a geographical location. For example, notifications may be delivered via a smart mirror, a smart faucet, or a smart shower head, a user computing platform (e.g., a smartphone, a smartwatch, etc.), diodes of the shaver and / or hub, etc.

[0030] Embodiments of the present disclosure use one or more artificial neural networks (ANNs). In some cases, the ANN can enhance a function that includes a large number of input parameters or features. Since the ANN can learn from all input parameters and features, the relationships may have higher dimensions. The ANN can enhance the ability to implicitly detect complex non-linear relationships between dependent and independent variables. The ANN can also enhance the ability to detect all possible interactions between input parameters and features.

[0031] Some embodiments of the present disclosure may use variations of ANNs, including deep neural networks and shallow neural networks. In some cases, deep neural networks may not rely on the generation of secondary (e.g., derived) features and may learn implicitly from raw data features. Deep neural networks consider possible interactions, although they may be difficult for the user to interpret. On the other hand, shallow neural networks may include secondary features predetermined by domain knowledge. Shallow neural networks may be easier to interpret but result in simpler models with fewer interactions.

[0032] As used herein, the terms 'device', 'personal care device', 'instrument', 'implement', and 'tool' may be used interchangeably. 'Stroke count' may refer to the number of strokes, such as a shaving stroke, per single use. 'Orientation' may refer to the vertical and horizontal positional changes and durations of a tool, implement, or device, such as a razor, and tendencies observed in the handle grip during use (e.g., while shaving). 'Temperature' may refer to the temperature of a tool, implement, or device, such as a razor and its components, during use. 'Duration' may refer to the typical time a consumer spends using a tool, implement, or device. 'Shave duration' may refer to the typical time a consumer spends on a single shave. 'Force' may refer to the amount of pressure applied to the face or body and the amount of drag across the surface while using a tool, instrument, or device, such as during grooming or shaving. 'Pivot measurement' may refer to the angle of the head of a tool (engaging against the face or body) being handled (held by the user) during personal care or grooming, such as handling the razor head while shaving. 'Rinse duration' may refer to the amount of time spent under a stream of water from a sink, bathtub, or shower. 'Date and Time of Grooming' may refer to the date and time of use and time records of grooming tools, instruments, or devices. 'Date & Time of Shave' may refer to shaving date and time records and records of the trends in shaving dates and times that typically occur for the user.'Count' may refer to the number of times a tool, instrument, or device is used during a specific period, such as daily, weekly, or monthly. 'Shave count' may refer to the number of shaves performed during a specific period, for example, monthly or weekly. 'Geographic location' may refer to 'geotagged' records of consumer personal care or grooming, such as shaving at home or while traveling.

[0033] As used herein, the processed data may include: date, period, number of strokes (obtained from force and / or IMU sensors), number of rinses (obtained from head capacitance sensors), total rinse time, ambient temperature, minimum, maximum, and / or average force, number of strokes in each orientation (e.g., vertical upward, vertical downward, horizontal) and / or histogram of the force of the strokes. The processed data may further include: strokes (e.g., start index, end index, maximum force, gravity axis), analysis graphs and / or filtered data.

[0034] As used herein, the processed shaving data may include: shaving date, shaving period, number of strokes (obtained from force and / or IMU sensors), number of rinses (obtained from head capacitance sensors), total rinse time, ambient temperature, minimum, maximum and average force, number of strokes in each orientation (e.g., vertical upward, vertical downward, horizontal) and / or a histogram of the force of the strokes. The processed shaving data may further include: strokes (e.g., start index, end index, maximum force, gravity axis), analysis graphs and / or filtered data.

[0035] FIG. 1 illustrates an example of a system for personal care and / or grooming according to aspects of the present disclosure. The illustrated example includes a user (100), a device (105), a communication network (110), a storage device (115), a server (120), and an artificial neural network (ANN) (125).

[0036] A user may use a personal care device and / or grooming device (tool, instrument, or tool) such as the device (105). In the illustrated example, the device (105) is a razor configured to perform shaving or other grooming activities, but may include, for example, a trimmer, a derma planning tool, a facial and body brush and scrubber, a water pick, a dental cleaning tool, a toothbrush, etc. The device (105) may communicate with a server (120) via a communication network (110). The communication network (110) may include a communication network such as a mobile network, a wireless network, and the Internet. The server (120) may include an ANN (125). In some examples of the present disclosure, the ANN (125) may be trained using data from a plurality of sensors.

[0037] For example, the device (105) may be a training device that provides data (d) from its (n) sensors (n is an integer) to train a machine learning algorithm to determine one or more shaving characteristics. However, the device (105) may be a production device that provides less data (e.g., d-1) from n sensors and / or fewer sensors (e.g., n-1) so that the machine learning algorithm (e.g., trained) predicts one or more shaving characteristics specifically for a production razor. Thus, for example, the ANN (125) may predict one or more shaving characteristics based on input data representing end-of-life information of the device (105), for example. In another example, the ANN (125) may predict feedback information related to brushing teeth, drying hair, or any other personal grooming process based on the type of device used (e.g., toothbrush, etc.), input data, etc.

[0038] The device (105) is an example of a corresponding element or aspect of the elements described with reference to FIG. 2, or may include such an example.

[0039] The communication network (110) may include a network such as a mobile network, a wireless network, the Internet, or any combination thereof. According to some embodiments of the present disclosure, information about the device (105) (e.g., sensor data) may pass through the communication network (110) and be collected by the ANN (125) of the server (120).

[0040] One embodiment of the present disclosure establishes a communication link between a hub and an off-the-shelf personal care or grooming device, such as a razor, using an inertial measurement unit (IMU) (or a set of sensors sufficient to collect information necessary for a decision) or a tool configured to generate raw use data. The communication link may be a BT link. Other communication interfaces are also possible. The communication link may be encrypted (e.g., unidirectional, bidirectional, etc.).

[0041] The storage device (115) is an example of a corresponding element or aspect of elements described with reference to FIGS. 3, FIG. 7 and FIG. 9, or may include such an example.

[0042] The server (120) may be an example of an aspect of the corresponding element or elements described with reference to FIGS. 6 and FIGS. 8, or may include such an example. The server (120) may use machine learning techniques such as an ANN (125) to identify input data from a source (e.g., sensor data), predict one or more features, such as shaving features, based on the input data, and provide feedback, such as cartridge replacement, e.g., shaving feedback and / or notification(s), device (105) end-of-life information, etc. One embodiment of the present disclosure is a feedback server (600) as illustrated in FIG. 6. The feedback server (600) may train a machine learning model that provides feedback, such as shaving feedback, based on one or more features, such as shaving features. In another embodiment, a sensor reduction server (800) may train a machine learning model that predicts one or more shaving features using data from a subset of sensors.

[0043] An ANN (125) may be an example of an aspect of a corresponding element or element described with reference to the training elements (630, 835) of FIGS. 6 and FIGS. 8, respectively, or may include such an example. An artificial neural network (ANN) may be a hardware or software component comprising a number of connected nodes (known as artificial neurons) that can loosely correspond to neurons in the human brain. Each connection, or edge, may transmit a signal from one node to another (like physical synapses in the brain). When a node receives a signal, it may process the signal and then transmit the processed signal to another connected node. In some cases, the signals between nodes may involve real numbers, and the output of each node may be calculated by a function of the sum of its inputs. Each node and edge may be associated with one or more node weights that determine how the signal is processed and transmitted.

[0044] During the training process, these weights can be adjusted to improve the accuracy of the results (i.e., by minimizing the loss function corresponding to the difference between the current result and the target result in some way). Edge weights can increase or decrease the strength of the signal transmitted between nodes. In some cases, nodes may have a threshold at which no signal is transmitted at all. Nodes may also be aggregated into layers. Different layers may perform different transformations on the input. The initial layer may be known as the input layer, and the final layer may be known as the output layer. In some cases, a signal may cross a specific layer multiple times.

[0045] The ANN (125) may consider primary features that include all raw sensor data features. The ANN may also consider secondary (e.g., derived) features. Secondary features (e.g., derived) may include features that are more interpretable from the raw sensor data. The ANN (125) may include deep learning models and shallow learning models. The ANN (125) may determine the shave length from the IMU sensor data. For example, the ANN (125) may use acceleration and time data to calculate the length of use for personal care or grooming routines, such as the length of a shave.

[0046] In some cases, the feedback predicted by the ANN (125) may be based on shaving of a specific area of ​​the body (i.e., face, legs, or other body parts). The shaving length may be used as data to predict the end of life of the device (105). Machine learning models, methods, and systems may use input data and features indicating the end of life, and sensors that generate the most valuable data.

[0047] Some embodiments may use deep learning structures including deep neural networks (DNNs) and recurrent neural networks (RNNs). In one embodiment of the present disclosure, a force prediction recurrent neural network (RNN) may be used. An RNN is a multi-layer perceptron that learns how to process data over time. An RNN may be designed to learn how to manage memory of time-series data. Filtered sensor data may be processed as input data to an RNN. In some cases, an RNN may process fixed-length time-series sensor data. Fixed-length time-series sensor data may be taken from IMU data and capacitive sensor data, or solely from IMU data. An RNN may output a detected force for a fixed-length input. In particular, IMU data may be the most important indicator of force, represented by the calculation of the mean absolute error (the average difference between the predicted value and the target value). A first model using IMU data and capacitive sensor data, and a second model using only IMU data, can be improved over several epochs (e.g., 50, etc.) of training (e.g., low error). Thus, for example, a force sensor may be unnecessary in the sensor configuration.

[0048] According to the embodiment, the dense neural network may be a multi-layer perceptron that performs similarly to a logistic regression model. The DNN may be designed to learn meaningful combinations of input features. Secondary features may be processed as inputs to the DNN.

[0049] One embodiment of the present disclosure describes a network structure for learning. The structure may be a stacked Long Short-Term Memory RNN. A Long Short-Term Memory (LSTM) network is a type of recurrent neural network capable of learning order dependencies in sequence prediction problems. An LSTM can be designed to learn how long each data point is retained. The width (number of nodes) and depth (number of layers) may be increased to trade off speed and accuracy. The width and depth of the neural network are generally adjusted according to the size and complexity of the dataset. For example, experiments are performed on a dataset regarding grooming tools, instruments, or devices such as device (105). For this dataset, a decrease in accuracy may occur at 2 or more layers × 24 nodes. Some embodiments may use a 2 × 50 stacked LSTM-RNN.

[0050] An LSTM-RNN can be approximated as a function of input data with weights and biases such as Y=(X, W, B). Input data can pass through an LSTM-RNN composed of a neural network initialized with, for example, a specific width and depth. As an example, as the network traverses from the input to the output of the LSTM-RNN, the number of nodes and / or links connecting nodes between layers decreases. For example, each node in the initial or input layer may have a link to each node in the immediately next layer, but a single node in the final layer or output layer may have only two input links from each of the two nodes in the immediately previous layer.

[0051] Additionally, each point or channel of the input data may be weighted with the same or different weights and biased with the same or different biases. For example, when input data (X) is passed to a model (F) and the model output (Y) is optionally compared with the expected output data (Ygt), the weight (W) and bias (B) at time 0 may be randomly initialized values ​​close to 0.0 selected by the Glorot Uniform Initialization algorithm. Additionally, or alternatively, the IMU x-acceleration data at time 1 may be weighted by weight (W1) and biased by bias (B), and the IMU z-gyroscope data at time 1 may be weighted by weight (W6) and biased with the same or different bias (B(n)), and each bias value may be maintained and updated independently during training. Initial and updated parameters, such as width, depth, links, weights, and / or biases, may be based on the usage features to be predicted (e.g., the weights and / or biases used may differ when training a model for force compared to a model for distance traveled). Additionally, preprocessing (e.g., filtering, etc.) and data processing (e.g., training, etc.) may occur at a single data point (or channel), but the same steps may be performed in windows or collections of data that are sequential or overlapping. For example, IMU x-acceleration data from time 1 may be used, or five points of IMU x-acceleration data from time 1–5 may be used. Reference data corresponds to the same temporary window as the input data.

[0052] During training, the model's output data can be compared with the reference data to determine the % error and / or confidence level from the comparison between the model output data (Y) and the reference data (Ygt). If the % error or confidence level is unacceptable, the weights and biases can be adjusted, and the input data can be passed back through an updated LSTM-RNN to determine whether an acceptable % error and / or confidence level has been reached. The process can be executed iteratively until an acceptable % error and / or confidence level is achieved with a trained machine learning model that generates the lowest or most acceptable error. Thus, in one example, variables Y, X, W, and B can be represented as tensors in computer memory or storage, which may refer to an N-dimensional collection of numerical data (e.g., normalized and scaled between [0, 1] or [-1, 1]). Meanwhile, the error (e.g., loss) may refer to the average of the calculated mean squared differences between the model output data (Y) and the reference data (Ygt). Therefore, a learning algorithm (e.g., Adam Optimizer) can take the calculated loss as input and determine the direction and magnitude to adjust each weight and bias of the training model.

[0053] Evaluation of the trained model may also be performed to verify repeatability before a model structure that generates the lowest or acceptable error is selected. For example, the total set of training data may be partitioned so that a relatively small portion of the training data can be used for evaluation rather than training. A relatively small portion of the training data is passed to a candidate trained LSTM-RNN to evaluate the error or confidence for a specific usage feature by comparing it to reference data (e.g., average error of accumulated force across all shaves, average error of total distance traveled across all shaves, etc.). The trained machine learning model can then be used to predict usage features and / or improve functionality using production input data (e.g., substantially in real time while using the device). In this regard, usage features may be predicted from a subset of sensors (e.g., using only IMU data), and, for example, a personal care device such as device (105) may exclude, disable, and / or not require traditional sensors to identify, derive, or predict usage features.

[0054] In one example for training a model for force prediction, the input training data passed through the LSTM-RNN consists only of time-series IMU data (e.g., acceleration (x, y, z) and gyroscope (x, y, z)) from a personal care device, such as device (105), and the reference data consists without limitation of sensor data, such as time-series force data, such as force applied to the head from the force sensor of the personal care device. If an acceptable % error and / or confidence level is achieved for force using only IMU data, the personal care device, such as device (105), may exclude, disable, and / or not require the force sensor to provide usage features (e.g., cumulative force, etc.). The predicted usage features of the force can then be used to predict other usage features and / or to enhance the user experience, including providing recommendations, controlling functions, supplements, etc.

[0055] In another example of training a model for predicting distance traveled, the input training data passed through the LSTM-RNN consists only of time-series IMU data (e.g., acceleration (x, y, z) and gyroscope (x, y, z)) from a personal care device, such as device (105), and the reference data consists of corresponding time-series data obtained using a 6-degrees-of-freedom (6DoF) tracking solution combined with the personal care device (e.g. via a serial connection). The tracking solution can be converted into actual distance traveled measurements. In one example for 146 seconds of time-series data, the actual distance is 4178.22 cm and the predicted distance is 4252.33 cm, with an error percentage of 1.77%. In particular, the trained machine learning model can provide an average error of 2.79% for the prediction of the total distance traveled by the personal care device. The usage characteristics of the distance traveled can then be used to enhance the user experience, including providing recommendations, control functions, supplements, etc., by predicting other usage characteristics and / or delivering production input data through deployed trained machine learning models.

[0056] In another example of training a model for predicting work performed by a personal care device, the input training data passed through an LSTM-RNN consists only of time-series IMU data from a personal care device, such as device (105), for example (e.g., acceleration (x, y, z) and gyroscope (x, y, z)), and the reference data consists of sensor data, such as time-series force data, such as the force applied to the head from the force sensor of the personal care device, and corresponding time-series data obtained using a 6-degrees-of-freedom (6DoF) tracking solution combined therewith (e.g. via serial connection). However, the work performed by the personal care device may be determined using an arithmetic logic unit (ALU) or another processor configured to independently multiply the predicted force and the predicted distance traveled. The predicted use characteristics of the work can then be used to predict other use characteristics, including providing recommendations, control functions, supplements, etc., by passing the production input data through the deployed trained machine learning model, and to improve the user experience.

[0057] In particular, data preparation may involve clock synchronization between sensor data and reference data using corresponding signals from two data streams for each data channel (e.g., IMU z-acceleration data and the corresponding reference z-position data from the tracking device, etc.). For example, the signal of the IMU data can be matched with the signal of the reference data to eliminate temporal offsets and align the two data sets. In one example, the same time offset used to synchronize one channel is applied to all data channels of the same data set. Thus, the clock can be synchronized between data sets (e.g., for each of the 100 shaves) by eliminating offsets between the data sets. However, each sensor platform may utilize its own internal time clock synchronized with the same external computing platform operating system. This type of synchronization can render a relatively higher but acceptable error rate for the distance traveled from a model configured to use IMU data as input for distance traveled prediction (e.g., during shaving, etc.) by passing the IMU data as production input data through deployed trained machine learning. Additionally or alternatively, inactive periods longer than a predetermined period (e.g., 5 seconds, etc.) in the reference data can be removed, and the remaining active periods can be split into subsets. Similar to clock synchronization, removal and splitting can be applied to all data channels of the same data set.

[0058] Therefore, an algorithm for predicting arbitrary usage features can generally be substantially similar to the differences associated with the specific usage features intended to be predicted. For example, a general algorithm for predicting arbitrary usage features may include requirements definition, data collection, data preparation, model development (e.g., model selection, model training, model optimization), model evaluation, and / or model deployment. Meanwhile, differences in the general algorithm may include differences in data collection (e.g., different subsets of data used for training, such as 6DoF motion data not required to train a force measurement model), differences in data preparation (e.g., force may be refined over a single time step (10ms), while distance may be calculated in each nested window of 128 time steps (1280ms)), and differences in the model (e.g., the values ​​of selected hyperparameters (model structure) and learned parameters (weights and biases) may be unique to each model), etc.

[0059] Components of a system, such as a razor system, will now be described according to an exemplary application of the system. The razor system is illustrated for illustrative purposes, and it is understood that one or more aspects of the razor system may be implemented by any system (e.g., toothbrush system, etc.), device (e.g., personal care system, hub, server, IoT device, etc.), and / or method (e.g., feature prediction, etc.) according to the embodiments.

[0060] FIG. 2 illustrates an example of an apparatus (105) according to an aspect of the present disclosure. The apparatus (105) may be an example of an aspect of a corresponding element or element described with reference to FIG. 1, or may include such an example.

[0061] The device (105) may include a cartridge (205), a cartridge holding assembly (210), a rinse sensor lug (215), an articulated head (220), a finger guard (225), a shaver body (230), a head hinge (235), a force sensor (240), a silicone pad (245), a logo insert (250), a battery (255), a microcontroller unit (MCU) (260), Bluetooth or Wi-Fi (265), a charging and / or data transfer port (270) such as a USB or USB C port, a power supply (275), and an end cap (280). The force sensor (240), the battery (255), and the MCU (260) may be or include aspects of the corresponding elements or elements described with reference to FIG. 3.

[0062] In one embodiment, an interactive razor such as the device (105) of FIG. 1 may have a plurality of sensors, and some of the sensors may be removed. For example, the force sensor (240) may be removed, and the device (105) may still be used to accurately predict the number of strokes and classify the stroke direction (e.g., horizontal, vertical upward, vertical downward, etc.). In one example, a model structure trained such that input production data passed through an LSTM-RNN (e.g., while the user is shaving) consists only of time-series IMU data (e.g., acceleration (x, y, z) and gyroscope (x, y, z)) to accurately predict the force may be selected for force prediction. As illustrated in FIG. 2 and FIG. 3, in one embodiment, the capacitive sensor (325) may be removed, and the device (105) may still be removed to accurately classify the rinse. In another embodiment, the shaving length may be determined from the IMU sensor (320) data. For example, the IMU sensor (320) may capture and / or provide acceleration and time data to calculate the shave length. In one example, a model structure trained for distance prediction may be selected such that it consists only of time-series IMU data (e.g., acceleration (x, y, z) and gyroscope (x, y, z)) to accurately predict the distance traveled by input production data (e.g., while the user is shaving) passed through the LSTM-RNN. In some embodiments, the shave length may be used as data for predicting the end of life. In another example, a model structure trained for work prediction may be selected such that it consists only of time-series IMU data (e.g., acceleration (x, y, z) and gyroscope (x, y, z)) to accurately predict the work done by input production data passed through the LSTM-RNN. However, the work done may be determined using an ALU or another processor configured to independently multiply the predicted force and the predicted distance traveled.

[0063] Accordingly, FIG. 2 illustrates the components of a razor according to an exemplary embodiment of the concept of the present invention. The razor of FIG. 2 is an exemplary implementation. That is, the razor may include more or fewer components than those shown in FIG. 2.

[0064] FIG. 3 illustrates an example of interactive razor hardware (310) according to aspects of the present disclosure. The illustrated example includes a programmer (300), interactive razor analysis software (305), and interactive razor hardware (310). It is understood that the razor is hardware illustrated for illustrative purposes, and that one or more aspects of the razor hardware may be implemented by any system (e.g., toothbrush system), device (e.g., personal care device, hub, server, IoT device, etc.), and / or method (e.g., feature prediction, etc.) according to the embodiment.

[0065] The interactive shaver hardware (310) may include a sensor input (315), a controller (350), a power manager (375), an external flash (392), and an RGB LED (394). The sensor input (315) may include an inertial measurement unit (IMU) (320), a capacitive sensor (325), a force (330), a rinse (335), a grip (340), and an NTC thermistor (345). The controller (350) may include an MCU (355), an RTC (360), a BLE radio (365), and an internal flash (370). The MCU (355) may be an example of an aspect of the corresponding element or elements described with reference to FIG. 2, or may include such an element.

[0066] The power management unit (375) may include a USB (380), a battery management unit (382), a battery (384), a source selection (386), a regulator (388), and a system power bus (390). The battery (384) may be an example of an aspect of the corresponding element or elements described with reference to FIG. 2, or may include such an element.

[0067] As illustrated in FIGS. 2 and 3, according to some embodiments, the force sensor (240) may be removed, and the device (105) of FIG. 1 may still be used, for example, to predict the number of strokes and classify the stroke direction (e.g., horizontal, vertically upward, vertically downward, etc.). The capacitive sensor (325) may be removed, and the device (105) of FIG. 1 may still be used, for example, to classify the rinse. In other embodiments, the shave length may be determined from IMU sensor (320) data. For example, the IMU sensor (320) may capture and / or provide acceleration and time data to calculate the shave length. In some embodiments, the shave length may be used as a prediction data for end-of-life.

[0068] Accordingly, FIG. 3 illustrates hardware for a shaver according to an exemplary embodiment of the concept of the present invention. As illustrated in FIG. 3, the hardware of the shaver includes, among other things, a sensor input unit (315), a controller (350) which may be a 2.4GHz wireless microcontroller supporting BLE (e.g., CC2640R2), and a power management unit (375).

[0069] As illustrated, the sensor input (315) may include, but is not limited to, an inertial measurement unit (IMU) (320), a capacitance-to-digital converter, a force sensor (330), a rinse sensor (335), a grip sensor (340), and a temperature sensor (e.g., an NTC thermistor (345)). The IMU (320) may include, for example, an accelerometer, a gyroscope, and / or a magnetometer. An IMU such as the IMU (320) may include at least one accelerometer per axis for three axes and / or at least one gyroscope per axis for three axes. For example, the IMU (320) may include six microelectromechanical systems (MEMS) modules, one of which corresponds to one accelerometer or gyroscope per axis (e.g., x, y, z axes, vertical axis, orthogonal axis, etc.).

[0070] Preferably, an IMU without a magnetometer includes a deactivated magnetometer and / or a magnetometer that is filtered so that data is ignored or otherwise does not reach an analog-to-digital converter, hardware registers, memory, storage device, processor, machine learning model to be trained or trained, or a combination thereof. In one example, the filter may include a bandpass filter. Additionally, the IMU may include a processor physically coupled to an accelerometer, a gyroscope, and / or a magnetometer. For example, the IMU (320) may be a module having a processor physically coupled to an accelerometer and a gyroscope (e.g., via a data connection, via an analog-to-digital converter, etc.). Additional sensors for using a shaver according to an exemplary embodiment of the concept of the present invention may include a resistive contact, a piezo-contact microphone, a Hall effect sensor, an external camera, a strain gauge, a load cell, a reed switch, a linear variable differential transformer, etc.

[0071] The wireless microcontroller may include a main central processing unit (e.g., MCU 355), a BLE radio (365), and internal flash memory (370), as well as other peripheral devices and modules (e.g., a real-time clock (RTC (360))). For example, the CC2640R2 SimpleLink™ BT microcontroller includes, among other features, a microcontroller (e.g., having a central processing unit, non-volatile memory, random access memory (RAM), cache memory), an ultra-low power sensor controller (e.g., a sensor controller engine, an analog-to-digital converter, RAM, etc.), and a radio frequency (RF) section (e.g., a transceiver, an RF interface, etc.). In one example, the MCU (355) may be physically coupled to the accelerometer and gyroscope of the IMU (320) through a connection between the controller (320) and the IMU (320). The wireless microcontroller can be connected to an external flash device (392) and a light-emitting element (e.g., a status RGB LED (394)). The wireless microcontroller can also be connected to a programmer (300) and analysis software (305).

[0072] When the previously described razor is used for shaving, raw sensor data may be collected. The raw data may be stored in the razor's non-volatile memory. In the implementation illustrated in FIG. 3, the raw data may be downloaded to a desktop application via a serial connection. However, as previously described, the raw data may be sent directly to a hub. The raw data transmitted between the razor and the hub may be encrypted. The raw data captured by the sensor includes, but is not limited to, time in seconds (e.g., UTC-0), IMU data (e.g., acceleration (x, y, z) and gyroscope (x, y, z)), force applied to the head, capacitance of the metal inserted in the head, case temperature, gravity axis (e.g., approximate orientation of the razor), and force sensing register (FSR) correction values.

[0073] When the razor described above is used for shaving in production, or more generally when a personal care device is used to perform its function outside of a training process (e.g., one or more cores, central processing units, graphics processing units, combinations thereof, etc.), it may be configured to run a machine learning model trained to predict usage features. For example, the processor of the IMU (320) and / or MCU (355) may be configured to run a trained LSTM-RNN selected to predict force such that input production data (e.g., while the user is shaving) passing through the LSTM-RNN consists only of time-series IMU data (e.g., acceleration (x, y, z) and gyroscope (x, y, z)). Additionally or alternatively, the processor of the IMU (320) and / or MCU (355) is configured to run a trained LSTM-RNN selected to predict the distance traveled, and the input sensor data passed through the LSTM-RNN (e.g., while a user is shaving) consists only of time-series IMU data (e.g., acceleration (x, y, z) and gyroscope (x, y, z)) to accurately predict the distance traveled. Additionally or alternatively, the processor of the IMU (320) and / or MCU (355) is configured to run a trained LSTM-RNN selected to predict the work performed by the razor, and the input sensor data passed through the LSTM-RNN (e.g., while a user is shaving) consists only of time-series IMU data (e.g., acceleration (x, y, z) and gyroscope (x, y, z)) to accurately predict the work. However, the processor of the IMU (320) and / or MCU (355) may be configured to independently calculate work based on the predicted force and the distance traveled (e.g., work = force * distance).

[0074] Similarly, the processor of the IMU (320) and / or MCU (355) may be configured to perform a transformation using the predicted usage features (e.g., predicted force, predicted distance, etc.) to predict other usage features related to the personal care device, such as stroke force (e.g., force generated in individual strokes, force generated in total strokes during shaving, etc.), stroke distance (e.g., distance traveled in individual strokes, distance traveled in total strokes during shaving, etc.), stroke count (e.g., number of strokes during shaving, total number of strokes during operation, etc.), rinse events (e.g., individual rinses of the device, number of rinse events during shaving, total number of rinse events during operation, etc.), end of life (e.g., prediction of when one or more components remain in an operable state, prediction of when replacement is recommended based on a threshold, etc.), or a combination thereof. The transformation may be achieved using a function (e.g., a database, a matrix, a pointer, a mathematical function, etc.) that associates the predicted usage features with other usage features to be predicted, such as stroke counts or end of life values. The processor of the IMU (320) and / or MCU (355) may also run an additional machine learning model trained to predict one or more other usage features based on one or more usage features, such as but not limited to when a model trained to predict stroke count or end of life based on predicted force, predicted distance, or a combination thereof is run.However, the processor of the IMU (320) and / or MCU (355) may be configured to run an additional machine learning model trained to predict one or more other usage features based on input production data (e.g., while the user is shaving) passed through a trained LSTM-RNN composed solely of time-series IMU data (e.g., acceleration (x, y, z) and gyroscope (x, y, z)) to accurately predict other features (e.g., stroke count, end of life, etc.).

[0075] Predicted usage features can then be used to provide feedback such as control functions for personal care devices (e.g., fluid delivery, pulse, etc.), use of recommended specifications (e.g., user guidance, etc.), control functions for IoT devices near the personal care device (e.g., faucets, lighting from fixtures, etc.), replenishment functions (e.g., automatic order placement, etc.), replacement information (e.g., cartridge usage level, etc.), or a combination thereof. For example, predicted usage features can be used to change the water flow or temperature from a faucet via an IoT faucet, deliver recommendations related to the use of the personal care device to the user (e.g., stroke recommendations via personal care devices, IoT devices such as smart mirrors, mobile devices, etc.), automatically replenish disposable components of the personal care device (e.g., ordering replacement cartridges, etc.), and notify the user of the predicted end of life of disposable components so that the user can change the disposable components.

[0076] Accordingly, for example, the processor of the IMU (320) and / or MCU (355) may be configured to initiate control commands for controlling elements of the personal care device. In one example, the control commands include an interrupt for controlling the operation of the elements, an enable signal for enabling the operation of the elements, a disable signal for disabling the operation of the elements, or a combination thereof. For example, the shaving device may include a cartridge having a razor blade and optionally a shaving aid composition. Additionally, the shaving device may include a dispenser that receives a personal care composition (e.g., shaving fluid, etc.) that may be optionally released to enhance the shaving experience. In this regard, the processor of the IMU (320) and / or MCU (355) may be configured to initiate control commands to a controller coupled to the cartridge (e.g., command generation, command delivery, making a call for command generation or delivery, etc.) to modify the manner in which one or more components of the cartridge come into contact with the user (e.g., blade contact, etc.) based on predicted usage characteristics (e.g., force, distance traveled, stroke, number of strokes, etc.). The processor of the IMU (320) and / or MCU (355) may also be configured to control the dispenser by initiating control commands to a controller coupled to the dispenser to selectively apply a personal care composition during shaving based on predicted usage characteristics (e.g., force, distance traveled, stroke, etc.).

[0077] A processor (e.g., one or more cores, a central processing unit, a graphics processing unit, etc., or a combination thereof) coupled to one or more sensors (e.g., an accelerometer, a gyroscope, etc.) may also be optionally coupled to an RF interface for raw data, prediction and / or feedback communication, a memory or storage device for storing raw data, prediction and / or feedback, a sensor controller that controls feedback based on prediction through an input / output device (e.g., a display, a speaker, a haptic device, a switch, etc.), or a combination thereof. In this regard, traditional communication connections include, but are not limited to, connections on a circuit board, serial connections between hardware modules, local area network connections, wide area network connections, or a combination thereof. Meanwhile, feedback signals may include, but are not limited to, interrupts, enable signals, disable signals, machine code instructions (e.g., assembly code, etc.), input / output device instructions, or a combination thereof.

[0078] Additionally, although the processor coupled to the accelerometer and / or gyroscope (or IMU) has been described as being physically coupled to the personal care device via a communication connection, the processor configured to execute one or more aspects of the embodiment (e.g., a machine learning model trained to predict usage features related to the personal care device, data collection, model training, feedback, etc., or a combination thereof) may be located on one or more other devices, such as a hub, a server, an IoT device, a mobile computing platform, or a combination thereof. Thus, for example, the processor may be communicably coupled to a set of sensors (e.g., accelerometer, gyroscope, etc.) on the personal care device via a connection between the hub and the personal care device. In another example, the processor may also be located on a server communicably coupled to the personal care device and / or a hub adjacent to the personal care device.

[0079] FIG. 4 illustrates the structure of an IoT shaver system according to an exemplary embodiment of the concept of the present invention. The structure illustrated in FIG. 4 is provided as an example, and embodiments of the concept of the present invention are not limited thereto.

[0080] As illustrated in FIG. 4, the IoT shaver system includes a consumer component (410), a hub (420), and a cloud component (430). The consumer component (410) includes a smart shaver and a mobile device. It is understood that the consumer component (410) may include additional devices equipped with wireless communication technology, such as a smart mirror, a smart faucet or shower head, etc. The smart shaver and the mobile device (e.g., a smartphone) may communicate wirelessly with each other via Bluetooth or Low Energy Bluetooth (BLE). It should be understood that various other short-range wireless communication technologies may be used by the smart shaver and the mobile device, including but not limited to ZigBee, Near Field Communication (NFC), Li-Fi, Ultra-Wideband (UWB), Radio Frequency Identification (RFID), etc. The smart shaver may correspond to one of the shavers described above with reference to FIG. 1 through 3.

[0081] The smart shaver wirelessly provides raw data to the hub (420) via, for example, Bluetooth or BLE. As described above, the raw data may be provided via a number of other short-range wireless communications. The raw data transmitted from the shaver to the hub (420) is data collected from the shaver's sensors. The sensors and raw data will be described in detail later. The hub (420) stores the raw data and transmits the raw data to the cloud component (430). For example, the hub (420) may transmit the raw data to the cloud component (430) via a dedicated and secure IP protocol. More specifically, the raw data may be transmitted to a dedicated IP address of a server in the cloud (e.g., a cloud computing network) via a long-range connection (e.g., a cellular connection, a fiber optic connection, a copper connection, etc.).

[0082] More specifically, in one embodiment, raw data is provided directly from the smart shaver to the hub (420), and the hub (420) provides the raw data directly to a server in the cloud. In this embodiment, a mobile device is not used as an intermediary for transmitting raw data between the hub (420) and the server. Also, a mobile device is not used as an intermediary for raw data communication between the smart shaver and the hub (420). As will be mentioned below, in such an embodiment, the mobile device is used only to receive information from the server. The mobile device does not receive raw shaver data.

[0083] In the cloud (430), the Platform as a Service (PaaS) provides an application gateway, e.g., a cloud gateway / IoT hub of FIG. 4, which receives raw data and preprocesses the raw data before inputting it into a neural network model, e.g., an ANN (255) of FIG. 1. The model used may correspond to one or more machine learning models described herein. In stream preprocessing, force and capacitive sensor data may be passed through a Butterworth filter to flatten the frequency response. Additionally, the input to the neural network may be normalized to increase the neural network speed. Additionally, the filtered force sensor values ​​may be reduced to mitigate the exploding gradient during training.

[0084] The model uses raw data (which may be preprocessed data) to predict shaving features. In one example, a neural network model uses preprocessed data to predict the effective lifespan of a razor cartridge. For instance, if filtered, time-series sensor data is used as input, the neural network model generates an output indicating the likelihood that the razor will reach the end of its lifespan. An example of a neural network model will be described in detail later.

[0085] The output of the neural network model is stored in the data store of FIG. 4, which may include a database, e.g., a hierarchical database, a relational database, an object-oriented database, etc. In one embodiment, the data store includes an Azure SQL database. The data output from the model may be used by the exemplified application service (app service) to create, deploy, and / or use an application to provide a replacement order for a new razor cartridge. The output of the model may also be used by the application service to create, deploy, and / or use an application to provide other feedback to the user. Although not illustrated, other service platforms or computing devices accessing the data store and / or application data (app data) may provide services (e.g., replacement services) and / or information (e.g., feedback) based on the predicted features. In addition, enterprise information gathering and / or application tools (Power Apps / BI) within a business analytics and / or subscription service platform (e.g., Microsoft 365) can build and / or use applications by utilizing the data output of the model, and enterprise information gathering and / or data mining tools can perform tasks such as gaining insights from data, making decisions, and generating reports.

[0086] Application services may facilitate communication with consumer components (410) to provide information to the user, for example. Personal care or grooming tools, devices, or tools, such as a razor, may receive information that a new cartridge is being shipped, for example, through a software application on a mobile device. For example, a display may provide a graphical user interface (GUI) that allows the user to view notifications such as one or more recommendation(s), replacement order confirmation(s) and / or acceptance(s), order status, end-of-life message(s) (e.g., predicted end date, usage level, etc.), or other usage information regarding the device or user. However, any other consumer components and / or hubs (420) may communicate with any service platform to receive feedback via a GUI, diode, and / or haptic device configured to provide information regarding replenishment, feedback, etc.

[0087] Therefore, while it has been described that the model is used to generate replacement orders for razor cartridges, it should be understood that the model can be used in various other applications. For example, a neural network model can utilize the provided data to identify usage habits (such as shaving habits) that can lead to personalized recommendations. More specifically, the neural network can be used to generate product recommendations, product usage adjustments, estimated numbers of remaining quality usage (such as shaving), etc. Thus, the provided services and / or feedback may relate to any other personal grooming process based on, for example, brushing teeth, drying hair, and / or the type of device used (e.g., toothbrush, etc.), training, sensor configuration, input data, use of parameters / features / metadata, etc., or a combination thereof. These additional features and other aspects of FIG. 4 will be discussed in detail later. Furthermore, the illustrated PaaS is one of the non-limiting embodiments of cloud infrastructure. Thus, one or more other infrastructures, such as Infrastructure as a Service (IaaS), Software as a Service (SaaS), etc., may be additionally or alternatively implemented. Similarly, one or more aspects of the cloud (430) may additionally or alternatively be implemented in a non-cloud computer network, such as, for example, an on-premises or in-house computer network.

[0088] FIG. 5 illustrates a hub (510) according to an exemplary embodiment of the concept of the present invention. As shown in FIG. 5, the hub (510) includes a short-range wireless communication module (512), a Wi-Fi module (514), a microcontroller (516), a memory (518), and a power supply (520).

[0089] The short-range wireless communication (512) may employ Bluetooth or BLE. It should be understood that various other short-range wireless communication technologies, including but not limited to Zigbee, NFC, UWB, RIFD, etc., may be employed by the hub (510). The short-range wireless communication module (512) enables the hub (510) to communicate with other peripheral devices such as a smart shaver (502), a mobile device (504), or other wireless communication devices (506) such as a smart mirror, a smart faucet, a shower head, a light, a thermostat, etc. According to an exemplary embodiment of the concept of the present invention, communication through the smart device (e.g., the shaver (502)) and the hub (510) may be encrypted. For example, the smart device (such as the shaver (502)) and the hub (510) may be paired using Secure Simple Pairing (SSP) to establish a BLE link, and once paired, encrypted raw data may be transmitted over the BLE link.

[0090] Data transmitted from a device (such as a shaver (502)) to a hub (510) can be stored in the hub's memory (518). The hub (510) can receive data in real time from the device (such as a shaver (502)) while the user is shaving. The transmission of data may be in response to a request from the hub (510). For example, the microcontroller (516) of the hub (510) may be configured to periodically request the device (such as a shaver (502)) to transmit raw data stored in the onboard memory. The periodic data requests may be based on the time of day (TOD), etc.

[0091] A Wi-Fi module (514) enables, for example, the hub (510) to communicate with a wireless home network (530). It should be understood that the hub (510) may include any other communication interface that enables communication with the network. For example, the hub (510) may include a module that enables power line communication (PLC).

[0092] The hub (510) can transmit raw data to a server in the cloud (534) via a wireless router (532). Data can be transmitted to the cloud (534) in real time when received from a device (such as a razor). Additionally, the hub (510) can queue data and transmit it to the cloud (534) after a predetermined time, after a predetermined amount of data has been collected, at a specific time of day, etc. This transmission time can be set by the user through an application on a mobile device. Data transmission may be a response to a data request from the cloud server. This data request can be transmitted after a predetermined time, after a predetermined amount of data has been collected, at a specific time of day, etc.

[0093] Communication between the hub (510) and the cloud server is secure. For example, all raw data provided from the hub (510) to the server can be encrypted. Additionally, any communication from the server to the hub (510) can be encrypted. More specifically, raw data can be transmitted through a secure channel existing between the hub (510) and the cloud server. The secure channel can use, for example, one of the Enhanced Message Queuing Protocol (AMQP), Message Queuing Telemeter Transport (MQTT), or Hypertext Transfer Protocol Secure (HTTPS). Simple Authentication and Security Layer (SASL) can be used for authentication and data security between the hub (510) and the cloud server.

[0094] The hub (510) may include its own power source, such as a battery that is rechargeable or non-rechargeable. Additionally, the hub (510) may include a USB connector for power, or the hub may utilize inductive charging. The USB connector may serve as a data storage device. The hub (510) may also include a plug that plugs into a power outlet. As illustrated in FIG. 5, the hub (510) is a standalone device similar to an Ethernet bridge. The hub (510) can be plugged into a power outlet, for example, so that a user can take the hub (510) with them while traveling. In one example, the hub (510) may have a substantially cube-shaped wall adapter form factor with a built-in power prong. The hub (510) may also be configured as a receptacle device for a tool, such as a shaving tool. For example, the hub (510) may be a dock, a stand, or a cradle. In this case, the hub (510) can be used to charge the device (such as the razor (502)), clean or sterilize the device (such as the razor (502)), as well as evaluate defects and / or deterioration of the device.

[0095] FIG. 6 illustrates an example of a feedback server (600) for a personal care system according to aspects of the present disclosure.

[0096] The server (600) is an example of an aspect of the corresponding element or elements described with reference to FIGS. 1 and FIG. 8, or may include such an example. The server (600) may include a processor component (605), a memory component (610), a data collection component (615), a feedback prediction component (620), a feedback component (625), and a training component (630).

[0097] A processor component (605) may be an example of an aspect of the corresponding element or elements described with reference to FIG. 8, or may include such an example. The processor may include intelligent hardware devices (e.g., a general-purpose processing component, a digital signal processor (DSP), a central processing unit (CPU), a graphics processing unit (GPU), a microcontroller, an application-specific integrated unit (ASIC), a field programmable gate array (FPGA), a programmable logic device, a discrete gate or transistor logic component, a discrete hardware component, or any combination thereof). In some cases, the processor may be configured to operate a memory array using a memory controller. In other cases, the memory controller may be integrated into the processor. The processor may be configured to execute computer-readable instructions stored in memory to perform various functions. In some examples, the processor may include special-purpose components for modem processing, baseband processing, digital signal processing, or transmission processing. In some examples, the processor may include a system-on-a-chip.

[0098] The memory component (610) may be an example of an aspect of the corresponding element or elements described with reference to FIG. 8, or may include such an example. Computer memory may store information for various programs and applications of a computing device. For example, the storage device may contain data for running an operating system. Memory may include both volatile memory and non-volatile memory. Volatile memory may include random access memory (RAM), and non-volatile memory may include read-only memory (ROM), flash memory, electrically erasable programmable read-only memory (EEPROM), digital tape, hard disk drive (HDD), and solid-state drive (SSD). Memory may include any combination of readable and / or writable volatile memory and / or non-volatile memory together with other storable storage devices.

[0099] The data collection component (615) can identify data from a shaver (e.g., one example of the device (105) of FIG. 1) that includes a set of sensors. The data collection component (615) may be an example of an aspect of the corresponding element or elements described with reference to FIG. 8, or may include such an element. Thus, in some examples, the set of sensors may include an IMU (320), a capacitance-to-digital converter, a force sensor (240), a rinse sensor (335), a grip sensor (340), a temperature sensor, an accelerometer, a gyroscope, a magnetometer, a resistance contact, a piezo contact microphone, a Hall effect sensor, an external camera, a strain gauge, a load cell, a reed switch, a linear variable differential transformer, or any combination thereof. The data collection component (615) may also receive data via a wireless communication connection to the shaver.

[0100] In one embodiment, the data collection component (615) collects data from a set of sensors on a device (e.g., a razor) such as the device (105) of FIG. 1. In some embodiments, the data collection component (615) may identify and collect other data through means such as user participation, surveys, personal subjective opinions, etc. In one example, a participant shaves periodically with the same cartridge using a razor (e.g., one example of the device (105) of FIG. 1) until the cartridge no longer provides an effective shaving experience. After each shave, the participant may answer survey questions regarding subjective opinions on the shaving experience. Thus, the data collected by the data collection component (615) may include sensor data from the device (105) shown in FIG. 1 (e.g., 60 Hz, 120 Hz, etc.) and / or hedonic survey data regarding the shaving experience (which may also be referred to as SIMS data). Additionally, the life shaving test may include two separate attempts (e.g., A and B). Trial A may be limited to 30 days, and Trial B may not be limited to a specific period. Both trials may include 26 unique participants, and the same participants participate in each trial.

[0101] In one embodiment, the data collection component (615) collects data from a set of sensors on a personal care device other than a razor (e.g., the device (105) of FIG. 1). For example, a participant may brush their teeth or perform personal grooming activities until the device indicates that it no longer provides an effective experience. The participant may answer survey questions after each session regarding subjective opinions about the experience.

[0102] In one example, ground measurement data may be collected from a razor (or other shaving tool) having a plurality of sensors (e.g., accelerometer, gyroscope, magnetometer, image capture device, Hall effect sensor, piezoelectric sensor, etc.). Ground measurement data may be captured by the sensors of the razor when a user, such as the user (100) of FIG. 1, uses the shaving tool. Ground measurement data may be data collected from the sensors. Thus, ground measurement data may include information that is not perceptible or detectable to the naked eye (e.g., minute movements, minute differences in each type of movement, various combinations of movement types, various movements captured simultaneously in real time, etc.). In another embodiment, ground measurement data may include a data set having data from an accelerometer, gyroscope, Hall effect sensor, etc.

[0103] In one embodiment of the present disclosure, the data collection component (615) may include data cleanup, integrity check (e.g., garbage in / garbage out), feature engineering, and extraction. The input data may be several subsets of filtered time-series sensor data. In another embodiment, the learning model may be developed through the creation of a neural network, where the input data is derived sensor data features (e.g., accumulated force, distance traveled, time of force received) and the output data is the likelihood of the razor's lifespan ending.

[0104] In some examples, the processed data may include the shaving date, shaving period, number of strokes (from force and IMU sensors), number of rinses (from head capacitance sensor), total rinse time, ambient temperature, minimum, maximum, and average force, number of strokes in each orientation (e.g., vertical upward, vertical downward, and horizontal), and a force histogram for the strokes. Additional processed data may include strokes (e.g., start index, end index, maximum force, gravity axis), analysis graphs, and filtered data.

[0105] In some embodiments, the processed data may include information related to personal grooming activities other than shaving, such as date, period, number of strokes (from force and IMU sensors), number of rinses (from head capacitance sensors), total rinse time, ambient temperature, minimum, maximum, and average force, number of strokes in each orientation (e.g., vertical upward, vertical downward, and horizontal), and force histograms for strokes other than shaving strokes (e.g., brush strokes for a toothbrush).

[0106] Referring to feature engineering, potential secondary (derived) features may include determining the start and end of stroke or rinse events, shaving characteristics, average force per stroke, cumulative force applied over the razor life, cumulative usage time over the razor life, wall time between shaves, survey response delta (i.e., difference between respondents regarding normalization of survey data), shaving length or distance traveled, and determining the amount of work over the razor life.

[0107] In one embodiment of the present disclosure, data limited to this period may be used to determine the start and end of stroke or rinse events for a better matrix. In some cases, usage characteristics, such as shaving characteristics, may include the number of strokes, stroke direction (e.g., horizontal, vertically upward, vertically downward), and the number of rinses. In one embodiment, responder survey response deltas may indicate device status, such as razor status, by analyzing the end-user experience. For example, if the quality of use of the device (e.g., shaving) decreases by 1 point for two consecutive times, the next use (e.g., shaving) is the last use (e.g., shaving). Additionally, if the quality of shaving suddenly decreases by 3 points, the razor will not operate. In one embodiment, the amount of work during the life of the razor may be calculated as Work = Force * Distance.

[0108] In some embodiments, the statistical correlation results may include secondary features extracted from the described data set. For example, the extracted secondary features may include average values ​​across all sensors per use, such as the deviation from the panelist's general sensor data per shave, the accumulated shaving time, and the accumulated force.

[0109] The feedback prediction component (620) can predict one or more shaving features based at least partially on input data from a razor (e.g., one example of the device (105) of FIG. 1) using a machine learning model trained using at least data from a set of sensors from the razor (e.g., one example of the device (105) of FIG. 1). For example, the input data may be taken from a razor configured to provide sensor data from a subset of available sensors, and / or from an IMU from the razor configured to provide a subset of sensor data only from the available sensor data and / or from a razor configured to provide sensor data from a subset of available sensors. The data used to train the machine learning model may include, for example, at least one additional sensor data than the input data (e.g., data from more sensors than the subset of available sensors, a larger subset of sensor data, and / or data from an IMU providing more data than the trained IMU, etc.). The IMU of a ready-made razor may also provide all raw sensor data to predict shaving features. In some embodiments, the server (600) can predict shaving features including the number of shaving strokes, shaving quality information, shaving life end information, or a combination thereof.

[0110] The feedback prediction component (620) may also predict one or more features related to grooming activities other than shaving based on input data from the device using a machine learning model trained using at least data from a plurality of sensors. For example, the input data may be taken from a device configured to provide sensor data from a subset of available sensors, such as the available sensor data and / or a subset of sensor data from only the IMU. The data used to train the machine learning model may include, for example, at least one additional sensor data in addition to the input data. In some embodiments, the server (600) may predict features including stroke count, quality information, end-of-life information, or a combination thereof.

[0111] The feedback component (625) may provide usage feedback, such as shaving feedback, based on one or more features of the device or usage, such as shaving features. In some examples, the feedback, such as shaving feedback, includes instructions for quality improvement, such as shaving quality. In some examples, the feedback, such as shaving feedback, includes replacement feedback for at least a limited life component of the device, such as razor (or razor cartridge) replacement feedback.

[0112] The feedback component (625) may provide feedback, such as shaving feedback, based on a machine learning model trained using data from a subset of sensors of a device (e.g., a razor). In one embodiment of the present disclosure, the feedback component (625) may include output data indicating the possibility of the end of life of a device, such as the end of life of a device (105), such as the device (105) used by the user (100) in FIG. 1 for shaving periodically.

[0113] The training component (630) may be an example of an aspect of the corresponding element or elements described with reference to FIG. 8, or may include such an example. In some examples, the machine learning model is trained using survey data collected from users, and the survey data includes shaving quality information, razor end-of-life information, or both. In one embodiment, the training component (630) may include a machine learning model such as the ANN (125) shown in FIG. 1. The ANN (125) can predict device end-of-life information. In another embodiment, the learning model may be developed through the generation of the ANN (125), where the input data is derived sensor data features (e.g., accumulated force, distance traveled, time of force received), and the output data is the probability of the razor reaching the end of its life.

[0114] In some examples, the machine learning model is trained using survey data collected from users, and the survey data includes quality information regarding grooming activities other than shaving, end-of-life information, or both. The ANN (125) can predict device end-of-life information. In other embodiments, the learning model may be developed through the generation of the ANN (125), where the input data is derived sensor data features (e.g., accumulated force, distance traveled, time of force applied), and the output data is the probability of end-of-life.

[0115] In another embodiment, the learning model can be developed by generating an ANN (125) as shown in FIG. 1, the input data is induced sensor data (e.g., accumulated force, distance traveled, time of receiving force), and the output data is the possibility of end of life (e.g., end of shaver life).

[0116] In one embodiment, ground actual data may be input into a machine learning (ML) algorithm to generate shaving implementation metadata. The ML algorithm may determine a portion of ground actual data that provides a specific confidence interval (CI) to accurately determine metadata for implementation usage parameters (e.g., shaving features). This may be an iterative process until a predetermined CI is reached. Thus, in each iteration, the ML algorithm may determine whether the CI is satisfied for the implementation usage parameters. If the result satisfies the CI, the ML algorithm may proceed; if the result does not satisfy the CI, the ML algorithm may repeat the input step. In one embodiment, data from multiple sensors may be used to accurately determine a stroke. Ground actual data is input into the ML algorithm to determine a dataset (and thus a sensor set) capable of accurately determining when a stroke occurs. In one example, to verify whether the dataset (sensor set) can accurately determine a stroke, the output data is compared with the predicted data of the ML algorithm. If the dataset cannot accurately determine the stroke, the ML algorithm can be repeated until a CI greater than n% (e.g., greater than 90%) is reached. The ML algorithm can use more data. The output data of the ML algorithm that satisfies the CI is the razor metadata. In some cases, the machine learning model can be trained using additional data from one or more additional sensors not included in the razor.

[0117] In one embodiment, ground measurement data may be input into an ML algorithm to generate implementation metadata for a device other than a razor. The ML algorithm may determine which part of the ground measurement data provides a CI in order to accurately determine metadata for implementation usage parameters. This may be an iterative process until a predetermined CI is reached. Thus, in each iteration, the ML algorithm may determine whether the CI is satisfied for the implementation usage parameters. If the result satisfies the CI, the ML algorithm may proceed, and if the result does not satisfy the CI, the ML algorithm may repeat the input step.

[0118] FIG. 7 illustrates an example of a process for feedback machine learning for a personal care device according to aspects of the present disclosure. In some examples, such operation may be performed by a system comprising a processor that executes a set of code to control functional elements of the device. Additionally or alternatively, the processor may be performed using special-purpose hardware. Generally, such operation may be performed according to the methods and processes described according to aspects of the present disclosure. For example, the operation may consist of various subsets or be performed together with other operations described herein.

[0119] In operation (700), the server (600) can identify data from a device including a set of sensors on a device such as the device (105) of FIG. 1 (e.g., a razor). In some cases, the operation of this step may be performed by or with reference to a data collection component (615 or 815) as described with reference to FIG. 6 and FIG. 8, respectively.

[0120] According to an embodiment of the present disclosure, the data collection component (615) may include data preparation, cleanup and integrity checks (e.g., garbage in / garbage out), feature engineering, and extraction. A machine learning model, such as the ANN (125) of FIG. 1, may include creating a deep neural network model to predict force sensor outputs. Input data may include IMU sensor data and capacitive sensor data, while output data may include force sensor data. The recorded raw data may include time in seconds (UTC-0), IMU data (e.g., acceleration (x, y, z), gyroscope (x, y, z)), force applied to the head, capacitance of the metal inserted in the head, case temperature, gravity axis (e.g., approximate orientation of a device such as a razor), and FSR correction values. The raw data may include data from a 6-degrees-of-freedom (6DoF) tracking device coupled to a personal care device that can simultaneously capture data about the personal care device along with data captured from one or more sensors of the personal care device (e.g., data tracked from a user group, etc.).

[0121] In some embodiments, the sensor set may include an IMU (320), a capacitance-to-digital converter, a force sensor (240), a rinse sensor (335), a grip sensor (340), a temperature sensor, an accelerometer, a gyroscope, a magnetometer, a resistance contact, a piezo contact microphone, a Hall effect sensor, an external camera, a strain gauge, a load cell, a reed switch, a linear variable differential transformer, or any combination thereof.

[0122] In some embodiments, the processed data may also include the shaving date, shaving period, number of strokes (from force and IMU sensors), number of rinses (from head capacitance sensors), total rinse time, ambient temperature, minimum, maximum, and average force, number of strokes in each orientation (e.g., vertical upward, vertical downward, and horizontal), and a force histogram for the strokes. Additional processed data may include strokes (e.g., start index, end index, maximum force, gravity axis), analysis graphs, and filtered data.

[0123] In some embodiments, the processed data may also include information on grooming activities other than shaving, such as session date, duration, number of strokes (from force and IMU sensors), number of rinses (from head capacitance sensors), total rinse time, ambient temperature, minimum, maximum, and average force, number of strokes in each orientation (e.g., vertical upward, vertical downward, and horizontal), and a force histogram for the strokes. Additional processed data may include strokes (e.g., start index, end index, maximum force, gravity axis), analysis graphs, and filtered data.

[0124] In one embodiment, force and capacitive sensor data may be passed through a signal processing filter, such as a Butterworth filter, to flatten the frequency response. Input data to a neural network (e.g., ANN (125)) may be normalized near the mean of 0. This allows the neural network to proceed quickly. For example, filtered force sensor values ​​may be reduced by ×100 to mitigate the exploding gradient during training.

[0125] In operation (705), a server such as the server (600) of FIG. 6 can predict one or more features (e.g., shaving features) based on data using a machine learning model trained using data from a sensor set. In some cases, the operation of this step may be performed by referring to or by a feedback prediction component as described with reference to FIG. 6. In some embodiments, one or more features may include stroke count, quality information, end-of-life information, or a combination thereof.

[0126] In operation (710), a server such as the server (600) of FIG. 1 may provide feedback based on one or more features. In some cases, the operation of this step may be performed by referring to or by the feedback component (625) as described with reference to FIG. 6.

[0127] In one embodiment, a server such as the server (600) of FIG. 1 may provide feedback based on one or more features. In some examples, the feedback may include commands to improve the quality of a grooming session. The feedback may also include replacement feedback. The feedback may be based on a machine learning model trained using data from a subset of sensors. In one embodiment of the present disclosure, the shaving feedback may include output data indicating the end-of-life possibility of the device (105).

[0128] In some cases, a server such as the server (600) of FIG. 1 may provide feedback other than shaving. For example, the safety razor may have a power supply, a (unidirectional) short-range communication device (e.g., Bluetooth), and an IMU (e.g., a multi-axis accelerometer or gyroscope). Raw data related to the movement of the razor (e.g., speed, orientation, acceleration) may be transmitted to a hub. The hub may transmit this raw data to a processor of the described network. The processor has an algorithm (e.g., machine learning or deep learning) that processes the data to determine the type of motion. Movements include normal strokes (e.g., shaving strokes), sudden drops, the user placing the device (e.g., razor) down in a storage location between uses, or the user placing the device (e.g., razor) under a faucet or shower head for rinsing. In the last case, the network may send a return signal to the hub to sequentially operate the smart faucet or shower head during the period when the device (e.g., a shaver) is identified as being in a "rinse" state. The time-coupled faucet or shower head or flow rate data may be transmitted to an application on the user's smart device (e.g., phone, tablet, etc.) to educate the user about water consumption and suggest ways to reduce it. Some embodiments may utilize additional features of an Internet of Things (IoT) communication network.

[0129] Although independent blocks and / or a specific order are illustrated for illustrative purposes, it should be understood that one or more blocks of the method of FIG. 7 or components of the device of FIG. 6 may flow in combination, omission, bypass, rearrangement, and / or any order, or be located in any device or a single device. For example, one or more components illustrated in FIG. 6 and / or blocks illustrated in FIG. 7 may additionally or alternatively be located or implemented on a personal care device, a hub, a server, an IoT device, or a combination thereof.

[0130] FIG. 8 illustrates an example of a sensor reduction server (800) for a personal care system according to an aspect of the present disclosure. The server (800) may be an example of an aspect of the corresponding element or elements described with reference to FIG. 1 and 6, or may include such an example.

[0131] The server (800) may include a processor component (805), a memory component (810), a data collection component (815), a feature selection component (820), a sensor selection component (825), a feature prediction component (830), and a training component (835).

[0132] The server (800) may include an interaction device (e.g., a razor) deep learning model that predicts characteristics (e.g., shaving characteristics) while limiting sensors on the device (e.g., device (105) of FIG. 1). According to an embodiment of the present disclosure, the server (800) can still predict the end of life of the device or its components (e.g., cartridges) even after removing some sensors. For example, a razor (e.g., one example of device (105) of FIG. 1) may have a plurality of sensors, and some sensors may be removed.

[0133] In one embodiment, a force sensor such as the force sensor (240) of FIG. 2 may be removed and still able to predict the number of strokes and classify the stroke direction (e.g., horizontal, vertically upward, vertically downward, etc.). In one embodiment, a capacitive sensor such as the capacitive sensor (325) of FIG. 3 may be removed and still able to classify the rinse. In another embodiment, a usage period or length, such as the duration of shaving or the length of shaving, may be determined from IMU sensor data. For example, a user or learning model may use acceleration and time data to calculate a length, such as the length of shaving. In some embodiments, a length, such as the length of shaving, may be used as a predictor of end-of-life.

[0134] The processor component (805) may be an example of an aspect of the corresponding element or elements described with reference to FIG. 6, or may include such an example. The processor component (805) may be an example of an aspect of the corresponding element or elements described with reference to FIG. 8, or may include such an example. The processor may include intelligent hardware devices (e.g., general-purpose processing components, digital signal processors (DSPs), central processing units (CPUs), graphics processing units (GPUs), microcontrollers, application-specific integrated devices (ASICs), field programmable gate arrays (FPGAs), programmable logic devices, discrete gate or transistor logic components, discrete hardware components, or any combination thereof). In some cases, the processor may be configured to operate a memory array using a memory controller. In other cases, the memory controller may be integrated into the processor. The processor may be configured to execute computer-readable instructions stored in memory to perform various functions. In some examples, the processor may include special-purpose components for modem processing, baseband processing, digital signal processing, or transmission processing. In some examples, the processor may include a system-on-chip.

[0135] The memory component (810) may be an example of an aspect of the corresponding element or elements described with reference to FIG. 6, or may include such an example. The memory component (810) may be an example of an aspect of the corresponding element or elements described with reference to FIG. 8, or may include such an example. Computer memory may store information for various programs and applications of a computing device. For example, a storage device may contain data for running an operating system. Memory may include both volatile memory and non-volatile memory. Volatile memory may include random access memory (RAM), and non-volatile memory may include read-only memory (ROM), flash memory, electrically erasable programmable read-only memory (EEPROM), digital tape, hard disk drive (HDD), and solid-state drive (SSD). Memory may include any combination of readable and / or writable volatile memory and / or non-volatile memory together with other storable storage devices.

[0136] The data collection component (815) is an example of an aspect of the corresponding element or elements described with reference to FIG. 6, or may include such an example.

[0137] Some embodiments of the present disclosure may enable one or more sensors configured based on implementation configuration data to generate raw data. This process may vary depending on the configuration of the implementation. For example, if the configuration includes manufacturing a tool, such as a shaving tool, with only the minimum number of sensors necessary to accurately determine the shaving implementation metadata for the implementation usage parameters, a user, such as the user (100) of FIG. 1, will turn on the device to provide power to the physically installed special-purpose sensors (which may be a minimum number). If the configuration includes selectively enabling or disabling the total number of sensors of the shaving tool for the minimum number of sensors necessary to accurately determine the shaving implementation metadata, a user, such as the user (100) of FIG. 1, will selectively turn on or off the power for the minimum sensors. If the configuration includes selectively programming one or more sensors to capture only a subset of data that can be captured (e.g., only a subset of accelerometer data, only a subset of gyroscope data such as a subset of axis data, etc.), a user such as the user (100) of FIG. 1 will selectively program one or more sensors to capture only a subset of data that can be captured (i.e., conventionally capturing data that can be captured (which may be a subset)). However, in the example described above, the device may be configured to additionally or alternatively enable, disable, and / or self-program automatically based on a set of instructions executed by its processor and / or stored in its memory.

[0138] Implementation metadata generated from the ML algorithm can be determined and used at the back end rather than the front end of the process. For example, a ready-made razor with an IMU such as the IMU (320) of FIG. 3 to collect as much raw data as possible, or a family of sensors sufficient to collect the information required for the decision, may be used. The raw data may be transmitted to a hub and then to a communication network such as the communication network (110) of FIG. 1. The output from the communication network (e.g., a signal of a subset of data that can accurately determine when a stroke occurs) may be used to evaluate the raw sensor data to determine whether or how the razor tool is used (e.g., a stroke) to make recommendations (e.g., replenishment, change of product, etc.) and / or to automatically transmit the product.

[0139] The feature selection component (820) can identify one or more features, such as shaving features, based on the data. In some examples, one or more features, such as shaving features, include stroke count, stroke orientation, or both.

[0140] The sensor selection component (825) can identify a subset of sensors, at least one of the sensors is not included in the subset of sensors. In some examples, at least one of the sensors includes a force sensor.

[0141] Embodiments of the present disclosure may include shaving tool metadata for configuring a shaving tool. The configuration may occur in software, hardware, or a combination of both. In some cases, the configuration of a tool, such as a shaving tool, may include physically installing one or more special-purpose sensors. For example, operations such as using microcode to reconfigure the sensors, using configurable hardware such as a field programmable gate array (FPGA), or using voltage rails to selectively turn the power on or off for the components may be performed. In some examples, the configuration of a tool, such as a shaving tool, may include manufacturing the shaving tool with only the minimum number of sensors required to accurately determine the shaving implementation metadata for the implementation usage parameters. The configuration may include selectively enabling or disabling the total number of sensors on the tool, such as the shaving tool, for the minimum number of sensors required to accurately determine the implementation metadata, such as the shaving implementation metadata. The configuration may also include selectively programming one or more sensors to capture only a subset of data that can be captured (e.g., only a subset of accelerometer data, only a subset of gyro data such as a subset of axis data, etc.). The advantages of the above configuration may include hardware stack simplification, lower power consumption, lower processing usage, etc. In one embodiment of the present disclosure, the IMU may be physically installed in a shaving tool. The IMU may be trained by a machine learning model (e.g., a deep neural network) to provide information that can be used to accurately determine the stroke.

[0142] The feature prediction component (830) can train a machine learning model to predict one or more features, such as shaving features, using data from a subset of sensors from a device, such as a razor (e.g., one example of the device (105) of FIG. 1). In some examples, the machine learning model can predict one or more features, such as shaving features including the number of strokes, stroke orientation, or both.

[0143] The training component (835) may be an example of an aspect of the corresponding element or elements described with reference to FIG. 6, or may include such an example. In some embodiments, the machine learning model is trained using additional data from at least one additional sensor not included in the shaver. In some examples, the at least one additional sensor includes a force sensor. In additional examples, the at least one additional sensor is located in a 6-degrees-of-freedom (6DoF) tracking device coupled to the personal care device, which is capable of simultaneously capturing data for the personal care device having data captured from one or more sensors on the personal care device (e.g., tracking data from a user group, etc.).

[0144] In one embodiment, ground measurement data may be input into an ML algorithm to determine a subset of data (and thus a subset of sensors) that can accurately determine when a stroke occurred. In one example, the output data is compared with the predicted data of the ML algorithm to verify whether the subset of data (the subset of sensors) can accurately determine the stroke. If the subset of data cannot accurately determine the stroke, the ML algorithm may be repeated until a CI greater than n% (e.g., greater than 90%) is reached. The ML algorithm may use more data. The output data of the ML algorithm that satisfies the CI is razor metadata.

[0145] The experiment was conducted on a sensor reduction server (800). In some examples, a participant shaves using a razor such as the device (105) of FIG. 1, while an observer records the shaving characteristics. The generated data may include sensor data from a razor such as the device (105) of FIG. 1 (e.g., 100 Hz) and ground actual data of shaving characteristics (e.g., number of strokes, stroke direction, number of rinses). A ground actual data set of 100 Hz may have a shaving data set equal to 98. The shaving data set may include 97 unique participants or other statistically significant groups. In another example, a ground actual data set of 100 Hz may have a shaving data set equal to 30. The shaving data set may include 15 unique participants. In particular, the results of the force prediction neural network improve significantly as the frequency of sensor data collection increases, and, for example, when an RNN (e.g., LSTM-RNN) is trained with 100 Hz sensor data, it provides accurate feature predictions. In one example, the predicted force curve follows the pattern of the actual force curve with significant peak alignment.

[0146] One embodiment describes a method for storing raw data on a server. In this case, the raw usage data may depend on one or more sensors configured based on the configuration of the tool or the tool configuration data. For example, if the configuration may involve manufacturing a shaving tool with only the minimum number of sensors necessary to accurately determine the shaving implementation metadata for the implementation usage parameters, the data may be all raw data from the IMUs provided to collect specific data (e.g., all acceleration data only, all gyroscope data only, etc.). For example, a user may turn on the device to power on physically installed special-purpose sensors (which may be at least a minimum number).

[0147] According to some embodiments, if the configuration includes selectively enabling or disabling the total number of sensors on the shaving tool for the minimum number of sensors required to accurately determine the shaving implementation metadata, the user selectively turns on or off the power to the sensors collecting data. If the configuration includes selectively programming one or more sensors to capture only a subset of data that can be captured (e.g., only a subset of accelerometer data, only a subset of gyroscope data, such as a subset of axis data), data may be collected from the selectively programmed sensors. Alternatively, an off-the-shelf shaving product may provide all the raw data of the function.

[0148] FIG. 9 illustrates an example of a process for sensor reduction machine learning for a personal care device according to aspects of the present disclosure. In some examples, such operation may be performed by a system comprising a processor that executes a set of code to control functional elements of the device. Additionally or alternatively, the process may be performed using special-purpose hardware. Generally, such operation may be performed in accordance with the methods and processes described according to aspects of the present disclosure. For example, the operation may consist of various subsets or may be performed in conjunction with other operations described herein.

[0149] In operation (900), the system collects data from a set of sensors on a device, such as a razor (e.g., one example of the device of FIG. 1). In some cases, the operation of this step may be performed by referring to or by the data collection component described with reference to FIG. 6 and FIG. 8.

[0150] In operation (905), the system identifies one or more features, such as shaving features, based on the data. In some cases, the operation of this step may be performed by referring to or by the feature selection component described with reference to FIG. 8.

[0151] In operation (910), the system identifies a subset of sensors, and at least one sensor is not included in the subset of sensors. In some cases, the operation of this step may be performed by referring to a sensor selection component as described with reference to FIG. 8.

[0152] In some examples, the configuration of the shaving tool may include manufacturing the shaving tool with only the minimum number of sensors necessary to accurately determine the shaving implementation metadata for the implementation usage parameters. The configuration may include selectively enabling or disabling the total number of sensors of the shaving tool with respect to the minimum number of sensors necessary to accurately determine the shaving implementation metadata. The configuration may also include selectively programming one or more sensors to capture only a subset of data that can be captured (e.g., only a subset of accelerometer data, only a subset of gyro data such as a subset of axis data, etc.). In one embodiment of the present disclosure, an IMU may be physically installed in the shaving tool. The IMU may be trained by a machine learning model (e.g., a deep neural network) to provide information that can be used to accurately determine the stroke.

[0153] In some examples, the configuration of a personal care device other than a razor may include manufacturing the device with the minimum number of sensors necessary to determine implementation metadata regarding implementation usage parameters. The configuration may include optionally enabling or disabling the total number of sensors of the implementation with the minimum number of sensors necessary to accurately determine the associated metadata. The configuration may also include optionally programming one or more sensors to capture only a subset of captureable data (e.g., only a subset of accelerometer data, only a subset of gyro data such as a subset of axis data, etc.).

[0154] In one embodiment, ground measurement data may be input into an ML algorithm to determine a subset of data (and thus a subset of sensors) that can accurately determine when a stroke occurred. In one example, the output data is compared with the predicted data of the ML algorithm to verify whether the subset of data (the subset of sensors) can accurately determine the stroke. If the subset of data cannot accurately determine the stroke, the ML algorithm may be repeated until a CI greater than n% (e.g., greater than 90%) is reached. The ML algorithm may use more data. The output data of the ML algorithm that satisfies the CI is implementation data, such as razor metadata.

[0155] In operation (915), the system trains a machine learning model to predict one or more features using data from a subset of sensors. In some cases, the operation of this step may be performed by referring to or by a feature prediction component as described with reference to FIG. 8.

[0156] One embodiment of the present disclosure describes inputting raw usage data into an algorithm to generate implementation usage data. There may be a step of comparing the raw usage data with reference data (e.g., shaving implementation metadata). For example, the raw data may be collected from a ready-made product (having a sufficient number of sensors for specific usage parameters) or a configured shaving tool. The raw usage data may be input into a comparator along with a metadata output (where a signal from a subset of ground actual data can accurately determine when a stroke occurs) to determine whether or how the shaving tool is used (e.g., strokes). The implementation usage data may be provided as a number (e.g., number of strokes, etc.) to maintain tracking of usage.

[0157] Embodiments of the present disclosure include obtaining a composition of existing data from a bank, live data from a user, live data from a group of users, data tracked over time from a user, data tracked from a group of users, or any combination thereof. The data may be combined to facilitate end-of-life prediction (i.e., data input by a user and data collected from a user).

[0158] One embodiment of the present disclosure describes a method for using data. A cloud server identical or different from the cloud server described herein may transmit data to another network or transmit data to memory (e.g., storage device, cache, etc.). In one example, the data usage process may include accessing user information or general reference information to determine whether it is appropriate to automatically deliver products to the user and provide recommendations. In one embodiment, a SQL server may be used to process information from a user. This may include receiving user preferences, payment information, calendar ratings, confirmations, etc., directly from the user, etc., or a combination thereof, from a user profile on an enterprise server or through a personal computing device (e.g., via an application).

[0159] In one embodiment, raw data may be input into the described system or network to accurately determine the % end-of-life for a component such as a device or razor cartridge. For example, tool usage data (e.g., indicating the number of strokes) is used in conjunction with user data collected from user profiles, real-time statistics, or user input through an application (e.g., order patterns, movement patterns, cartridge preferences, actual strokes / cartridge, preference for using strokes / cartridges, etc.). Additionally or alternatively, tool usage data may be compared with general reference data (e.g., average strokes per tool (e.g., cartridge), average strokes per tool (e.g., cartridge) for a specific age group, etc.). Strokes may be used to determine the available life for a percentage of the life calculation (e.g., 90% of life remaining) used for recommendations (other products, different types of cartridges, companion products, etc.) or supplements.

[0160] Although independent blocks and / or a specific order are depicted for illustrative purposes, it should be understood that one or more blocks of any method of FIG. 9 or components of the device of FIG. 8 may be combined, omitted, bypassed, rearranged, and / or flow in any order or be located in any device or a single device. For example, one or more components shown in FIG. 8 and / or one or more blocks shown in FIG. 9 may additionally or alternatively be located or implemented on a personal care device, a hub, a server, an IoT device, or a combination thereof.

[0161] Additional features and examples

[0162] Additional examples include a system, device, non-transient computer-readable medium, a method for training a machine learning model, running a trained machine learning model, capturing and / or processing data for a machine learning model, providing feedback based on the output of a trained machine learning model, or a combination thereof.

[0163] Example 1 may include a plurality of sensors and a processor coupled to the plurality of sensors, and the processor is configured to execute a machine learning model trained to predict usage features based on data from the plurality of sensors.

[0164] Example 2 may include the system of Example 1, in which a plurality of sensors include an accelerometer and a gyroscope located on an inertial measurement unit (IMU) of a personal care device.

[0165] Example 3 may include any system of Example 1 or Example 2, wherein the IMU includes at least one accelerometer per axis for three axes and at least one gyroscope per axis for three axes.

[0166] Example 4 may include any of the systems in Examples 1 through 3, wherein at least one accelerometer and at least one gyroscope comprise a microelectromechanical system (MEMS).

[0167] Example 5 may include any of the systems in Examples 1 through 4, in which the IMU does not have a magnetometer.

[0168] Example 6 may include any of the systems from Examples 1 to 5 and further includes a filter that prevents data from a magnetometer from reaching one or more of the machine learning model to be trained or the machine learning model that has been trained.

[0169] Example 7 may include any of Examples 1 through 6, wherein the processor is located on the IMU and is physically coupled to the accelerometer and gyroscope through a connection on the IMU.

[0170] Example 8 may include any of Examples 1 through 7, wherein a processor is located on a controller of a personal care device and is physically coupled to a plurality of sensors through a connection between the controller and an IMU.

[0171] Example 9 may include any of Examples 1 through 8, wherein a processor is located on a hub adjacent to a personal care device and is coupled to communicate with a plurality of sensors through a connection between the hub and the personal care device.

[0172] Example 10 may include any of Examples 1 through 9, wherein a processor is located on a hub adjacent to a server or personal care device that is communically coupled to one or more personal care devices, and the processor is communically coupled to a plurality of sensors through a connection between the server and one or more personal care devices or the hub.

[0173] Example 11 may include any of the systems in Examples 1 through 10, wherein the personal care device is for providing data from a plurality of sensors to a hub via a short-range wireless connection, and the hub is for providing data to a server located on a computing network via a long-range connection.

[0174] Example 12 may include any system of Examples 1 through 11, which is a trained stacked long short-term memory recurrent neural network (LSTM-RNN).

[0175] Example 13 may include any of Examples 1 to 12, wherein the usage features include a force applied by the personal care device, a distance traveled by the personal care device, work performed by the personal care device, or a combination thereof.

[0176] Example 14 may include any system among Examples 1 to 13, wherein the processor is configured to execute one or more of the following: a first machine learning model for predicting a first use feature related to a personal care device based solely on accelerometer data and gyroscope data input to the machine learning model to be trained, wherein the first use feature includes force; a second machine learning model for predicting a second use feature related to a personal care device based solely on accelerometer data and gyroscope data input to the machine learning model to be trained, wherein the second use feature includes distance traveled; or a third machine learning model for predicting a third use feature related to a personal care device based solely on accelerometer data and gyroscope data input to the machine learning model to be trained, wherein the third use feature includes work performed by the personal care device.

[0177] Example 15 may include any of the systems in Examples 1 through 14, wherein the processor is configured to predict one or more of the number of strokes, stroke force, stroke distance, rinse event, or end of life associated with a personal care device based on usage characteristics.

[0178] Example 16 may include any system of Examples 1 through 15, wherein the processor is configured to perform one or more of the following: using a function to predict one or more of the stroke count, stroke force, stroke distance, rinse event, or end of life based on usage characteristics, or running an additional machine learning model trained to determine one or more of the stroke count, stroke force, stroke distance, rinse event, or end of life based on usage characteristics.

[0179] Example 17 may include any system of Examples 1 through 16, in which the personal care device includes a razor device.

[0180] Example 18 may include any system of Examples 1 through 17, in which a razor device is coupled to communicate in close proximity to an Internet of Things (IoT) device.

[0181] Example 19 may include any of the systems in Examples 1 through 18, and further includes a radio frequency interface, memory, storage device, sensor controller, input / output device, or a combination thereof.

[0182] Example 20 may include any of the systems in Examples 1 through 19, wherein the processor is configured to perform one or more of the following: training a machine learning model to predict usage characteristics related to a personal care device based on data from multiple sensors, collecting data from multiple sensors, or providing feedback based on usage characteristics.

[0183] Example 21 may include at least one computer-readable storage medium comprising an instruction set that, when executed by a processor, causes the processor to implement any of the systems of Examples 1 through 20. For example, at least one computer-readable storage medium comprises a set of instructions that, when executed, causes the processor to execute a machine learning model trained to predict usage characteristics related to a personal care device based on a plurality of sensors.

[0184] Example 22 may include a method for implementing the operation of any of the systems in Examples 1 through 20. For example, the method may include capturing data through a plurality of sensors of a personal care device and running a machine learning model trained to predict usage characteristics related to the personal care device based on the data.

[0185] Example 23 may include a method for implementing the operation of any of the systems in Examples 1 through 20. For example, the method may include means for capturing data through a plurality of sensors of a personal care device, and means for executing a machine learning model trained to predict usage characteristics related to the personal care device based on the data.

[0186] Example 24 may include a feature prediction component configured to train a machine learning model to predict usage features related to a personal care device using data from a subset of sensor sets, and the feature prediction component is configured to initialize the machine learning model with a set of initial model parameters, input data from a subset of sensor sets into the machine learning model, run the machine learning model to generate output data corresponding to usage features, compare the output data with reference data to determine whether the machine learning model predicts usage features, iteratively update at least one model parameter, iteratively input data from a subset of sensor sets, and iteratively compare the output data with reference data until the machine learning model predicts usage features related to the personal care device.

[0187] Example 25 may include the feature prediction device of Example 24, in which the initial model parameters include model width, model length, model link connections between model nodes, weights to be applied to input data, input data to be applied to input data, or a combination thereof.

[0188] Example 26 may include any of Examples 24 and 25, wherein a subset of the sensor set includes an accelerometer and a gyroscope located on a personal care device, and the feature prediction component is configured to initialize a machine learning model with a set of initial model parameters, input data from the accelerometer and gyroscope into the machine learning model, run the machine learning model to generate output data corresponding to a use feature, compare the output data with reference data to determine whether the machine learning model predicts a use feature, iteratively update at least one model parameter, iteratively input data from the accelerometer and gyroscope, and iteratively compare the output data with reference data until the machine learning model predicts a use feature related to the personal care device.

[0189] Example 27 may include any of the feature prediction devices of Examples 24 through 26, wherein the accelerometer and gyroscope are positioned on the inertial measurement unit (IMU) of the personal care device.

[0190] Example 28 may include any of Examples 24 through 27, wherein the feature prediction device is located on a server that is communically coupled to one or more of a personal care device or a hub adjacent to the personal care device, and the feature prediction component is communically coupled to a subset of a sensor set through a connection between the server and one or more personal care devices and hubs.

[0191] Example 29 may include any of the feature prediction devices of Examples 24 to 28, wherein the personal care device provides data from a plurality of sensors to a hub via a short-range wireless connection, and the hub provides data to a server located on a cloud computing network via a long-range connection.

[0192] Example 30 may include any of the feature prediction devices of Examples 24 through 29, wherein the feature prediction component is located in one or more of a personal care device or a hub adjacent to a personal care device.

[0193] Example 31 may include any feature prediction device among Examples 24 to 30, in which the machine learning model is a stacked long short-term memory recurrent neural network (LSTN-RNN).

[0194] Example 32 may include any of the feature prediction devices in Examples 24 through 31, wherein the use features include a force applied by the personal care device, a distance traveled by the personal care device, work performed by the personal care device, or a combination thereof.

[0195] Example 33 describes a feature prediction component that initializes and executes a first machine learning model to predict a first use feature related to a personal care device based solely on accelerometer and gyroscope data inputs to the machine learning model to be trained, wherein the first use feature includes force, and reference data includes data from a force sensor on the personal care device, data from a capacitive sensor on the personal care device, or a combination thereof; initializes and executes a second machine learning model to predict a second use feature related to a personal care device based solely on accelerometer and gyroscope data inputs to the machine learning model to be trained, wherein the second use feature includes distance traveled, and reference data includes data from a 6-degrees-of-freedom (6DoF) tracking device coupled to the personal care device; or initializes and executes a third machine learning model to predict a third use feature related to a personal care device based solely on accelerometer and gyroscope data inputs to the machine learning model to be trained, wherein the third use feature includes work performed by the personal care device, and reference data includes data from a force sensor on the personal care device, the personal care device Any of the feature prediction devices in Examples 24 to 32 may be configured to perform one or more of initializing and executing a third machine learning model comprising data from a capacitive sensor, data from a 6DoF tracking device, or a combination thereof.

[0196] Example 34 may include any of the feature prediction devices of Examples 24 to 33, wherein the feature prediction component is configured to predict one or more of the number of strokes, stroke force, stroke distance, rinse event, and end of life based on the use feature.

[0197] Example 35 may include any of Examples 24 through 34, wherein the feature prediction component is configured to perform one or more of the following: using a function to predict one or more of the stroke count, stroke force, stroke distance, rinse event, and end of life based on the feature used, or running an additional machine learning model trained to determine one or more of the stroke count, stroke force, stroke distance, rinse event, and end of life based on the feature used.

[0198] Example 36 may include any feature prediction device among Examples 24 to 35, in which the personal care device includes a razor device.

[0199] Example 37 may include any feature prediction device among Examples 24 to 35, and further includes one or more data collection devices comprising a data collection component configured to collect data from a sensor set, wherein the sensor set comprises an accelerometer of a personal care device, a gyroscope of a personal care device, and additional sensors located on a personal care device, additional sensors located on a tracking device coupled to a personal care device, or a combination thereof, and a feedback component configured to collect user profile data used to train a feedback device configured to prepare data before reaching a machine learning model, or to provide feedback based on a machine learning model or usage features related to the personal care device.

[0200] Example 38 may include at least one computer-readable storage medium comprising a set of instructions that, when executed by a processor, cause the processor to implement any of the feature prediction devices of Examples 24 through 37. For example, at least one computer-readable storage medium comprises a set of instructions that, when executed, cause a process to implement a feature prediction component for training a machine learning model to predict usage features related to a personal care device using data from a subset of a sensor set.

[0201] Example 39 may include a method for implementing any of the feature prediction devices of Examples 24 through 37. For example, the method may include training a machine learning model to predict usage features associated with a personal care device using data from a subset of a sensor set.

[0202] Example 40 may include a method comprising means for implementing any of the feature prediction devices of Examples 24 through 37. For example, the method may include means for training a machine learning model to predict usage features associated with a personal care device using data from a subset of a sensor set.

[0203] Example 41 may include a data collection device comprising a data collection component configured to collect data from a sensor set, wherein the sensor set includes an accelerometer on a personal care device, a gyroscope on a personal care device, and additional sensors located on a personal care device, additional sensors located on a tracking device coupled to a personal care device, or a combination thereof.

[0204] Example 42 may include the data collection device of Example 41, the data collection component provides access to the data, and the data is provided to one or more comparators configured to receive output data from a machine learning model to be trained, the machine learning model to be trained receives data from a subset of a sensor set, receives reference data from a sensor set, and is trained to compare the output data with the reference data, or the machine learning model is trained to predict usage characteristics associated with a personal care device based on data from a subset of a sensor set.

[0205] Example 43 may include any of the data set devices of Examples 41 and 42, wherein the sensor set comprises an IMU, a capacitance-to-digital converter, a force sensor, a rinse sensor, a grip sensor, a temperature sensor, an accelerometer, a gyroscope, a magnetometer, a resistance contact, a piezo contact microphone, a Hall effect sensor, an external camera, a strain gauge, a load cell, a reed switch, a linear variable differential transformer, or a combination thereof.

[0206] Example 44 may include any of the data set devices of Examples 41 through 43, wherein the data collection component is configured to prepare the data before the data reaches the machine learning model.

[0207] Example 45 may include any of the data set devices of Examples 41 through 44, wherein the data collection component is configured to implement flattening, normalization, rescaling, clock synchronization, elimination-and-splitting, or a combination thereof.

[0208] Example 46 may include any of the data set devices of Examples 41 through 45, wherein flattening includes passing the data through a Butterworth filter to flatten the frequency response, normalization includes normalizing the input data around the mean value, and rescaling includes reducing the input data to minimize the exploding gradient during training.

[0209] Example 47 may include any of the data set devices of Examples 41 to 46, wherein clock synchronization includes using a corresponding signal between the input data and the reference data to eliminate the offset between the input data and the reference data.

[0210] Example 48 may include any of the data set devices of Examples 41 through 47, wherein the removal-and-partitioning comprises removing an inactive period longer than a predetermined period from the reference data and partitioning the remaining active period into a subset of the active period.

[0211] Example 49 may include any of the data set devices of Examples 41 through 48, wherein the data collection component is configured to collect user profile data to be used to train a machine learning model.

[0212] Example 50 may include any of the data set devices of Examples 41 to 49, wherein user profile data includes survey data.

[0213] Example 51 may include any of the data collection devices of Examples 41 through 50, wherein the data collection component is located on a personal care device, a hub coupled in close proximity to the personal care device, or one or more servers coupled to one or more of the personal care device or the hub.

[0214] Example 52 may include at least one computer-readable storage medium comprising a set of instructions that, when executed by a processor, cause the processor to implement any of the data set devices of Examples 41 through 51. For example, at least one computer-readable storage medium comprises a set of instructions that, when executed, cause the processor to implement a data collection component to collect data from a sensor set, wherein the sensor set comprises an accelerometer on a personal care device, a gyroscope on a personal care device, an additional sensor located on a personal care device, an additional sensor located on a tracking device coupled to a personal care device, or a combination thereof.

[0215] Example 53 includes a method for implementing the operation of any of the data collection devices in Examples 41 through 51. For example, the method may include collecting data from a set of sensors, the set of sensors including an accelerometer on a personal care device, a gyroscope on a personal care device, additional sensors located on a personal care device, additional sensors located on a tracking device coupled to a personal care device, or a combination thereof.

[0216] Example 54 includes a method comprising means for implementing the operation of any of the data collection devices in Examples 41 through 51. For example, the method may include means for collecting data from a sensor set, the sensor set comprising an accelerometer on a personal care device, a gyroscope on a personal care device, an additional sensor located on a personal care device, an additional sensor located on a tracking device coupled to a personal care device, or a combination thereof.

[0217] Example 55 may include a feedback device comprising a feedback component configured to provide feedback based on usage characteristics related to a personal care device, and the usage characteristics are predicted using a machine learning model trained using data from a subset of a sensor set.

[0218] Example 56 may include the feedback device of Example 55, the feedback including a control function on the personal care device, a control function for an Internet of Things (IoT) device near the personal care device, a use recommendation, a supplement function, replacement information, or a combination thereof.

[0219] Example 57 may include any of Example 55 and Example 57, wherein the feedback component is configured to initiate a control command for controlling an element of a personal care device.

[0220] Example 58 may include any of Examples 55 through 57, wherein the personal care device comprises a razor device, the element comprises one or more of the cartridge of the razor device or the dispenser of the razor device, and the feedback component is configured to cause a control command to reach a controller module coupled to the cartridge of the razor device to modify the way one or more cartridge components come into contact with the user, or a control command to reach a controller module coupled to the dispenser of the razor device to modify the way a personal care composition is dispensed from the razor device.

[0221] Example 59 may include any of the feedback devices of Examples 55 through 58, wherein the feedback component is located on a personal care device, a hub coupled in proximity to the personal care device, or one or more servers coupled to one or more of the personal care device or the hub.

[0222] Example 60 may include at least one computer-readable storage medium comprising a set of instructions that, when executed by a processor, cause the processor to implement any of the data acquisition devices of Examples 55 through 59. For example, at least one computer-readable storage medium comprises a set of instructions that, when executed, cause the processor to implement a feedback component for providing feedback based on usage characteristics related to a personal care device, wherein the usage characteristics are predicted using a machine learning model trained using data from a subset of a sensor set.

[0223] Example 61 may include a method for implementing the operation of any of the feedback devices in Examples 55 through 59. For example, the method may include providing feedback based on usage characteristics related to a personal care device, and the usage characteristics are predicted using a machine learning model trained using data from a subset of a sensor set.

[0224] Example 62 may include a method comprising means for implementing the operation of any of the feedback devices in Examples 55 through 59. For example, the method may include means for providing feedback based on usage characteristics related to a personal care device, and the usage characteristics are predicted using a machine learning model trained using data from a subset of a sensor set.

[0225] The description and drawings described herein represent exemplary configurations and do not represent all implementations within the scope of the claims. For example, operations and steps may be rearranged, combined, or otherwise modified. Additionally, structures and devices may be represented in the form of block diagrams to indicate relationships between components and to avoid obscuring the described concepts. Similar configurations or functions may have the same designation but different reference numerals corresponding to different drawings.

[0226] Some modifications to the present disclosure may be readily apparent to a person skilled in the art, and the principles defined herein may be applied to other variations without departing from the scope of the present disclosure. Accordingly, the present disclosure is not limited to the examples and designs described herein, but should be applied to the broadest extent consistent with the principles and novel features disclosed herein.

[0227] The described methods and / or components may be implemented or performed by a general-purpose processor, a digital signal processor (DSP), an application-specific integrated unit (ASIC), a field programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof. A general-purpose processor may be a microprocessor, a conventional processor, a controller, a microcontroller, or a state machine. A processor may also be implemented as a combination of computing devices (e.g., a DSP and a microprocessor, a multi-microprocessor, one or more microprocessors with a DSP core, or any other combination of such configurations). Accordingly, the functions described herein may be implemented in hardware or software and may be executed by a processor, firmware, or a combination thereof. When configured as software executed by a processor, the functions may be stored in the form of instructions or code on a computer-readable medium.

[0228] Computer-readable media include both non-transient computer storage media and communication media that include any match that facilitates the transmission of code or data. Non-transient storage media may be any available medium accessible by a computer. For example, non-transient computer-readable media may include random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), compact disc (CD) or other optical disc storage devices, magnetic disc storage devices, or any other non-transient medium capable of carrying or storing data or code.

[0229] Additionally, connection components may be appropriately referred to as computer-readable media. For example, if code or data is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technology such as infrared, radio, or microwave signals, coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technology is included in the definition of the medium.

[0230] In the present disclosure and the following claims, the word "or" indicates a inclusion list, for example, X, Y, or Z is a list of X, Y, or Z, or XY, or XZ, or YZ, or XYZ. Similarly, a list of one or more of X, Y, or Z, or a list of X, Y, or Z, or a combination thereof, means X, Y, or Z, or XY, or XZ, or YZ, or XYZ. Additionally, the phrase "based on" is not used to indicate a set of closed conditions. For example, a step described as "based on condition A" may be based on both condition A and condition B. That is, the phrase "based on" should be interpreted to mean "based at least in part on." Additionally, "a" or "an" indicates "at least one."

Claims

Claim 1 A personal care device comprising a plurality of sensors, wherein the plurality of sensors include an accelerometer and a gyroscope located in an inertial measurement unit (IMU) of the personal care device, and the personal care device is a personal care grooming device or a washing device configured to contact a user's body to groom or wash the body; A system comprising a processor coupled to the plurality of sensors, wherein the processor is configured to execute at least one machine learning model including an artificial neural network trained to predict usage characteristics related to the personal care device based on data from the plurality of sensors, wherein the usage characteristics include a force applied by the personal care device to the body of the user, a distance moved by the personal care device on the body of the user, or a combination thereof, and when the at least one machine learning model including the artificial neural network is executed by the processor, i) the force applied by the personal care device to the body of the user is predicted based only on accelerometer data and gyroscope data from an IMU input to a machine learning model trained to predict and output a measurement of the force, ii) the distance moved by the personal care device on the body of the user is predicted based only on accelerometer data and gyroscope data from an IMU input to a machine learning model trained to predict and output a measurement of the distance moved, or both i) and ii) are predicted. Claim 2 A system according to claim 1, wherein the IMU has no magnetometer, or the system further comprises a filter that prevents magnetometer data from the magnetometer from reaching one or more of the machine learning model to be trained or the machine learning model trained to predict the usage features. Claim 3 A system according to claim 1, wherein the processor is: located on the IMU and physically coupled to the accelerometer and the gyroscope through a connection on the IMU; located on the controller of the personal care device and physically coupled to the plurality of sensors through a connection between the controller and the IMU; located on a hub adjacent to the personal care device and communicably coupled to the plurality of sensors through a connection between the hub and the personal care device; or located on a server communicably coupled to one or more of the personal care device or the hub adjacent to the personal care device, and the processor is communicably coupled to the plurality of sensors through a connection between the server and one or more of the personal care device or the hub. Claim 4 A system according to claim 1, wherein the machine learning model comprising the artificial neural network is a stacked long short-term memory recurrent neural network (LSTM-RNN) trained using time series IMU training data. Claim 5 A system according to claim 1, wherein the processor is configured to calculate work performed by the personal care device based on the force and the distance traveled, which are predicted independently of each other. Claim 6 A system according to claim 1, wherein the processor is configured to further execute a machine learning model that predicts and outputs a measurement of work performed by the personal care device based only on the accelerometer data and gyroscope data from the IMU input to the machine learning model. Claim 7 A system according to claim 1, wherein the processor is configured to predict one or more of the number of strokes, stroke force, stroke distance, rinse event, or end of life associated with the personal care device based on the usage features. Claim 8 A system according to claim 7, wherein the processor is configured to use a function that predicts one or more of the stroke count, the stroke force, the stroke distance, the rinse event, or the end of life based on the usage feature; or configured to execute an additional machine learning model trained to determine one or more of the stroke count, the stroke force, the stroke distance, the rinse event, or the end of life based on the usage feature. Claim 9 A system according to claim 1, wherein the personal care device comprises a shaving device, and the usage feature comprises a force applied to the user's body by the personal care device and a distance moved on the user's body by the personal care device. Claim 10 In claim 9, the shaving device is a system coupled to communicate in close proximity to an Internet of Things (IoT) device. Claim 11 A system according to claim 1, further comprising a radio frequency interface, memory, storage device, sensor controller, input / output device, or a combination thereof. Claim 12 A system according to claim 4, wherein the machine learning model is trained to predict the usage features associated with the personal care device based on the time-series IMU training data from the plurality of sensors and reference data from a 6-degrees-of-freedom (6DoF) tracking solution coupled to the personal care device. Claim 13 A computer-readable storage medium comprising at least one instruction set that causes the processor to execute at least one machine learning model, which includes an artificial neural network trained to predict usage characteristics related to a personal care device based on data from a plurality of sensors, including accelerometers and gyroscopes located on an inertial measurement unit (IMU) of the personal care device, wherein the personal care device is a personal care grooming device or a washing device configured to contact the body of a user to groom or wash the body, and the usage characteristics include a force applied by the personal care device to the body of the user, a distance moved by the personal care device on the body of the user, or a combination thereof, and when the at least one machine learning model including the artificial neural network is executed by the processor, i) the force applied by the personal care device to the body of the user is predicted based solely on accelerometer data and gyroscope data from the IMU input to a machine learning model trained to predict and output a measurement of the force, or ii) the distance moved by the personal care device on the body of the user is a measurement of the distance moved At least one computer-readable storage medium that predicts based solely on accelerometer and gyroscope data from the IMU, which are input to a machine learning model trained to predict and output, or where both i) and ii) are predicted. Claim 14 A method for executing at least one machine learning model comprising an artificial neural network trained to predict usage characteristics related to a personal care device based on data from a plurality of sensors, including accelerometers and gyroscopes located on an inertial measurement unit (IMU) of the personal care device, wherein the personal care device is a personal care grooming device or a washing device configured to contact a user's body to groom or wash the body, and the usage characteristics include a force applied by the personal care device to the user's body, a distance moved by the personal care device on the user's body, or a combination thereof, and when the at least one machine learning model comprising the artificial neural network is executed, i) the force applied by the personal care device to the user's body is predicted based solely on accelerometer data and gyroscope data from the IMU input to a machine learning model trained to predict and output a measurement of the force, or ii) the distance moved by the personal care device on the user's body is predicted based on accelerometer data from the IMU input to a machine learning model trained to predict and output a measurement of the distance moved and A method in which prediction is made based solely on gyroscope data, or in which both i) and ii) are predicted. Claim 15 delete

Citation Information

Patent Citations

  • Electric toothbrush

    JP2009240760A

  • Personal care device

    JP2019171060A