KR20070071917A - Apparatus and method for learning adaptively grip patterns of mobile devices - Google Patents

Abstract

An apparatus and a method for adaptively learning grip patterns in a mobile terminal are provided to reflect personal characteristics for grip patterns by adaptively updating grip pattern recognition models using a user's feedback in transferring function modes according to grip patterns. An apparatus for adaptively learning grip patterns in a mobile terminal comprises a detection part(100), a comparing and driving part(200), a judgement part(300), and an update part(400). The detection part(100) detects a contact area with the mobile terminal when the user grips the mobile terminal. The comparing and driving part(200) compares a recognition model, stored in a recognition model DB part(500), with information about the contact area and drives a function mode of the mobile terminal. The judgement part(300) judges whether the driven function mode is matched with the user's intended function mode. The update part(400) updates the recognition model DB part(500) by a judgement result of the judgement part(300).

Description

Apparatus and method for adaptive learning of mobile terminal gripping pattern {APPARATUS AND METHOD FOR LEARNING ADAPTIVELY GRIP PATTERNS OF MOBILE DEVICES}

1 is a diagram illustrating a gripping pattern according to a function mode of a mobile terminal.

2 is a view showing the overall configuration of the adaptive learning device of the mobile terminal gripping pattern according to an embodiment of the present invention.

3 is a diagram illustrating a configuration of a sensing unit of the configuration of FIG. 2.

4 is a diagram illustrating modeling of a reinforcement learning theory algorithm applied to an embodiment of the present invention.

5 is a view showing the overall flow of the adaptive learning method of the mobile terminal gripping pattern according to an embodiment of the present invention.

* Description of the main parts of the drawings *

100: detection unit 110: contact detection sensor

120: contact signal processor 200: comparison driver

300: determination unit 400: update unit

500: recognition model database

The present invention relates to an adaptive learning apparatus and method for a mobile terminal gripping pattern, and more particularly, a model for recognizing a gripping pattern using feedback of a user in switching a function mode of a mobile terminal according to a gripping pattern of a mobile terminal. The present invention relates to an adaptive learning apparatus and method for a mobile terminal gripping pattern that can reflect an individual characteristic of a gripping pattern by adaptively updating.

Recently, with the spread of mobile terminals such as mobile phones and PDAs (Personal Digital Assistants), one portable terminal has been manufactured to perform various functions. For example, a mobile phone is basically a terminal having a call function, but a recent mobile phone is further equipped with a function such as SMS (Short Message Service), a digital camera, an MP3 player, and a video playback. As the functions of the mobile terminals are diversified as described above, the user can be provided with a complex service using a single portable terminal.

In general, in order to operate a mobile terminal having a plurality of functions in a specific function mode, the user operates a keypad or a touch pad provided in the mobile terminal to search for a function mode to perform a desired function and then return to the function mode. Should be instructed to switch. For example, while the user checks a menu screen provided through the display unit of the portable terminal, the user operates a direction key provided in the portable terminal to search for a desired function mode. When the function mode desired by the user is found through the display unit, the user may select a searched function mode by operating a selection button, and the mobile terminal is switched to the function mode selected by the user. According to the conventional technology, in order to switch the portable terminal to each function mode, there is inconvenience that a user must go through a multi-step key input process.

On the other hand, in order to simplify the switching of the function mode of the portable terminal, the portable terminal may be provided with a shortcut key. A key value for switching the portable terminal to a specific function mode is mapped to the shortcut key, and when the user uses the shortcut key, the user can switch the function mode of the portable terminal by a single key operation. However, even in the case of using a shortcut key, a user who is not familiar with the use of the mobile terminal has the inconvenience of knowing a shortcut key mapped to a specific function mode. In addition, the number of shortcut keys that can be provided in the portable terminal is limited, so dedicating the shortcut keys individually for each function of the portable terminal causes a waste of key interface space that can be provided in the portable terminal. .

Therefore, a patent application (10-2004-0095293) has been filed by the present applicant to solve the inconvenience of having to go through a multi-step terminal operation process to switch the function mode of the portable terminal as described above. However, the above application has a limitation in that a recognition model, which is a reference for recognizing a function mode of a mobile terminal, must be built in advance, and in actual use environments, differences in gripping patterns and changes in user habits according to individual characteristics of users. There is a limit that does not reflect the change in the holding posture.

The present invention has been devised to solve the above problems, and it is possible to adaptively change a recognition model by using a user's feedback on a function mode of a mobile terminal according to the mobile terminal's holding pattern. An adaptive learning apparatus and method are provided.

The objects of the present invention are not limited to the above-mentioned objects, and other objects that are not mentioned will be clearly understood by those skilled in the art from the following description.

According to an aspect of the present invention, there is provided an adaptive learning apparatus for a mobile terminal holding pattern, the sensing unit sensing a contact area with the mobile terminal as a user grips the mobile terminal; A comparison of driving the function mode of the portable terminal by comparing the information on the contact area with the recognition model in a recognition model database unit storing a recognition model which is a reference for recognizing the functional mode of the portable terminal A drive unit; A determination unit which determines whether the driven function mode matches the function mode intended by the user; And an updating unit for updating the recognition model database unit according to the determination result of the determination unit.

In addition, the adaptive learning method of a mobile terminal holding pattern according to an embodiment of the present invention, the sensing step of detecting a contact area with the mobile terminal as the user grips the mobile terminal; A comparison of driving the function mode of the portable terminal by comparing the information on the contact area with the recognition model in a recognition model database unit storing a recognition model which is a reference for recognizing the functional mode of the portable terminal Driving step; A determination step of determining whether the driven function mode matches the function mode intended by the user; And updating the recognition model database unit according to the determination result.

Specific details of other embodiments are included in the detailed description and the drawings. Advantages and features of the present invention and methods for achieving them will be apparent with reference to the embodiments described below in detail with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be embodied in various different forms, and the present embodiments merely make the disclosure of the present invention complete, and are common in the art to which the present invention pertains. It is provided to fully inform those skilled in the art of the scope of the invention, which is to be defined only by the scope of the claims. Like reference numerals refer to like elements throughout.

Hereinafter, with reference to block diagrams or flowcharts for describing preferred embodiments of the present invention will be described in more detail.

1 is a diagram illustrating a gripping pattern according to a function mode of a mobile terminal.

The portable terminal may have various function modes such as a call mode, an SMS mode, and a digital camera mode. As illustrated in FIG. 1, when the user uses the portable terminal, there is a difference in the pattern of gripping the portable terminal according to the function mode of the portable terminal. In the portable terminal according to an embodiment of the present invention, the user recognizes a function mode intended by the user, and switches to a recognized function mode. As a control information for the switching, information about the shape and posture of the user holding the portable terminal ( Hereinafter, referred to as "gripping information") can be used. That is, the function mode of the mobile terminal intended by the user may be recognized according to the grip information of the user. In this case, as the user grips the portable terminal, information about a contact area of the outer surface of the portable terminal in contact with the user's hand (hereinafter referred to as 'contact information') may be used as the grip information.

In addition, when the user grips the portable terminal, when the portable terminal rotates based on a predetermined reference axis set in the portable terminal or when the reference axis is inclined at a specific angle with respect to the gravity direction, Information about the rotation angle or directionality (hereinafter, such information is referred to as 'movement information') may also be used as the gripping information. In addition, the 'removable state' of the external device which can be attached to and detached from the portable terminal may also be used as information for recognizing a function mode of the portable terminal desired by the user together with the gripping information (contact information and motion information).

2 is a view showing the overall configuration of the adaptive learning device of the mobile terminal gripping pattern according to an embodiment of the present invention. Referring to FIG. 2, the detector 100, the comparison driver 200, the determiner 300, the updater 400, and the recognition model database 500 are included.

The detector 100 detects a contact area with the portable terminal as the user grips the portable terminal. To this end, the sensing unit 100 may include one or more touch sensing sensors capable of sensing whether a touch exists, and each touch sensing sensor may be located on the surface of the portable terminal. The sensor 100 will be described with reference to FIG. 3.

3 is a diagram illustrating a configuration of the sensing unit 100 in the configuration of FIG. 2. Referring to FIG. 3, the detector 100 may include at least one touch sensor 110 and a touch signal processor 120.

The touch sensor 110 detects whether or not the user touches the grip, and outputs a signal indicating whether the user contacts the touch signal processor 120 to be described later. The touch sensor 110 may be a touch sensor that detects whether a user directly contacts the user, but the present invention is not limited thereto. Therefore, the touch detection sensor 110 may be a non-contact sensor that can detect that a body part such as a user's hand is located within a predetermined distance from the portable terminal even if there is no direct contact with the user and determine that contact has occurred.

In addition, as the user grips the portable terminal, the sensing unit 100 detects an angle at which the portable terminal is rotated or an angle at which the reference axis forms a gravity direction based on a predetermined reference axis set in the portable terminal. You may. To this end, the sensing unit 100 may include an inertial sensor or an acceleration sensor manufactured by mechanical or micro electro-mechanical systems (MEMS) technology. For example, if the sensing unit 100 is configured using a gyroscope, the sensing unit 100 may measure a rotation angle due to the rotational movement of the portable terminal. Meanwhile, if the sensing unit 100 is configured using the MEMS type acceleration sensor, the angle between the specific reference axis of the mobile terminal and the direction of gravity due to the tilting of the mobile terminal as well as the rotation angle due to the rotational motion of the mobile terminal may be measured. It may be. In this case, the comparison driving unit 200 may additionally use the information on the detected angle in recognizing the function mode of the portable terminal.

The touch signal processor 120 combines the signals output by the touch sensor 110 to generate contact information for the contact area, and provides the generated contact information to the comparison driver 200 to be described later.

Referring back to FIG. 2, the comparison driving unit 200 stores the recognition model and the recognition model in the recognition model database unit 500, which stores a recognition model that is a reference for recognizing a functional mode of the portable terminal. The contact information is compared to serve to drive the function mode of the portable terminal.

In this case, the comparison driving unit 200 receives the contact information from the contact signal processing unit 120, and when there is a corresponding recognition model in the built-in recognition model database unit 500, the corresponding function according to the recognition model. It will run by switching to the mode. However, there is a difference in the pattern of gripping the portable terminal for each individual, and the gripping pattern may change according to the environment as each individual uses the portable terminal. In this case, the recognition model for the gripping pattern should be updated by using a feedback process that can reflect the changed gripping pattern of the user each time the portable terminal is used.

A reinforcement learning theory algorithm may be used to perform the update, which will be described later in the description of FIG. 4.

Determination unit 300 by the comparison driving unit 200 It determines whether the function mode of the driven portable terminal matches the function mode intended by the user. In this case, the determination unit 300 determines whether a match is made by using a predetermined action that the user inputs to the mobile terminal.

For example, when the user grips the mobile terminal to take a picture by using the digital camera function, but the grip position of the mode 2 (call mode) of FIG. The mode will be driven. Since then, the user has habitually grasped the posture of the mode 2 of FIG. 1 for taking a picture, and it is cumbersome to switch the call function mode that is driven each time back to the picture taking mode every time. It was hoped that the shooting mode would be activated.

In this case, when the user grips in the posture of mode 2 and the call mode function is displayed on the portable terminal, the user may take the above input action and input a feedback signal having a rejection meaning to the determination unit 300 of the portable terminal. The input action may be an action of shaking the portable terminal or pressing a key button provided in the portable terminal. Such an action is only an embodiment of the present invention and is not limited to such an action.

The determination unit 300 receives an action indicating a meaning of rejection of the user, and determines whether the function mode of the driven portable terminal matches the function mode originally intended by the user. The updater 400 controls to update the recognition model.

Update unit 400 is The recognition model database unit 500 is updated according to the determination result of the determination unit 300.

4 is a diagram illustrating modeling of a reinforcement learning theory algorithm applied to an embodiment of the present invention. Referring to FIG. 4, a modeling process of the comparison driver 200 represented as a neural network using a Q-learning algorithm among reinforcement learning theory algorithms will be described. The following description is an embodiment of the present invention, and can be extended to various reinforcement learning algorithms such as SARSA as well as the Q-learning algorithm, and also various function approximators such as a support vector machine as well as a neural network. It can be extended to the Function Approximator) model.

First, a state space may be represented by {x ₁ , ..., x _N , z, θ '}, wherein x ₁ to x _N Is the N sensor input data sensed by the sensing unit 100 and means data about the actual gripping pattern, and z denotes the time taken from the moment of the user's gripping to the operation of the function mode corresponding to the gripping pattern. Delay counter (delay counter), the θ 'means the functional mode of the previous stage of the portable terminal.

In addition, since θ means a function mode which is a kind of action that the portable terminal can presently present, the action space will be {θ}. Θ of the action space may take a ready screen mode immediately before being driven into the function mode, a function mode screen by the ready screen, and a screen mode which is neither a ready screen mode nor a function mode screen. Q denotes a function value by a Q-learning algorithm, and w _i denotes a weight for the i-th input data. And the net _Q Represents the sum of all input elements for a particular function mode. The larger the net _Q is, the higher the probability that the corresponding function mode is driven.

In one embodiment of the present invention, the probability that the function mode is driven may be driven by Equation 1 below.

<Equation 1>

Herein, T denotes a temperature parameter inversely proportional to the number of times the action is performed, and Q (x _1, ..., x _{N ,} z, θ ') is output by a Q-learning algorithm. It means a function value.

Meanwhile, the process of updating the recognition model of the function mode by the updater 400 according to the determination result of the determiner 300 may be updated by Equation 2 below.

<Equation 2>

Here, Δw _i denotes an amount of change in weight for updating the recognition model, γ denotes a depreciation ratio, and α denotes a proportional constant with respect to Equation (2). And, R denotes an instantaneous compensation value. In this case, the R value is set to 0 when the driven function mode is not rejected because it matches the function mode intended by the user, and -1 when it does not match. Can be set to

값으로 결정되고, 상기 z에 대해서는

값으로 결정되고, 상기 θ'에 대해서는

값으로 결정된다. 여기서, 상기 net_j는 상기 j 번째 기능 모드에 대한 가중치 w_j에 대한 수식

에 의해 산출될 수 있다. 여기서, w_j0 는 바이어스 텀(bias term)으로서 상기 net_j 값을 특정하는 역할을 한다. The Γ means a weight for Equation 2, which is input differently depending on which parameter is connected. For x _i , i-th sensor input data detected by the detector 100,

Determined by a value, for z

Value, and for the

It is determined by the value. Here, the net _j is a formula for the weight w _j for the j th function mode

Can be calculated by Here, w _j0 serves to specify the net _j value as a bias term.

Therefore, when Δw _i value calculated by Equation 2 becomes 0, it means that there is no feedback of the rejection of the gripping pattern of the portable terminal, and thus the recognition model for the corresponding function mode is maintained. However, when the Δw _i value calculated by Equation 2 exists, the recognition model is updated. The larger the Δw _i value is, the higher the probability that the user is driven to the function mode to be changed.

As described above, if the continuous determination process of the determination unit 300 and the update process of the update unit 400 using Equation 2 are repeated for the function mode desired by the user, the recognition stored in the recognition model database unit 500 is recognized. The model will be updated to suit your intentions.

On the other hand, the term '~ part' used in the embodiments of the present invention as described above, that is, '~ module' or '~ table' and the like, may include software, a field programmable gate array (FPGA), or an application specific integrated circuit, A hardware component, such as an ASIC, a module performs certain functions. However, modules are not meant to be limited to software or hardware. The module may be configured to be in an addressable storage medium and may be configured to play one or more processors. Thus, as an example, a module may include components such as software components, object-oriented software components, class components, and task components, and processes, functions, properties, procedures, subroutines. , Segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functionality provided within the components and modules may be combined into a smaller number of components and modules or further separated into additional components and modules. In addition, the components and modules may be implemented to reproduce one or more CPUs in a device.

Hereinafter, an operation process according to an embodiment of the present invention will be described with reference to FIG. 5.

5 is a view showing the overall flow of the adaptive learning method of the mobile terminal gripping pattern according to an embodiment of the present invention.

First, when the user grips the portable terminal in a specific gripping pattern (S100), the sensing unit 100 detects a contact area of the user with the portable terminal (S110). In this case, the touch detection sensor 110 in the detection unit 100 outputs a signal indicating whether or not the user touches the grip, and the contact signal processing unit 120 combines the signals to generate the contact information.

The comparison driver 200 compares the contact information with the recognition model in the recognition model database unit 500 that stores a recognition model that is a reference for recognizing a function mode of the portable terminal, and compares the contact information of the portable terminal. The function mode is driven (S120).

On the other hand, the detection unit 100 may detect the angle of the rotation of the mobile terminal or the angle formed by the reference axis with respect to the gravity direction on the basis of a predetermined reference axis set in the mobile terminal, in this case the comparison driving unit 200 ) May additionally use information about the detected angle in recognizing the function mode of the portable terminal.

The determination unit 300 determines whether the driven function mode matches the function mode intended by the user (S130). in this case, The determination may be made by using an input action of the user, which may be determined by shaking the portable terminal or pressing a key button provided in the portable terminal.

According to the determination result, the updater 400 updates the recognition model database 500 (S140). Under the control of the updater 400, the recognition model of the recognition model database unit 500 is updated, and the comparison driving unit 200 newly drives the function mode by reflecting the updated result, using a reinforcement learning theory algorithm. I can drive it. Since the function mode may be driven in proportion to the probability value according to Equation 1, it has already been described above, and thus description thereof will be omitted.

In addition, since the updating process of the recognition model by the updater 400 may be performed by Equation 2, the description thereof will be omitted.

Although embodiments of the present invention have been described above with reference to the accompanying drawings, those skilled in the art to which the present invention pertains have various permutations and modifications without departing from the spirit or essential features of the present invention. It is to be understood that the present invention may be practiced in other specific forms, since modifications may be made. Therefore, it should be understood that the embodiments described above are exemplary in all respects and not restrictive. The scope of the present invention is indicated by the scope of the following claims rather than the detailed description, and all changes or modifications derived from the meaning and scope of the claims and the equivalent concept are included in the scope of the present invention. Should be interpreted.

According to an adaptive learning apparatus and method for a mobile terminal gripping pattern according to an embodiment of the present invention, in the switching of a function mode according to the gripping pattern of the mobile terminal, the recognition model of the gripping pattern is adaptively updated using user feedback. By doing so, there is an effect that can reflect the individual characteristics of the phage pattern.

The effects of the present invention are not limited to the above-mentioned effects, and other effects not mentioned will be clearly understood by those skilled in the art from the description of the claims.

Claims (20)

A detector configured to detect a contact area with the portable terminal as the user grips the portable terminal; A comparison driver configured to drive the functional mode of the portable terminal by comparing the recognition model stored in the recognition model database unit with the information on the contact area as a reference for recognizing the functional mode of the portable terminal; A determination unit which determines whether the driven function mode matches the function mode intended by the user; And an updating unit for updating the recognition model database unit according to the determination result of the determination unit. The method of claim 1, The detection unit, At least one touch sensing sensor configured to output a signal indicating whether the user contacts the grip; And And a contact signal processor for combining the signals to generate information about the contact area. The method of claim 1, The detection unit, As the user grips the portable terminal, the angle at which the portable terminal is rotated or the reference axis is formed in the direction of gravity based on a predetermined reference axis set in the portable terminal, and the comparison driving unit detects the portable terminal. And adaptively use information on the sensed angle in recognizing a functional mode of the terminal. The method of claim 1, The determination unit, And determining whether the driven function mode coincides with a function mode intended by the user, by using the user's input behavior. The method of claim 4, wherein The input action, And shaking the portable terminal or pressing a key button provided in the portable terminal. The method of claim 1, The comparison driving unit, An adaptive learning apparatus for a mobile terminal holding pattern, which drives the function mode using a reinforcement learning theory algorithm. The method of claim 6, The function mode,

Is driven according to a probability value corresponding to, wherein θ is a parameter representing a function mode which is an action that the portable terminal can presently represent, and θ 'is a function of an action immediately preceding the current function mode which is the action indicated by θ. Z is a delay counter indicating the length of time it takes to drive the function mode, and the x ₁ to the x _N Denotes N pieces of input data sensed by the detector, and T denotes a parameter that is inversely proportional to the number of times the action is performed, and Q (x ₁ , ..., x _{N ,} z, θ ' , θ) means a function value by a Q learning algorithm. The method of claim 7, wherein The update unit,

Update the recognition model database unit by using the Δw _i means the amount of change of the weight for the recognition model, R is the instantaneous compensation value, γ is the depreciation ratio, and Γ is Means a weight for a formula, and α denotes a proportionality constant for the formula. The method of claim 8, The R value is, And if the driven function mode matches the function mode intended by the user, sets the value to 0, and sets it to -1 if it does not match. The method of claim 9, The Γ value is, For x _i , i-th input data detected by the detector,

For z,

Is determined by, and for θ '

The net _j is determined by the formula for the weight w _ji when the x _i is input for the j th function mode

An adaptive learning apparatus for a mobile terminal gripping pattern, calculated by. Detecting a contact area with the portable terminal as the user grips the portable terminal; A comparison driving step of driving a functional mode of the portable terminal by comparing the recognition model stored in a recognition model database unit with information on the contact area as a reference for recognizing the functional mode of the portable terminal; A determination step of determining whether the driven function mode matches the function mode intended by the user; And updating the recognition model database unit according to the determination result. The method of claim 11, The detecting step, Outputting a signal indicating whether or not the user contacts the grip; And And combining the signals to generate information on the contact area. The method of claim 11, The detecting step, Detecting an angle of rotation of the mobile terminal or an angle formed by the reference axis with a direction of gravity based on a predetermined reference axis set in the portable terminal as the user grips the portable terminal; And the driving step additionally uses the information on the sensed angle in recognizing the functional mode of the portable terminal. The method of claim 11, The determining step, And determining whether the driven function mode corresponds to a function mode intended by the user, by using the user's input behavior. The method of claim 14, The input action, And shaking the portable terminal or pressing a key button provided in the portable terminal. The method of claim 11, The comparison driving step, Driving the function mode using a reinforcement learning theory algorithm. The method of claim 16, The function mode,

Is driven according to a probability value corresponding to, wherein θ is a parameter representing a function mode which is an action that the portable terminal can presently represent, and θ 'is a function of an action immediately preceding the current function mode which is the action indicated by θ. Z denotes a delay counter indicating a length of time required to drive the function mode, and wherein x is ₁ to x _N. Denotes N pieces of input data detected by the sensing step, and T denotes a parameter inversely proportional to the number of times the action is performed, and Q (x _1, ..., x _{N ,} z, θ ' , θ) means a function value by a Q learning algorithm. The method of claim 17, The updating step,

Updating the recognition model database unit using, wherein Δw _i means the amount of change of the weight for the recognition model, R is the instantaneous compensation value, γ is the depreciation rate, Where Γ is a weight for the formula, α is a proportional constant for the formula, adaptive learning method of a mobile terminal holding pattern. The method of claim 18, The R value is, If the driven function mode matches the function mode intended by the user, set to 0; otherwise, set to -1. The method of claim 19, The Γ value is, For x _i , i-th input data detected by the detector,

For z,

Is determined by, and for θ '

The net _j is determined by the formula for the weight w _ji when the x _i is input for the j th function mode

An adaptive learning method of a mobile terminal gripping pattern, calculated by.

Images