Page clock flipping driving method and system supporting voice interaction control

Abstract

The present application relates to the technical field of page-turning clock, and particularly relates to a page-turning clock turning driving method and system supporting voice interaction control, which collects historical operation data of a driving unit, extracts voice analysis response data to determine a response deviation factor of instruction analysis, quantifies the uncertainty of voice interaction, acquires historical pose deviation data, performs fuzzy mapping by using an error membership function, calculates a deviation membership degree and determines a correction cost of flexible closed-loop feedback, comprehensively feeds back the correction cost and the response deviation factor to dynamically solve a driving compensation coefficient, dynamically regulates the output torque of a page-turning driving module based on the coefficient, realizes full-link adaptive optimization from instruction receiving to action execution, effectively solves the problems of inaccurate instruction response, rigid driving process and large noise of a traditional page-turning clock in a complex sound field, and significantly improves the smoothness, accuracy and system robustness of a page-turning action.

Description

Technical Field

[0001] This invention relates to the field of page turner technology, and in particular to a page turner flipping drive method and system that supports voice interaction control. Background Technology

[0002] As a time display device that combines practicality and aesthetic appeal, the core of a flip clock lies in achieving smooth, accurate, and quiet page-turning action. Traditional flip clocks mostly use mechanical cams or stepper motors for drive, controlling the page-turning action through a preset fixed timing sequence. This results in problems such as rigid drive process, poor response flexibility, and relatively high noise, making it difficult to meet the high requirements of natural human-computer interaction and operational quality in modern smart home environments. With the development of intelligent control technology, some flip clocks have begun to introduce electronic drive and simple feedback mechanisms, but most are still limited to open-loop control or rigid closed-loop adjustment, lacking the ability to adapt to complex operating conditions and external interference.

[0003] Meanwhile, far-field voice interaction technology is widely used in smart devices, providing a contactless and convenient control interface for page-turning clocks. However, far-field voice recognition is susceptible to environmental noise, sound source distance, and speech ambiguity, leading to delays or misjudgments in command parsing, directly affecting the accuracy and consistency of page-turning actions. Furthermore, existing page-turning drive systems typically execute preset action sequences directly upon receiving voice commands, lacking a dynamic perception and compensation mechanism for actual positional deviations during execution, resulting in frequent issues such as incomplete page turning, stuttering, or excessive impact.

[0004] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this invention is to provide a page-turning clock flipping drive method and system that supports voice interaction control. It aims to solve the technical problems of existing page-turning clock drive technology, which is difficult to effectively integrate the uncertainty of far-field voice interaction with the dynamic response characteristics of mechanical execution, lack of adaptive flexible closed-loop feedback mechanism, resulting in instruction parsing deviation, stiff page-turning action, inaccurate posture control and high operating noise under complex working conditions.

[0006] To achieve the above objectives, the present invention provides a method for driving the flip clock to rotate, which supports voice interaction control, the method comprising: Collect the interactive operation records of the drive control unit during the operation of the page-turning clock, and then obtain historical operation data; Extract the interactive response data of each voice parsing unit of the page-turning clock voice interaction module when processing far-field voice commands from the historical operation data, and then determine the response deviation factor in the command parsing process through all the interactive response data; Historical pose deviation data during the page-turning continuously variable drive process is obtained from the historical operation data. The error membership function in the function library is used to perform fuzzy mapping on the historical pose deviation data to obtain the deviation membership degree corresponding to each input quantity in the continuously variable drive control logic. Then, the feedback correction cost of the flexible closed-loop feedback mechanism is determined based on all the deviation membership degrees. The drive compensation coefficient when the page-turning clock performs stepless page-turning drive is determined by the feedback correction cost and the response deviation factor. The output torque of each page-turning drive module of the page-turning clock is dynamically adjusted based on the drive compensation coefficient.

[0007] Optionally, determining the response deviation factor during the instruction parsing process using all interactive response data includes: Extract the environmental noise features of far-field voice interaction during each command parsing in all interactive response data, and then obtain the noise feature vector; The ambiguity of command recognition after processing by the corresponding speech parsing unit is determined based on each interactive response data. The response deviation factor for parsing the page-turning clock voice commands is determined by the noise feature vector and all command recognition ambiguities.

[0008] Optionally, the step of using the error membership function from the function library to perform fuzzy mapping on the historical pose deviation data to obtain the deviation membership degree corresponding to each input quantity in the continuously variable drive control logic includes: All drive output deviations in the historical pose deviation data are aggregated and divided into multiple fuzzy deviation intervals. Each driving deviation in the historical pose deviation data is mapped to the corresponding deviation fuzzy interval to obtain the deviation mapping domain; By mapping and judging each driving deviation using the deviation mapping domain and the error membership function in the function library, the deviation membership degree corresponding to each driving deviation in the historical pose deviation data is obtained.

[0009] Optionally, the step of determining the feedback correction cost of the flexible closed-loop feedback mechanism based on all deviation membership degrees includes: The deviation bound of the output pose is determined by all the deviation memberships; Typical deviation values ​​are selected from all deviation memberships, and then the typical deviation values ​​are used to determine the typical pose values ​​of the output pose after the page turning action is completed. The feedback correction cost of the flexible closed-loop feedback mechanism is obtained by using the typical pose value and the deviation boundary.

[0010] Optionally, determining the drive compensation coefficient when the page-turning clock performs stepless page-turning drive using the feedback correction cost and the response deviation factor includes: Acquire real-time pose records during the far-field voice-triggered page-turning action; For each driven pose in the real-time pose record, the historical reference pose corresponding to the driven pose is obtained from the historical pose deviation data; the current actual pose corresponding to the driven pose is obtained from the real-time pose record, and then the pose deviation between the historical reference pose and the current actual pose is determined. The torque hysteresis of the page-turning drive axis is determined by the feedback correction cost and the pose deviation, and then the torque hysteresis of each drive pose in the real-time pose record is obtained. The drive compensation coefficient during the continuously variable drive process of the page-turning clock is determined based on all torque hysteresis and the response deviation factor.

[0011] Optionally, the dynamic adjustment of the output torque of each page-turning drive module of the page-turning clock based on the drive compensation coefficient includes: The torque is compensated for each driving pose in the real-time pose record by the driving compensation coefficient to obtain all the compensated output torques. The output torque of the drive control unit is adjusted based on the drive error in each torque compensation and the corresponding output torque after compensation during the stepless drive of the page-turning clock.

[0012] Optionally, the step of performing torque compensation on each driving pose in the real-time pose record using the driving compensation coefficient to obtain all compensated output torques includes: For each driven pose in the real-time pose record, the driving error is determined based on the driving compensation coefficient and the current pose deviation. The initial output torque is compensated using the drive error to obtain the compensated output torque, and then all compensated output torques are determined.

[0013] Furthermore, to achieve the above objectives, the present invention also provides a page-turning clock flipping drive system supporting voice interaction control, the system comprising: The data acquisition module is used to collect the interactive operation records of the drive control unit during the operation of the page-turning clock, thereby obtaining historical operation data; The voice evaluation module is used to extract the interaction response data of each voice parsing unit of the page-turning clock voice interaction module when processing far-field voice commands in the historical operation data, and then determine the response deviation factor in the command parsing process through all the interaction response data. The fuzzy mapping module is used to obtain historical pose deviation data in the page-turning continuously variable drive process from the historical running data, and to perform fuzzy mapping on the historical pose deviation data using the error membership function in the function library to obtain the deviation membership degree corresponding to each input quantity in the continuously variable drive control logic. Then, based on all the deviation membership degrees, the feedback correction cost of the flexible closed-loop feedback mechanism is determined. The coefficient calculation module is used to determine the drive compensation coefficient when the page-turning clock performs stepless page-turning drive by using the feedback correction cost and the response deviation factor; The torque control module is used to dynamically control the output torque of each page-turning drive module of the page-turning clock based on the drive compensation coefficient.

[0014] Furthermore, to achieve the above objectives, the present invention also provides a page-turning clock flipping drive device supporting voice interaction control. The device includes: a memory, a processor, and a page-turning clock flipping driver program supporting voice interaction control stored in the memory and executable on the processor. The page-turning clock flipping driver program supporting voice interaction control is configured to implement the steps of the page-turning clock flipping drive method supporting voice interaction control as described above.

[0015] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing a page-turning clock driver program that supports voice interaction control. When the page-turning clock driver program that supports voice interaction control is executed by a processor, it implements the steps of the page-turning clock driver program that supports voice interaction control as described above.

[0016] This invention provides a method for driving the flip clock that supports voice interaction control. This method extracts the interactive response data from the voice parsing unit and calculates the response deviation factor. This method quantifies the impact of external factors such as environmental noise and distance attenuation on voice command recognition. This allows the system to dynamically adjust the driving strategy based on the credibility of the command, rather than blindly executing commands. This effectively reduces erroneous page-turning actions caused by misrecognition or ambiguous commands, improving the reliability of human-computer interaction. Traditional flip clocks often suffer from stiff movements or impact noise due to fixed torque output. This solution utilizes error membership functions and fuzzy mapping to handle pose deviations, transforming rigid position errors into flexible feedback correction costs. Combined with a driving compensation coefficient, the output torque is dynamically adjusted, allowing the page-turning action to adjust its force in real time according to actual resistance and smoothness requirements. This achieves stepless speed control from start to stop, significantly reducing operating noise and extending the lifespan of the mechanical structure. This method innovatively integrates the uncertainty of voice interaction (response deviation factor) with the dynamism of mechanical execution (feedback correction cost), jointly determining the final driving compensation coefficient. This means the system can not only correct positional deviations during mechanical execution, such as card jamming or changes in friction, but also sense the quality of upstream commands. This dual-dimensional closed-loop feedback mechanism enables the page-turning clock to maintain high-precision positional control and stable operation even under varying environmental temperatures, humidity levels, or aging conditions. Dynamic control based on historical operating data and fuzzy logic avoids overdrive or repeated trial-and-error phenomena common in traditional control, providing precisely matched torque compensation only when needed. This ensures accurate page-turning while avoiding unnecessary energy waste, thus improving the overall system's energy efficiency ratio. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating an embodiment of the page-turning clock flipping drive method supporting voice interaction control according to the present invention; Figure 2 This is a structural block diagram of an embodiment of the page-turning clock flipping drive system that supports voice interaction control according to the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] Reference Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the page-turning clock flipping drive method supporting voice interaction control according to the present invention. An embodiment of the page-turning clock flipping drive method supporting voice interaction control according to the present invention is presented.

[0021] In one embodiment, the method for driving the flip clock with voice interaction control includes: Step S100: Collect the interactive operation records of the drive control unit during the operation of the page-turning clock, and then obtain historical operation data.

[0022] The drive control unit can be a central control component responsible for coordinating the operation of various functional modules of the page-turning clock and executing drive strategies. It can be used to uniformly manage the logical relationships between voice interaction, pose feedback, and torque output, achieving dynamic drive regulation. In this embodiment, the drive control unit can run control algorithms through an embedded processor, receive sensor data and interaction commands, and output drive signals. Furthermore, the drive control unit can exchange command parsing results with the voice interaction module and transmit torque control parameters with the page-turning drive module. The interactive operation records of the drive control unit during the page-turning clock's operation can be various interactive events and status parameters recorded by the drive control unit during operation. These can be used as the original source of historical operation data for subsequent feature extraction and model training. Historical operation data can be a dataset containing voice interaction and mechanical execution information accumulated by the system in past operations, which can be used to support offline or online calculations of response deviation factors and feedback correction costs.

[0023] Collecting the interaction operation records of the drive control unit during the operation of the page-turning clock can be achieved by continuously recording the interaction events and status parameters between the drive control unit and each module during system operation. For example, the collection of the interaction operation records of the drive control unit during the operation of the page-turning clock can be achieved by writing log files to storage media, caching in real time through a memory buffer, and uploading periodically, thereby forming a historical operation data foundation that can be used for subsequent analysis.

[0024] Step S200: Extract the interaction response data of each voice parsing unit of the page-turning clock voice interaction module when processing far-field voice commands from the historical operation data, and then determine the response deviation factor in the command parsing process through all the interaction response data.

[0025] The voice interaction module can be a hardware and software integrated unit that processes far-field voice input and converts it into executable instructions. It can provide a contactless human-computer interaction interface, supporting users to control page-turning actions via natural language. In an exemplary embodiment, the voice interaction module may include a voice acquisition submodule, a voice recognition submodule, and a semantic parsing submodule. Far-field voice commands can be voice control commands issued by the user at a certain distance, requiring noise reduction and enhancement processing. They can be used as the trigger source for page-turning actions, and their quality directly affects the system's response accuracy. The voice parsing unit can be the processing unit in the voice interaction module responsible for converting the raw voice signal into structured instructions. It can be used to output the instruction content and its confidence-related parameters as the basis for calculating the response deviation factor. Furthermore, the voice parsing unit can decode and semantically map voice features based on acoustic and language models. In a specific embodiment, the voice parsing unit may include an acoustic decoding unit, a language understanding unit, and an intent classification unit.

[0026] Interactive response data can be intermediate and final output parameters generated by the speech parsing unit during instruction processing. It can reflect uncertainties in the speech recognition process and quantify the impact of external interference. The response deviation factor can be a quantitative indicator characterizing the degree of environmental interference affecting the speech instruction recognition result. It can be used to measure instruction credibility and guide whether and how strongly the driving strategy is executed. In this embodiment, the response deviation factor can be obtained by normalizing or weighting the interactive response data (such as confidence level, signal-to-noise ratio, and keyword matching degree).

[0027] Extracting the interactive response data of each speech parsing unit in the page-turning clock speech interaction module when processing far-field speech commands from historical operating data can be achieved by filtering records related to speech interaction from historical operating data and parsing the output parameters of each speech parsing unit. For example, this operation can be achieved by associating speech events with parsing results by timestamp and extracting corresponding response data by command ID index, thereby obtaining raw features reflecting speech recognition quality and providing a basis for calculating the response deviation factor. Determining the response deviation factor in the command parsing process through all interactive response data can be achieved by performing statistical analysis or model inference on the interactive response data to generate a single value representing the credibility of the command. Furthermore, this operation can be achieved by using a weighted average method to fuse multiple confidence indicators and using a machine learning model to predict the deviation level, thereby quantifying the impact of environmental noise, distance attenuation, and other factors on speech recognition.

[0028] Step S300: Obtain historical pose deviation data during the page-turning continuously variable drive process from historical operating data; use the error membership function in the function library to perform fuzzy mapping on the historical pose deviation data to obtain the deviation membership degree corresponding to each input quantity in the continuously variable drive control logic; and then determine the feedback correction cost of the flexible closed-loop feedback mechanism based on all deviation membership degrees.

[0029] The continuously variable drive (CVD) page-turning process can be a drive mode where the output torque is continuously adjustable and the speed changes smoothly during the page-turning action. This can be used to avoid the impact and noise caused by fixed torque, achieving smooth motion from start to stop. In an exemplary embodiment, the CVD page-turning process can include drive states corresponding to acceleration, constant speed, and deceleration phases. Historical pose deviation data can be historical error records between the actual pose and the target pose during the CVD page-turning process. This data can be used as input to the error membership function to construct a fuzzy feedback model. The error membership function in the function library can be a collection of various error membership function instances pre-stored in the system. This data can be used to provide optional fuzzy mapping tools to adapt to the deviation processing requirements under different working conditions. The error membership function can be a mathematical function that maps pose deviation values ​​to membership degrees in a fuzzy set. This function can be used to convert rigid errors into continuous membership degree values ​​that can be used for fuzzy inference. In this embodiment, the error membership function can perform a nonlinear transformation on the input deviation based on a preset fuzzy rule base.

[0030] The inputs in the continuously variable drive (CVD) control logic can be various state variables or parameters participating in the CCD decision-making process. These can form the input space of a flexible closed-loop feedback mechanism, influencing the calculation of deviation membership degrees. Deviation membership degrees can be fuzzy set membership values ​​obtained by mapping historical pose deviations through an error membership function. These can be used as inputs for fuzzy inference, participating in the generation of feedback correction costs. The flexible closed-loop feedback mechanism can be a feedback control structure that dynamically adjusts drive parameters based on fuzzy mapping results. It can be used to achieve compliant compensation for mechanical execution deviations, avoiding overshoot or oscillations caused by hard corrections. Furthermore, the flexible closed-loop feedback mechanism can receive deviation membership degrees as input and output feedback correction costs for calculating drive compensation coefficients. The feedback correction cost of the flexible closed-loop feedback mechanism can be the intensity of mechanical execution correction demand derived from a comprehensive evaluation of deviation membership degrees. This can be used to quantify the influence weight of the current pose deviation on the drive output, for calculating drive compensation coefficients.

[0031] Obtaining historical pose deviation data during the continuously variable page-turning process from historical operational data can be achieved by extracting the difference sequence between the actual pose and the target pose during page-turning execution from historical operational data. In a specific embodiment, this operation can be implemented by calculating the deviation through encoder feedback and comparing the target with the actual position through a visual sensor, thereby obtaining historical samples of mechanical execution deviations for constructing a fuzzy feedback model. Fuzzy mapping of the historical pose deviation data using error membership functions from a function library can be achieved by inputting the historical pose deviation data into a preset error membership function and outputting the corresponding membership degree value. Further, this operation can be implemented by selecting a triangular membership function for linear mapping or a Gaussian membership function for nonlinear smooth mapping, thereby converting discrete or continuous error values ​​into membership degree representations that can be processed by fuzzy logic. Determining the feedback correction cost of the flexible closed-loop feedback mechanism based on all deviation membership degrees can be achieved by integrating multiple deviation membership degrees and generating the correction cost through fuzzy rules or weighted summation. For example, this operation can be implemented by selecting the dominant deviation level using the maximum membership principle and calculating the comprehensive correction value using the center of gravity method, thereby forming a comprehensive assessment of the current mechanical state deviation to drive compensation decisions.

[0032] Step S400: Determine the drive compensation coefficient when the page-turning clock performs stepless page-turning drive by using feedback correction cost and response deviation factor.

[0033] The drive compensation coefficient can be a torque adjustment factor generated by integrating the credibility of the voice command and the mechanical execution state. It can be used as a scaling parameter for the final torque output to achieve dual-dimensional adaptive drive control. In this embodiment, the drive compensation coefficient can be obtained by weighted fusion or fuzzy rule inference of the response deviation factor and the feedback correction cost. Determining the drive compensation coefficient when the page-turning clock performs stepless page-turning drive by using the feedback correction cost and the response deviation factor can be achieved by fusing the feedback correction cost and the response deviation factor to generate the final drive compensation coefficient. Furthermore, this operation can be implemented by emphasizing dual constraints through product fusion and using a fuzzy inference engine for rule-based synthesis, thereby enabling joint decision-making between the uncertainty of the voice layer and the dynamics of the execution layer.

[0034] Step S500: Dynamically adjust the output torque of each page-turning drive module of the page-turning clock based on the drive compensation coefficient.

[0035] The page-turning drive module can be an electromechanical actuator that directly drives the page-turning mechanism to complete physical actions. It can be used to convert electrical signals into mechanical page-turning actions, and its output torque is dynamically adjusted by a drive compensation coefficient. In an exemplary embodiment, the page-turning drive module may include a DC brushed motor module, a stepper motor module, a brushless motor module, etc. The output torque of the page-turning drive module can be the actual rotational torque applied to the page-turning mechanism by the page-turning drive module, which can be used to directly determine the speed, force, and smoothness of the page-turning action, and is dynamically adjusted by a drive compensation coefficient. Dynamically controlling the output torque of each page-turning drive module of the page-turning clock based on the drive compensation coefficient can be achieved by using the drive compensation coefficient as a scaling factor to adjust the current or voltage output of the page-turning drive module in real time. For example, this operation can be achieved by adjusting the motor drive voltage through the PWM duty cycle and adjusting the output torque through the current loop setting value, so that the page-turning action force can adaptively change with the command quality and mechanical state, achieving stepless speed regulation and silent operation.

[0036] Taking voice-controlled page turning in a high-noise home environment as an example, the page-turning clock flipping drive method supporting voice interaction control in this embodiment can be as follows: The user issues a voice command to turn to tomorrow in the living room where a TV program is playing. Due to background noise, the confidence of the keywords in the voice interaction module is low, and the interactive response data output by the voice parsing unit shows a low matching score. Based on this, the system calculates a high response deviation factor. At the same time, the page-turning drive module detects that the pose deviation increases due to slight card jamming. After error membership function mapping, a medium deviation membership degree is generated, and then the feedback correction cost is obtained. After the two are combined, the drive compensation coefficient is appropriately reduced. The page-turning drive module starts slowly with a low initial torque. After confirming that the card has successfully disengaged from the jamming, the torque is gradually increased to complete the page turning, avoiding erroneous actions or mechanical damage caused by misjudged commands or forced drive.

[0037] In one embodiment, the response deviation factor during the instruction parsing process is determined using all interactive response data, including: Extract the environmental noise features of far-field voice interaction during each command parsing in all interactive response data, and then obtain the noise feature vector; The ambiguity of command recognition after processing by the corresponding speech parsing unit is determined based on each interactive response data. The response bias factor for parsing the page-turning clock's voice commands is determined by using noise feature vectors and all command recognition ambiguities.

[0038] The noise feature vector can be a multi-dimensional numerical vector representing the environmental noise characteristics at each moment of voice command parsing. It can be used to quantify the intensity and type of external acoustic interference on the speech recognition process, serving as one of the components of the response deviation factor. In this embodiment, the noise feature vector can be extracted from the interactive response data by noise-related acoustic parameters (such as spectral energy distribution, signal-to-noise ratio, and background noise type identifier) ​​and combined into a vector form. Furthermore, the noise feature vector can be an external interference characterization variable, participating in the construction of the response deviation factor together with the command recognition ambiguity. For example, the noise feature vector can include, but is not limited to, one or more of the following: time-domain noise energy features, frequency-domain masking effect features, and spatial reverberation features. Extracting the environmental noise features of far-field voice interaction at each command parsing in all interactive response data can be achieved by filtering parameters related to noise perception from the interactive response data, aligning them according to the command parsing time point, and structuring them into a feature set. Furthermore, this operation can be achieved by extracting the real-time signal-to-noise ratio and noise classification labels from the voice front-end module and analyzing the background noise spectrum through the residual signal after microphone array beamforming, thereby obtaining a quantifiable description of environmental interference to support the construction of the noise feature vector.

[0039] Command recognition ambiguity can be a quantitative indicator measuring the uncertainty of the output command result of the speech parsing unit. It can be used to reflect the reliability of the internal processing results of speech recognition and to assess whether the corresponding page-turning action should be executed or weakened. In an exemplary embodiment, command recognition ambiguity can be calculated based on parameters such as the confidence score of the speech parsing unit, the number of candidate intents, and the keyword matching ambiguity. In a specific embodiment, command recognition ambiguity can be an internal recognition quality representation variable in the speech parsing unit, which is input together with the noise feature vector into the response bias factor generation mechanism. For example, command recognition ambiguity can include, but is not limited to, one or more of semantic ambiguity ambiguity, pronunciation variation ambiguity, and lexical boundary ambiguity. Determining the command recognition ambiguity of the speech interaction module after processing by the corresponding speech parsing unit based on each interaction response data can be achieved by parsing the intermediate and final results output by each speech parsing unit, extracting multiple indicators reflecting recognition uncertainty, and fusing them into a single ambiguity value. Furthermore, this operation can be achieved by statistically analyzing the score differences between N-best candidate intents to measure the degree of ambiguity and combining the perplexity of the language model to assess the rationality of the statement, thereby enabling fine-grained evaluation of the semantic reliability of speech commands and avoiding reliance on a single confidence threshold for judgment.

[0040] The response deviation factor for parsing the page-turning clock voice command is determined by using the noise feature vector and all command recognition ambiguities as inputs. This can be achieved by generating the response deviation factor through weighted fusion, lookup table mapping, or fuzzy inference. In one embodiment, the formula for calculating the response deviation factor can be as follows: In the formula, For response bias factor; The modulus of the environmental noise feature vector represents the noise level. The larger the modulus, the stronger the background noise (such as interference from other electrical appliances or human voices), and the worse the quality of the speech signal collected by the microphone, which directly affects the initial accuracy of speech recognition. The mean of the ambiguity in command recognition represents the uncertainty of recognition. Even if the user speaks clearly, if there is ambiguity in the language or unclear pronunciation, the parsing result will be ambiguous. This parameter reflects the confidence level at the algorithm level. , This is a weighting coefficient used to balance the impact of noise and recognition ambiguity on the total bias. For example, it may be increased in outdoor scenes. They value ambient noise more, and might turn up the volume in a quiet environment like a library. Furthermore, this operation places greater emphasis on semantic clarity. It can be further implemented by using a pre-trained regression model to map the two types of features into deviation factors, and by using a fuzzy rule base to determine the deviation level based on a combination of noise level and fuzziness. This allows for the construction of a comprehensive credibility index that integrates external interference and internal recognition quality, providing a preliminary judgment basis for driving strategies.

[0041] Taking voice-controlled page turning under multi-source noise interference as an example, the page-turning clock flipping method supporting voice interaction control in this embodiment can be as follows: When the user is cooking in the kitchen, they issue a voice command to turn to the day after tomorrow. At the same time, the range hood and faucet generate broadband noise. The voice interaction module records the high background noise energy and the low-frequency dominant noise feature vector. The voice parsing unit outputs multiple high-scoring candidates because the keywords "the day after tomorrow" and "tomorrow" are similar in pronunciation, which increases the ambiguity of command recognition. The system fuses the noise feature vector with the high ambiguity value to generate a higher response deviation factor. Based on this, the drive control unit temporarily suspends the page turning action and triggers a voice confirmation prompt. After the user confirms a second time, the page turning is completed with a conservative torque, which effectively avoids misoperation caused by the combined effect of environmental noise and semantic ambiguity.

[0042] In one embodiment, the error membership function from the function library is used to perform fuzzy mapping on the historical pose deviation data to obtain the deviation membership degree corresponding to each input quantity in the continuously variable drive control logic, including: All drive output deviations in the historical pose deviation data are aggregated and divided into multiple fuzzy deviation intervals. The drive output deviation can be the difference between the actual output pose and the target pose of the drive module during execution. It can serve as a basic unit for constructing fuzzy mapping inputs, representing historical pose deviation data. In an exemplary embodiment, the drive output deviation may include, but is not limited to, one or more of positive overshoot deviation, negative hysteresis deviation, and zero steady-state deviation. Multiple deviation fuzzy intervals can be a set of non-mutually exclusive intervals formed by semantically dividing the numerical range of the drive output deviation. These intervals can provide a semantic classification basis for subsequent fuzzy mapping, making the numerical deviations interpretable. Furthermore, the boundaries of the multiple deviation fuzzy intervals can be set based on historical data distribution or engineering experience, with each interval corresponding to a deviation semantic label. For example, the multiple deviation fuzzy intervals can be small deviation intervals, medium deviation intervals, and severe deviation intervals.

[0043] Data aggregation of all drive output deviations in historical pose deviation data can be achieved by extracting all drive output deviation values ​​from the historical pose deviation data and integrating them into a unified dataset. Furthermore, this aggregation can be achieved by aggregating deviation data by time window or grouping by page-turning action type, thus forming a structured deviation sample pool, facilitating subsequent interval partitioning and mapping. Multiple fuzzy deviation intervals can be obtained by dividing the range of drive output deviation values ​​into several semantic intervals based on preset rules or statistical distributions. Further, these multiple fuzzy deviation intervals can be obtained by partitioning intervals based on equal-width binning strategies and determining interval boundaries based on clustering results, thereby assigning semantic meaning to numerical deviations and supporting subsequent fuzzy logic reasoning.

[0044] Each driving deviation in the historical pose deviation data is mapped to the corresponding deviation fuzzy interval to obtain the deviation mapping domain; Each driving deviation can be a single, specific instance of the driving output deviation in the historical pose deviation data, serving as the smallest processing unit for fuzzy mapping and participating in membership calculation. The deviation mapping domain can be a set of mapping relationships between all driving deviations and their respective deviation fuzzy intervals, used to establish the association between numerical deviations and semantic intervals, providing the scope for the error membership function. In a specific embodiment, the deviation mapping domain can serve as the input context for the error membership function, limiting the applicable scope of the mapping judgment. Mapping each driving deviation in the historical pose deviation data to its corresponding deviation fuzzy interval can involve determining which one or more deviation fuzzy intervals each driving deviation belongs to. Furthermore, mapping each driving deviation in the historical pose deviation data to its corresponding deviation fuzzy interval can be achieved by using hard partitioning to only belong to the nearest neighbor interval and soft partitioning to allow simultaneous belonging to multiple intervals, thereby establishing a preliminary association between specific deviation values ​​and semantic labels. The resulting deviation mapping domain can be a summary of the mapping relationships between all driving deviations and their corresponding deviation fuzzy intervals. Furthermore, the obtained deviation mapping domain can be realized by storing the mapping relationship in key-value pairs and representing the many-to-many membership structure in matrix form, thereby forming a complete fuzzy classification framework as the basis for the application of membership functions.

[0045] By mapping and judging each driving deviation using the deviation mapping domain and the error membership function in the function library, the deviation membership degree corresponding to each driving deviation in the historical pose deviation data is obtained.

[0046] Mapping and determining each driving deviation using the deviation mapping domain and error membership functions in the function library can be achieved by calling the corresponding error membership function to calculate the membership degree of the driving deviation within the interval defined by the deviation mapping domain. Furthermore, mapping and determining each driving deviation using the deviation mapping domain and error membership functions in the function library can be achieved by calculating the membership degree for each interval separately and then weighted and fused, or by calculating a single membership degree only within the main interval. This allows for precise quantization from semantic intervals to continuous membership degree values. Specifically, the expression for the deviation membership degree can be: In the formula, is the deviation membership degree corresponding to the i-th driving deviation, used to convert the physical error into a decision weight between 0 and 1, which is used to calculate the feedback correction cost in the subsequent calculation. The i-th historical pose deviation value is the input variable of the system, which is the difference between the actual position and the theoretical position after a page-turning action. This is the error source of the control system. The more accurate the data, the more effective the feedback. The center of the fuzzy deviation interval, i.e., the typical deviation value, defines the ideal state or the typical deviation state. The smaller the value, the more the current state conforms to a typical pattern; The width parameter determines the coverage of the membership function, controlling the fault tolerance range or sensitivity of the controller; The larger the value, the flatter the membership function curve, indicating a higher tolerance for deviations and a stronger degree of fuzziness in the system. The smaller the value, the sharper the curve, and the more sensitive the system is to minute deviations. The deviation membership degree corresponding to each driving deviation in the historical pose deviation data can be obtained. It can be the membership degree value in the interval [0, 1] after fuzzy mapping, so that rigid error can be transformed into a flexible variable that can be used for fuzzy control, supporting the generation of feedback correction cost.

[0047] For example, in a page-turning scenario where increased friction is caused by mechanical aging, the page-turning clock flipping drive method supporting voice interaction control in this embodiment can be as follows: As usage time increases, the drive mechanism experiences increased starting resistance due to reduced lubrication, resulting in a 0.8 mm positional lag during a page turn; the system extracts this drive output deviation from historical pose deviation data and categorizes it into the overlapping area of ​​the moderate and severe deviation ranges; the error membership function data calculates a membership degree of 0.6 in the moderate deviation range and 0.3 in the severe deviation range; after comprehensive analysis, a higher feedback correction cost is generated, and the drive compensation coefficient is correspondingly increased. The drive module applies a larger initial torque in advance during the next action to ensure the card flips smoothly into position and avoids page-turning failure due to insufficient compensation.

[0048] In one embodiment, the feedback correction cost of the flexible closed-loop feedback mechanism is determined based on all deviation memberships, including: The deviation bounds of the output pose are determined by using all the deviation memberships.

[0049] The deviation boundary of the output pose after the page-turning action can be the maximum possible deviation range of the actual pose from the target pose at the moment of page-turning completion, determined based on all deviation membership degrees. This can be used to provide an upper limit for error tolerance in feedback correction, avoiding overcompensation for small or reasonable deviations. In this embodiment, the deviation boundary of the output pose after the page-turning action can be obtained by performing boundary analysis (e.g., taking the extreme values ​​of the non-zero intervals of membership degrees) or confidence interval estimation on all deviation membership degrees. Furthermore, the deviation boundary of the output pose after the page-turning action can participate in the construction of the feedback correction cost in conjunction with the typical pose value. For example, the deviation boundary of the output pose after the page-turning action can include, but is not limited to, one or more of the following: positive overshoot boundary, negative hysteresis boundary, and comprehensive absolute deviation boundary. Determining the deviation boundary of the output pose after the page-turning action by analyzing the support set or effective interval of all deviation membership degrees can determine the upper and lower limits of the pose deviation. Furthermore, by determining the deviation boundary of the output pose after the page-turning action is completed through all the deviation membership degrees, the deviation boundary can be determined by taking the deviation range with membership degrees greater than a preset threshold as the deviation boundary and using the α-cut of fuzzy sets. This can establish an error tolerance boundary and provide a safe threshold basis for setting the subsequent correction intensity.

[0050] Typical deviation values ​​are selected from all deviation memberships, and then the typical deviation values ​​are used to determine the typical pose values ​​of the output pose after the page-turning action is completed.

[0051] The typical deviation value can be the most representative membership sample selected from all deviation memberships, or its corresponding original deviation value. It can be used to reflect the most frequently occurring deviation patterns in historical execution and to derive typical pose states. In an exemplary embodiment, the typical deviation value can extract the central trend from the fuzzy set through methods such as clustering, mode extraction, maximum membership corresponding point, or weighted averaging. Furthermore, the typical deviation value can be combined with the target pose to generate a typical pose value. In a specific embodiment, the typical deviation value may include, but is not limited to, the maximum membership corresponding deviation, the fuzzy mean deviation, and the fuzzy median deviation. Selecting the typical deviation value from all deviation memberships can be achieved by identifying statistically representative or semantically significant sample points from the deviation membership set. Further, selecting the typical deviation value from all deviation memberships can be achieved by selecting the deviation with the highest membership as the typical value and calculating the expected deviation after weighting the memberships as the typical value, thereby extracting the core patterns of historical deviations and avoiding the use of outliers caused by noise interference.

[0052] The typical pose value of the output pose after the page-turning action can be the most likely actual pose state after the page-turning action, obtained by mapping the typical deviation value. It can be used as a benchmark for the system's expected execution result and participate in the construction of feedback correction costs. In this embodiment, the typical pose value of the output pose after the page-turning action can be superimposed on the typical deviation value to the target pose, or converted into a specific angle / position value through an inverse kinematics model. For example, the typical pose value of the output pose after the page-turning action can include, but is not limited to, typical angle values, typical linear displacement values, and typical multi-degree-of-freedom joint pose values. Determining the typical pose value of the output pose after the page-turning action through the typical deviation value can be achieved by synthesizing the typical deviation value with the target pose to obtain the predicted actual output pose. Furthermore, determining the typical pose value of the output pose after the page-turning action through the typical deviation value can be achieved by directly superimposing the typical deviation value onto the target pose or by inversely deriving the typical pose value through a system dynamics model, thereby forming a typical state estimate of the execution result and enhancing the predictability of the feedback mechanism.

[0053] The feedback correction cost of the flexible closed-loop feedback mechanism is obtained by using the typical pose value and the deviation boundary.

[0054] The feedback correction cost of the flexible closed-loop feedback mechanism can be obtained by combining the typical pose values ​​and the deviation bounds, and generating the correction cost through function mapping or rule inference. Furthermore, the feedback correction cost of the flexible closed-loop feedback mechanism can be calculated by using the degree of deviation of the typical pose values ​​from the target as the main weight and the deviation bounds as the constraint; or by using a fuzzy rule base to output the correction level based on a combination of both. This allows the feedback correction cost to simultaneously include typical behavioral trends and boundary constraint information, improving compensation accuracy and stability.

[0055] For example, in an adaptive compensation scenario where mechanical aging causes incomplete page turning, the page-turning clock flipping drive method supported by voice interaction control in this embodiment can be as follows: After long-term use, the friction of the transmission mechanism of the page-turning clock increases, and historical posture deviation data shows that most page-turning actions stop about 2 degrees before the target position, and the deviation distribution is concentrated in the range of 1 to 3 degrees; the system determines the deviation boundary as 0 to 4 degrees through the deviation membership degree, selects the 2 degrees with the highest membership degree as the typical deviation value, and then obtains the typical posture value as the target minus 2 degrees; the feedback correction cost is set to medium to high accordingly, and the drive compensation coefficient is increased accordingly to ensure that the card is completely flipped to the correct position without triggering overshoot, thus avoiding gradual failure due to aging accumulation.

[0056] In one embodiment, the drive compensation coefficient for the page-turning clock when performing stepless page-turning is determined by the feedback correction cost and the response deviation factor, including: Acquire the real-time pose record during the far-field voice-triggered page-turning action.

[0057] Real-time pose recording refers to the sequence of driving poses collected in real time by sensors during the page-turning action. This sequence provides dynamic data of the current execution state, which can be compared with historical reference poses to calculate real-time deviations. In this embodiment, real-time pose recording can be obtained by continuously sampling position and angle information during the page-turning process using an encoder, Hall sensor, or visual feedback device. For example, real-time pose recording can include, but is not limited to, one or more of angular pose recording, linear displacement recording, and composite motion trajectory recording. Obtaining real-time pose recording during the far-field voice-triggered page-turning action can be achieved by activating a pose sensor to continuously sample and record the page-turning driving process after the voice command is triggered. Further, this operation can be implemented by using a rotary encoder to collect angle data at a fixed frequency and by using a magnetic position sensor to obtain discrete keypoint poses, thereby establishing a dynamic state baseline for the current execution process and providing a data foundation for real-time deviation analysis.

[0058] For each driven pose in the real-time pose record, the historical reference pose corresponding to the driven pose is obtained from the historical pose deviation data; the current actual pose corresponding to the driven pose is obtained from the real-time pose record, and then the pose deviation between the historical reference pose and the current actual pose is determined.

[0059] The historical reference pose can be an ideal or typical pose value corresponding to the current driving pose in historical pose deviation data. It can be used as a reference standard to evaluate the accuracy of the current execution and to generate the pose deviation. In an exemplary embodiment, the historical reference pose can be obtained by retrieving a reference pose that matches the current driving pose stage from historical operation data. Furthermore, the historical reference pose can be input together with the current actual pose into the pose deviation calculation process. The current actual pose can be the actual physical position or angle of the driving mechanism at a certain moment during the page-turning action. It can be used to reflect the true state of the mechanical system under disturbance and is the practical basis for deviation calculation. In one embodiment, the current actual pose can be obtained by extracting the measurement value at the corresponding time point from the pose sensor in the real-time pose record. The pose deviation can be the difference between the historical reference pose and the current actual pose. It can be used to quantify the instantaneous error in the current execution process and to dynamically generate the torque hysteresis. For example, the pose deviation can be obtained by calculating the degree of deviation between the two through vector difference or angle difference.

[0060] For each driving pose in the real-time pose record, the historical reference pose corresponding to that driving pose is obtained from the historical pose deviation data. This can be achieved by matching the timestamp or stage label of the current driving pose with the historical data index to extract the corresponding reference value. Furthermore, this operation can be implemented by matching the same stage in the page-turning cycle based on a phase alignment strategy and generating the corresponding reference pose from the historical trajectory through interpolation, thereby achieving alignment between the current state and the historical baseline and ensuring the comparability of the deviation calculation. The current actual pose corresponding to that driving pose is obtained from the real-time pose record, which can be achieved by extracting the specific pose value by index or time point from the acquired real-time pose record. Furthermore, this operation can be achieved by directly reading the sensor's raw output and the smoothed pose value after filtering, thereby obtaining the actual execution data for deviation calculation. The pose deviation between the historical reference pose and the current actual pose is determined by calculating the algebraic difference or Euclidean distance between the two in the angular or positional dimensions. Furthermore, this operation can be implemented using absolute deviation calculation and a directional deviation representation with a direction sign, thereby generating a quantitative indicator reflecting the degree of deviation in the current execution.

[0061] The torque hysteresis of the page-turning drive axis is determined by the feedback correction cost and the pose deviation, and then the torque hysteresis of each drive pose in the real-time pose record is obtained.

[0062] The page-turning drive shaft can be a mechanical transmission shaft that directly transmits torque to drive the page-turning card to rotate. It can be used as the physical carrier for torque output, and its resistance characteristics affect the calculation of torque hysteresis. For example, the page-turning drive shaft can be one or more of the following: single-axis direct-drive type, multi-stage reduction coupling shaft, flexible coupling integrated shaft, etc. Torque hysteresis can be the additional torque compensation value required to overcome the mechanical resistance under the current pose. It can be used to characterize the dynamic compensation intensity required for each driving pose point and is an intermediate component of the driving compensation coefficient. In a specific embodiment, torque hysteresis can be obtained by jointly determining the feedback correction cost and pose deviation through a mapping function or lookup table method. The expression for torque hysteresis can be: In the formula, Let be the torque hysteresis of the k-th driving pose; The feedback correction cost is used to adjust the baseline strength of the hysteresis calculation. If the overall correction cost of the page-turning clock is high and the historical data performance is poor, it indicates that the page-turning clock has high rigidity or inertia, and the predicted hysteresis will also increase accordingly. The absolute value of the pose deviation is used for real-time error input. In the current page-turning action, the greater the deviation of the actual position from the target position, the greater the difference in the resistance or deformation of the mechanical structure, which will affect the hysteresis. The hysteresis coefficient is an inherent property determined by the mechanical structure of the page turner, such as gear friction and belt elasticity. The shape factor, i.e., the nonlinear adjustment parameter, determines the rate at which the hysteresis increases as the error increases. A large value means that even a small error can cause the hysteresis to spike rapidly. The torque hysteresis of each driving pose in the real-time pose record can be a set of torque hysteresis values ​​corresponding one-to-one with each sampling point in the real-time pose record. This can be used to form the profile of the compensation requirements throughout the entire process, supporting the refined generation of the driving compensation coefficients.

[0063] Determining the torque hysteresis of the page-turning drive axis by using feedback correction cost and pose deviation can be achieved by nonlinearly mapping or table lookup with the pose deviation, using the feedback correction cost as a weighting factor. Furthermore, this operation can utilize a pre-stored torque-deviation mapping table for interpolation and synthesize the hysteresis using a fuzzy inference engine, thereby fusing static feedback evaluation with dynamic execution deviation to generate a resistance compensation value for the current pose. Obtaining the torque hysteresis for each drive pose in the real-time pose record can be achieved by repeatedly performing the torque hysteresis calculation process on all pose points in the real-time pose record. Further, this operation can be implemented by calculating point-by-point and caching the results, and using a sliding window to update the hysteresis sequence online, thus constructing a dynamic compensation demand curve covering the entire page-turning stroke.

[0064] The drive compensation coefficient for the continuously variable drive process of the page-turning clock is determined based on all torque hysteresis and response deviation factors.

[0065] Here, all torque hysteresis can be the complete set of torque hysteresis corresponding to all driving poses in this page-turning action, which can be used as the mechanical input for calculating the drive compensation coefficient, reflecting the resistance change characteristics throughout the entire process. The drive compensation coefficient in the continuously variable drive process of the page-turning clock is determined based on all torque hysteresis and the response deviation factor. This can be achieved by fusing the torque hysteresis sequence with the response deviation factor to generate compensation coefficients suitable for the entire drive process. Furthermore, this operation can be performed by averaging the torque hysteresis and then weighting the product with the response deviation factor, using the response deviation factor as a gating signal to adjust the effective gain of the hysteresis. This allows for a globally optimal drive strategy under the dual constraints of voice layer credibility and mechanical layer resistance characteristics. For example, the expression for the drive compensation coefficient could be: In the formula, λ is the final driving compensation coefficient. The base compensation gain is the factory default compensation level of the page-turning clock, which is the theoretically optimal compensation parameter under ideal conditions of no noise and no deviation. As an interactive influencing factor, it serves as a bridge connecting voice interaction and mechanical actuation, determining... How much influence does (voice interpretation deviation) have on mechanical control? The larger the number, the more likely the page-turning clock is to be more cautious / forceful because it doesn't hear clearly. In response to the deviation factor, which is used to introduce the nondeterminism of the voice end, the compensation coefficient is dynamically adjusted when the voice command is ambiguous to prevent misoperation or incomplete action; Let be the torque hysteresis at the k-th driving pose. The total torque hysteresis reflects the overall physical load or resistance accumulation of the drive shaft. The larger the total hysteresis, the greater the resistance that the motor needs to overcome, and therefore a larger compensation coefficient λ is needed to offset these resistances.

[0066] For example, in the scenario of precise page-turning control under mechanical aging conditions, the page-turning clock flipping drive method supporting voice interaction control in this embodiment can be as follows: after long-term use, the gear set of the page-turning clock experiences slight wear, resulting in increased resistance in a specific angle range. When the user issues a voice command, the system begins to record the real-time pose. When turning the page to the high resistance section, the current actual pose lags significantly behind the historical reference pose, and the pose deviation increases significantly. Combining the previously calculated feedback correction cost, the system dynamically generates a higher torque hysteresis. At the same time, if the voice command has high confidence due to the quiet environment (low response deviation factor), the final drive compensation coefficient is appropriately increased, enhancing the output torque only in the resistance section, while maintaining low power consumption operation in other stages. This ensures that the page turns in place while avoiding noise and energy consumption caused by high torque throughout the process.

[0067] In one embodiment, the output torque of each page-turning drive module of the page-turning clock is dynamically adjusted based on the drive compensation coefficient, including: Torque compensation is performed on each driven pose in the real-time pose record by driving compensation coefficient to obtain all the compensated output torques. The output torque of the drive control unit is adjusted based on the drive error in each torque compensation and the corresponding output torque after compensation during the stepless drive of the page-turning clock.

[0068] The driving error can be the residual between the desired pose and the actual pose achieved, reflecting the effectiveness of the current torque compensation and serving as feedback for the output control of the drive control unit. In an exemplary embodiment, the driving error can be calculated by comparing the target driving pose with the current actual pose after execution. Further, the driving error can be one or more of the following, categorized by error source: friction disturbance error, transmission backlash error, motor response delay error, etc. All compensated output torques can be a set of torque commands obtained by applying the driving compensation coefficient to each driving pose in the real-time pose record. This set can be used to construct a dynamic torque planning curve for the entire page-turning action, replacing the traditional fixed torque output. For example, all compensated output torques can be input together with the driving error into the control logic of the drive control unit.

[0069] Torque compensation is applied to each drive pose in the real-time pose record using a drive compensation coefficient. This can be achieved by using the drive compensation coefficient as a scaling factor, applied to the base torque setting value corresponding to each drive pose. Further, this operation generates the compensated torque through point-by-point multiplication and obtains the nonlinear compensation curve through lookup table interpolation. This allows for the generation of a personalized torque output sequence that matches the current operating condition, achieving pose-level precision drive adaptation. Obtaining all compensated output torques can be achieved by summing the compensated torque values ​​of all drive poses to form a complete output sequence. In one specific embodiment, this operation can be implemented by caching the data in memory as an array and preloading it as a timestamp-associated instruction queue. This allows for the establishment of dynamic torque planning covering the entire process, providing a benchmark for closed-loop control.

[0070] The driving error in each torque compensation can be the execution deviation generated after applying compensation to the output torque at a certain pose point. It can be used to evaluate the adequacy of the compensation strategy at that point and guide the output adjustment of subsequent pose points. The corresponding compensated output torque can be the applied torque compensation value associated with a certain driving error, which can be used as an input item for the error-torque mapping relationship to support the online learning or correction of the drive control unit. The output torque of the drive control unit during the stepless drive of the page-turning clock can be the real-time torque command finally issued by the drive control unit to the page-turning drive module during the stepless drive stage. It can directly determine the force and smoothness of the page-turning action and is jointly controlled by the driving error and the compensated output torque. Based on the driving error in each torque compensation and the corresponding compensated output torque, the output torque of the drive control unit during the stepless drive of the page-turning clock can be controlled by jointly analyzing the driving error and the applied compensated output torque to dynamically correct the subsequent output torque. Furthermore, this operation can be achieved by using a proportional-integral controller to adjust the output at the next moment according to the error and constructing an error-torque mapping model for feedforward compensation, thereby realizing online closed-loop correction during the execution process and suppressing page-turning deviations caused by instantaneous disturbances.

[0071] For example, in an adaptive page-turning scenario where the card is slightly stuck, the page-turning clock flipping drive method supported by voice interaction control in this embodiment can be as follows: during the page-turning process, a card may become slightly stuck due to changes in humidity. When the card reaches this position, the system applies a driving force according to the compensated output torque. However, the actual position lags behind the target, resulting in a positive driving error. The drive control unit detects that this error does not match its corresponding compensated output torque, determines that the current resistance is higher than expected, and then increases the output torque at the next driving position. At the same time, because the voice command is clear (with a low response deviation factor), the system allows for a moderate increase in compensation intensity, and finally successfully turns over the stuck point without causing impact or repeated trial and error.

[0072] In one embodiment, torque compensation is performed on each driven pose in the real-time pose record using a drive compensation coefficient to obtain all compensated output torques, including: For each driven pose in the real-time pose record, the driving error is determined based on the driving compensation coefficient and the current pose deviation. The initial output torque is compensated using the drive error to obtain the compensated output torque, and then all compensated output torques are determined.

[0073] The current pose deviation can be the instantaneous deviation between the current driving pose and the target pose during real-time execution. It can be used as a key input for generating driving errors, reflecting the accuracy of the current page-turning action. In this embodiment, the current pose deviation can be obtained by calculating the difference between the actual pose collected by the sensor and the corresponding target pose in the planned trajectory. Furthermore, the current pose deviation can include, but is not limited to, one or more of the following: lead deviation, lag deviation, and oscillation deviation. The initial output torque can be the original torque value set based on the basic driving strategy before compensation is applied. It can be used as a reference value for torque compensation and the final output is generated after driving error correction. For example, the initial output torque can be preset by the continuously variable drive control logic according to the page-turning stage or inherited from the output of the previous cycle. In a specific embodiment, the initial output torque can be a stage-constant initial torque, a ramp-changing initial torque, or a zero initial torque.

[0074] For each driven pose in the real-time pose record, the driving error is determined based on the driving compensation coefficient and the current pose deviation. This can be achieved by mapping or calculating the driving compensation coefficient and the current pose deviation to generate a driving error characterizing the influence of composite disturbances. Furthermore, this operation can be implemented by scaling the current pose deviation using the driving compensation coefficient as a weight, or by synthesizing a multi-dimensional error level using fuzzy rules, thereby constructing an error index that integrates semantic credibility and physical state, surpassing the limitations of traditional pure position errors. Compensating the initial output torque using the driving error can be achieved by superimposing or modulating the driving error as a correction amount onto the initial output torque. In an exemplary embodiment, this operation can be implemented using additive compensation (initial torque plus error multiplied by gain) or multiplicative modulation (initial torque multiplied by an error function), ensuring that the output torque matches the actual requirements and achieving dynamic force control based on composite errors. For example, the expression for the compensated output torque can be: The initial output torque is the theoretical output used for open-loop control. It is the standard torque calculated by the page-turning clock basic control algorithm based on the current page-turning number, without considering environmental interference and mechanical wear. The current pose error, which is the direct basis for closed-loop feedback, is determined by subtracting the current position from the target position, indicating how much further distance or angle needs to be traveled. For historical reference position, This represents the current actual pose; The knob used to adjust the correction intensity is used to drive the compensation coefficient. It determines the system's response speed and strength to errors and is an intelligent decision result that integrates voice interference and mechanical hysteresis. The compensated output torque is ultimately applied to the motor control quantity. This is the actual torque executed by the drive module, ensuring that the page-turning clock can still accurately and smoothly complete the page-turning action even under far-field voice interference. Obtaining the compensated output torque can involve correcting the torque at a single drive pose point and outputting the final drive command at that moment. Furthermore, this operation can be directly output to the motor driver or sent after smoothing and filtering to avoid abrupt changes, thereby generating a precise drive force adapted to the current voice quality and mechanical state. Determining all compensated output torques can be achieved by traversing all drive poses in the real-time pose record and repeating the aforementioned compensation process to form a complete torque sequence. For example, this operation can be implemented by offline batch generation and execution, or online point-by-point calculation and immediate application, thereby establishing an adaptive torque planning covering the entire page-turning stroke and supporting smooth execution throughout.

[0075] For example, in a scenario where cautious page turning is required under low-confidence voice commands, the page-turning clock flipping drive method supported by voice interaction control in this embodiment can be as follows: the user issues a page-turning command in a noisy kitchen, the low confidence of the voice analysis leads to a high response deviation factor, and the drive compensation coefficient is suppressed; after page turning is initiated, the system detects a slight lag in posture (the current posture deviation is negative), but because the drive compensation coefficient is small, the generated drive error amplitude is limited; the initial output torque is only slightly increased to avoid strong drive due to possible misrecognition; finally, the system completes the page turning tentatively at a low speed, and if the subsequent confirmation that the command is correct continues, otherwise it can be quickly stopped, effectively preventing erroneous actions and mechanical impact.

[0076] In addition, refer to Figure 2 To achieve the above objectives, the present invention also provides a page-turning clock flipping drive system supporting voice interaction control, the system comprising: The data acquisition module 10 is used to collect the interactive operation records of the drive control unit during the operation of the page-turning clock, and thus obtain historical operation data; The voice evaluation module 20 is used to extract the interaction response data of each voice parsing unit of the page-turning clock voice interaction module when processing far-field voice commands in the historical operation data, and then determine the response deviation factor in the command parsing process through all the interaction response data. Fuzzy mapping module 30 is used to obtain historical pose deviation data in the page turning continuously driven process from the historical running data, use the error membership function in the function library to perform fuzzy mapping on the historical pose deviation data, obtain the deviation membership degree corresponding to each input quantity in the continuously driven control logic, and then determine the feedback correction cost of the flexible closed-loop feedback mechanism based on all the deviation membership degrees. The coefficient calculation module 40 is used to determine the drive compensation coefficient when the page-turning clock performs stepless page-turning drive through the feedback correction cost and the response deviation factor; The torque regulation module 50 is used to dynamically regulate the output torque of each page-turning drive module of the page-turning clock based on the drive compensation coefficient.

[0077] Other embodiments or specific implementations of the page-turning clock flipping drive system supporting voice interaction control described in this invention can be referred to the above-described method embodiments, and will not be repeated here.

[0078] Furthermore, to achieve the above objectives, the present invention also provides a page-turning clock flipping drive device supporting voice interaction control. The device includes: a memory, a processor, and a page-turning clock flipping driver program supporting voice interaction control stored in the memory and executable on the processor. The page-turning clock flipping driver program supporting voice interaction control is configured to implement the steps of the page-turning clock flipping drive method supporting voice interaction control as described above.

[0079] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing a page-turning clock driver program that supports voice interaction control. When the page-turning clock driver program that supports voice interaction control is executed by a processor, it implements the steps of the page-turning clock driver program that supports voice interaction control as described above.

[0080] The above are merely preferred embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for driving the flip clock to rotate, supporting voice interaction control, characterized in that, The method includes: Collect the interactive operation records of the drive control unit during the operation of the page-turning clock, and then obtain historical operation data; Extract the interactive response data of each voice parsing unit of the page-turning clock voice interaction module when processing far-field voice commands from the historical operation data, and then determine the response deviation factor in the command parsing process through all the interactive response data; Historical pose deviation data during the page-turning continuously variable drive process is obtained from the historical operation data. The error membership function in the function library is used to perform fuzzy mapping on the historical pose deviation data to obtain the deviation membership degree corresponding to each input quantity in the continuously variable drive control logic. Then, the feedback correction cost of the flexible closed-loop feedback mechanism is determined based on all the deviation membership degrees. The drive compensation coefficient when the page-turning clock performs stepless page-turning drive is determined by the feedback correction cost and the response deviation factor. The output torque of each page-turning drive module of the page-turning clock is dynamically adjusted based on the drive compensation coefficient.

2. The page-turning clock flipping drive method supporting voice interaction control as described in claim 1, characterized in that, The process of determining the response deviation factor during instruction parsing by using all interactive response data includes: Extract the environmental noise features of far-field voice interaction during each command parsing in all interactive response data, and then obtain the noise feature vector; The ambiguity of command recognition after processing by the corresponding speech parsing unit is determined based on each interactive response data. The response deviation factor for parsing the page-turning clock voice commands is determined by the noise feature vector and all command recognition ambiguities.

3. The page-turning clock flipping drive method supporting voice interaction control as described in claim 1, characterized in that, The step of using error membership functions from the function library to perform fuzzy mapping on the historical pose deviation data to obtain the deviation membership degree corresponding to each input quantity in the continuously variable drive control logic includes: All drive output deviations in the historical pose deviation data are aggregated and divided into multiple fuzzy deviation intervals. Each driving deviation in the historical pose deviation data is mapped to the corresponding deviation fuzzy interval to obtain the deviation mapping domain; By mapping and judging each driving deviation using the deviation mapping domain and the error membership function in the function library, the deviation membership degree corresponding to each driving deviation in the historical pose deviation data is obtained.

4. The page-turning clock flipping drive method supporting voice interaction control as described in claim 1, characterized in that, The feedback correction cost of determining the flexible closed-loop feedback mechanism based on all deviation membership degrees includes: The deviation bound of the output pose is determined by all the deviation memberships; Typical deviation values ​​are selected from all deviation memberships, and then the typical deviation values ​​are used to determine the typical pose values ​​of the output pose after the page turning action is completed. The feedback correction cost of the flexible closed-loop feedback mechanism is obtained by using the typical pose value and the deviation boundary.

5. The page-turning clock flipping drive method supporting voice interaction control as described in claim 1, characterized in that, The determination of the drive compensation coefficient when the page-turning clock performs stepless page-turning drive through the feedback correction cost and the response deviation factor includes: Acquire real-time pose records during the far-field voice-triggered page-turning action; For each driven pose in the real-time pose record, the historical reference pose corresponding to the driven pose is obtained from the historical pose deviation data; the current actual pose corresponding to the driven pose is obtained from the real-time pose record, and then the pose deviation between the historical reference pose and the current actual pose is determined. The torque hysteresis of the page-turning drive axis is determined by the feedback correction cost and the pose deviation, and then the torque hysteresis of each drive pose in the real-time pose record is obtained. The drive compensation coefficient during the continuously variable drive process of the page-turning clock is determined based on all torque hysteresis and the response deviation factor.

6. The page-turning clock flipping drive method supporting voice interaction control as described in claim 5, characterized in that, The dynamic adjustment of the output torque of each page-turning drive module of the page-turning clock based on the drive compensation coefficient includes: The torque is compensated for each driving pose in the real-time pose record by the driving compensation coefficient to obtain all the compensated output torques. The output torque of the drive control unit is adjusted based on the drive error in each torque compensation and the corresponding output torque after compensation during the stepless drive of the page-turning clock.

7. The page-turning clock flipping drive method supporting voice interaction control as described in claim 6, characterized in that, The process involves applying torque compensation to each driving pose in the real-time pose record using the driving compensation coefficient to obtain all compensated output torques, including: For each driven pose in the real-time pose record, the driving error is determined based on the driving compensation coefficient and the current pose deviation. The initial output torque is compensated using the drive error to obtain the compensated output torque, and then all compensated output torques are determined.

8. A page-turning clock flipping drive system supporting voice interaction control, characterized in that, The system includes: The data acquisition module is used to collect the interactive operation records of the drive control unit during the operation of the page-turning clock, thereby obtaining historical operation data; The voice evaluation module is used to extract the interaction response data of each voice parsing unit of the page-turning clock voice interaction module when processing far-field voice commands in the historical operation data, and then determine the response deviation factor in the command parsing process through all the interaction response data. The fuzzy mapping module is used to obtain historical pose deviation data in the page-turning continuously variable drive process from the historical running data, and to perform fuzzy mapping on the historical pose deviation data using the error membership function in the function library to obtain the deviation membership degree corresponding to each input quantity in the continuously variable drive control logic. Then, based on all the deviation membership degrees, the feedback correction cost of the flexible closed-loop feedback mechanism is determined. The coefficient calculation module is used to determine the drive compensation coefficient when the page-turning clock performs stepless page-turning drive by using the feedback correction cost and the response deviation factor; The torque control module is used to dynamically control the output torque of each page-turning drive module of the page-turning clock based on the drive compensation coefficient.

9. A page-turning clock flipping drive device supporting voice interaction control, characterized in that, The device includes: a memory, a processor, and a voice-interactive control-enabled page-turning clock driver program stored in the memory and executable on the processor, wherein the voice-interactive control-enabled page-turning clock driver program is configured to implement the steps of the voice-interactive control-enabled page-turning clock driver method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a page-turning clock driver program that supports voice interaction control. When the page-turning clock driver program that supports voice interaction control is executed by a processor, it implements the steps of the page-turning clock driver program that supports voice interaction control as described in any one of claims 1 to 7.

Images