A robot touch interaction control method

Abstract

The present invention relates to the field of interactive technology for teaching equipment, and in particular to a method for controlling robot touch interaction. The method obtains touch points and touch position areas on a preschool education device; obtains time periods consisting of continuous sampling moments with and without touch points as touch time periods and blank time periods respectively; analyzes changes in touch points within two touch time periods adjacent to the blank time periods, and updates the touch time periods; constructs multiple connectivity graphs based on the updated touch time periods; combines the structural distribution characteristics of the connectivity graph, performs weighted adjustment on node similarity, and determines clusters based on a connectivity graph splitting clustering algorithm; extracts sampling moments within the clusters, performs interpolation and supplementation on the touch position areas, and obtains interaction trajectories. The present invention improves the accuracy of human-computer interaction of preschool education equipment by extracting the changing characteristics of the touch action position and analyzing the correlation between the structural distribution of nodes.

Description

Technical Field

[0001] The present invention relates to the technical field of teaching equipment interaction, and in particular to a robot touch interaction control method. Background Art

[0002] Robotic touch interaction refers to an interactive method of using fingers or palms to input and operate on a touch screen. For example, human-computer interaction in early childhood education equipment based on infrared touch is a type of robotic touch interaction. Early childhood education equipment based on infrared touch can provide a rich variety of learning content, such as text, graphics, sound, video, etc. By touching the screen, young children can learn basic knowledge such as numbers, letters, shapes, and colors. They can also participate in various learning and interactive activities such as enlightenment education, fun games, and painting. The user interactive experience of its human-computer interaction is closely related to the accuracy of motion recognition of the infrared touchpad. When there are water stains or oil stains on the touchpad surface or the user's fingers, it is easy to cause related problems such as disconnection, affecting the real-time and accuracy of human-computer interaction. Therefore, it is necessary to improve the accuracy of motion recognition based on the changing trajectory of the user's touch motion on the touchpad to enhance the human-computer interaction experience.

[0003] Currently, a common method for identifying human-computer interaction in early childhood education equipment is to cluster the touch action position data using a connected graph splitting clustering algorithm. This method can effectively analyze the correlation characteristics of the temporal trajectory changes of the user's touch actions on the touchpad, so as to achieve accurate identification of touch actions and thereby enhance the human-computer interaction experience of early childhood education equipment. However, in the process of processing the touch action position data using the connected graph splitting clustering algorithm, there are multiple touch operations that are similar to the performance when the touch is disconnected. When multiple touch operations and the touch is disconnected, it will be considered that there is no touch operation, resulting in inaccurate analysis results. Summary of the Invention

[0004] In order to solve the technical problem that when processing touch action position data using a connected graph splitting clustering algorithm, multiple touch operations and disconnected touches occur, resulting in inaccurate analysis results of human-computer interaction, the present invention aims to provide a robot touch interaction control method. The technical solutions adopted are as follows:

[0005] Acquire touch points generated when performing touch interaction on the early childhood education device of the infrared touch panel, and form a touch position area from the touch points;

[0006] Respectively obtaining a time period consisting of continuous sampling moments in which a touch point exists on the early childhood education device and a time period consisting of continuous sampling moments in which no touch point exists as a touch time period and a blank time period; performing an action change correlation analysis on touch point changes in two touch time periods adjacent to the blank time period to determine an action change correlation of the blank time period; screening the blank time periods according to the action change correlation of the blank time period to obtain a correlated time period; and updating the touch time period based on the correlated time period;

[0007] Based on the updated touch time period, multiple connectivity graphs are constructed; node distinction confidences of nodes in the connectivity graphs are determined based on the structural distribution characteristics of the connectivity graphs; node similarities are weighted using the node distinction confidences as weights, and clusters are determined based on a connectivity graph divisive clustering algorithm;

[0008] The sampling moments belonging to the associated time period in the cluster are extracted as interaction moments; and the touch position areas are interpolated and supplemented for the interaction moments to obtain a final interaction trajectory formed by the touch position areas.

[0009] Preferably, performing motion change correlation analysis on touch point changes in two touch time periods adjacent to the blank time period to determine the motion change correlation of the blank time period includes:

[0010] Obtaining the touch position change degree at each sampling moment in two touch time periods adjacent to the blank time period;

[0011] Determining the correlation between the interval changes of the touch positions at different sampling moments according to the change degrees of the touch positions at different sampling moments within the touch time period and the length of the touch time period;

[0012] The motion change correlation of the blank time period is determined by combining the interval change correlation of the touch positions corresponding to the sampling moments belonging to the touch time period at the left and right end points of the blank time period.

[0013] Preferably, obtaining the touch position change degree at each sampling moment in two touch time periods adjacent to the blank time period includes:

[0014] Any sampling moment within the touch time period is taken as the target moment, the difference in the number of touch points between the target moment and the previous sampling moment is taken as the numerator, the time difference between the target moment and the previous sampling moment is taken as the denominator, and the fraction formed by the numerator and denominator is taken as the touch position change degree at the target moment.

[0015] Preferably, the calculation formula for the interval change correlation is:

[0016]

[0017] in, is the correlation degree of the touch position interval change at the i-th sampling moment within the touch time period; is the normalization function; is the length of the touch time period at the i-th sampling moment; is the touch position change degree at the i-th sampling moment within the touch time period; is an exponential function with the natural constant e as the base; is the number of sampling moments in the touch time period of the i-th sampling moment; is the average value of the touch position change at all sampling moments within the touch time period.

[0018] Preferably, the step of determining the motion change correlation of the blank time period by combining the interval change correlation of the touch positions corresponding to the sampling moments belonging to the touch time period at the left and right endpoints of the blank time period includes:

[0019] The difference in interval change correlation between the touch positions corresponding to the sampling moments belonging to the touch time period at the left and right endpoints of the blank time period is calculated as the interval change degree; the negative correlation normalized value of the interval change degree is used as the action change correlation degree of the blank time interval.

[0020] Preferably, constructing multiple connectivity graphs based on the updated touch time period includes:

[0021] The centroid point of the touch position area corresponding to a single sampling moment within the updated touch time period is used as a node of the connectivity graph, the number of nodes in the action time period within the touch time period is used as the preset incremental clustering increment value, and the preset incremental clustering increment value is used as the number of nodes in the touch time period. Multiple connectivity graphs are constructed according to the connectivity graph splitting clustering algorithm.

[0022] Preferably, the calculation formula for the node differentiation confidence is:

[0023] ;

[0024] ;

[0025] in, The node discrimination of the connectivity graph constructed for the i-th node in the r-th incremental clustering; Node discrimination confidence for the i-th node; is the correlation degree of the touch position interval change at the i-th sampling moment within the touch time period; The number of node connection edges of the first part of the connected graph constructed by incremental clustering for the i-th node; The number of nodes in the first part of the connected graph constructed by incremental clustering that is divided for the i-th node; The number of nodes in the second part of the connected graph constructed by incremental clustering divided for the i-th node; The number of nodes connecting the second part of the connected graph constructed by incremental clustering for the i-th node; R is the number of incremental clustering times; norm is the normalization function; exp is the exponential function with the natural constant e as the base; is the absolute value symbol.

[0026] Preferably, the node similarity is weighted using the node distinction confidence as a weight, including:

[0027] The calculation formula for weighted node similarity is:

[0028]

[0029] in, The weighted node similarity between the a-th node and the b-th node; Calculate the node distinction confidence between the a-th node and the k-th node in the b-th node; is the initial node similarity between the a-th node and the b-th node.

[0030] Preferably, updating the touch time period based on the associated time period includes:

[0031] The associated time period and the adjacent touch time period are merged and updated to obtain an updated touch time period.

[0032] Preferably, the step of screening the blank time periods according to the action change correlation of the blank time periods to obtain the correlated time periods includes:

[0033] The blank time period in which the corresponding interval change correlation is greater than a preset correlation threshold is used as the correlation time period.

[0034] The embodiments of the present invention have at least the following beneficial effects:

[0035] Acquire the touch points and touch position areas generated when performing touch interaction on the early childhood education device; respectively acquire the time periods consisting of continuous sampling moments with and without touch points as touch time periods and blank time periods. The sampling moments are divided into touch time periods and blank time periods in order to subsequently analyze from the blank time periods the abnormal situation that there are no touch points on the early childhood education device due to multiple touch operations or disconnection; analyze the touch point changes in the two touch time periods adjacent to the blank time period, update the touch time period, and realize the area change characteristics of the touch point set in the time sequence, and filter out the abnormal sampling moments of the non-existence of touch points on the early childhood education device due to multiple touch operations or disconnection from the blank time period as the associated time period. Based on the updated touch time period, multiple connectivity graphs are constructed; then, combined with the structural distribution characteristics of the connectivity graph, the node similarity is weighted and adjusted, and clusters are determined based on the connectivity graph split clustering algorithm, so that the final connectivity graph split clustering algorithm's accurate processing results can achieve accurate judgment of touch accuracy affected by environmental factors such as touch disconnection; the sampling moments within the clusters are extracted, and the touch position area is interpolated and supplemented to obtain the interaction trajectory. The interaction trajectory constructed based on the touch position area after linear interpolation is more accurate. Real-time human-computer interaction feedback is performed based on the interaction trajectory of the final touch position area obtained, effectively avoiding the impact of environmental factors such as touch disconnection on the human-computer interaction experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1 A flowchart of a robot touch interaction control method provided by one embodiment of the present invention;

[0038] Figure 2 This is a flow chart of a method for determining the correlation degree of action changes in a blank time period provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0039] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effectiveness of a robot touch interaction control method proposed in accordance with the present invention. In the following description, different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0040] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0041] The embodiment of the present invention provides a specific implementation method of a robot touch interaction control method, which is applicable to human-computer interaction scenarios of infrared touch preschool education equipment. In order to solve the technical problem that in the process of processing touch action position data using the connected graph split clustering algorithm, there are multiple touch operations and touch disconnection, which will lead to inaccurate analysis results of human-computer interaction. The present invention extracts the characteristics of touch action position changes, conducts a precise analysis of the structural distribution correlation between nodes, and improves the accuracy of action recognition, so that the final precise processing result of the connected graph split clustering algorithm can achieve accurate judgment of touch accuracy affected by environmental factors such as touch disconnection; extracts the sampling time within the cluster, performs interpolation and supplementation of the touch position area, and obtains the interaction trajectory. The interaction trajectory constructed according to the touch position area after linear interpolation is more accurate, so as to enhance the human-computer interaction experience.

[0042] The specific scheme of the robot touch interaction control method provided by the present invention is described in detail below with reference to the accompanying drawings.

[0043] See also Figure 1 , which shows a flowchart of a robot touch interaction control method provided by one embodiment of the present invention, the method comprising the following steps:

[0044] Step S100: acquiring touch points generated when touch interaction is performed on the early childhood education device of the infrared touch panel, and forming a touch position area by the touch points.

[0045] In an embodiment of the present invention, an infrared sensor array within a preschool education device with an infrared touchpad collects touch points during user touch interactions and obtains the position coordinates of the touch points. The sampling frequency is 150 Hz, and each sampling captures a touch location region, resulting in the acquisition of the position coordinates of multiple touch points. This yields multiple sets of touch location coordinates over time. Specifically, the touch points form a touch location region. Specifically, the position coordinates of the touch points are extracted to obtain the corresponding touch location region.

[0046] When there are water stains or oil stains on the touchpad surface or the user's fingers, the infrared touchpad sensor is blocked, resulting in loss or instability of the touch signal, and related problems such as touch disconnection, which affects the user's human-computer interaction experience. The touch disconnection signal will show characteristics similar to multiple touch interactions, but the area change and position change characteristics of the touch area are different between the touch disconnection and multiple input processes.

[0047] Step S200, respectively obtain a time period consisting of continuous sampling moments when touch points exist on the early childhood education device and a time period consisting of continuous sampling moments when no touch points exist, as a touch time period and a blank time period; perform an action change correlation analysis on touch point changes in two touch time periods adjacent to the blank time period to determine the action change correlation of the blank time period; screen the blank time periods according to the action change correlation of the blank time period to obtain a correlated time period; and update the touch time period based on the correlated time period.

[0048] Therefore, in order to solve the problem of abnormal human-computer interaction when there is a disconnection or multiple inputs, the embodiment of the present invention first divides the continuous sampling time period into a touch time period and a blank time period, wherein the touch time period is the time period when there is a touch point on the early childhood education device, and the blank time period is the time period when there is no touch point on the early childhood education device. However, in addition to the moments when there is no real touch, the blank time period also includes moments when there is real touch due to disconnection or multiple touch interactions, but the touch point is not detected by the early childhood education device. Therefore, after dividing the touch time period and the blank time period, further, combined with the action change correlation analysis of the touch time period adjacent to the blank time period, the time period associated with the touch time period is filtered out from the blank time period as the associated time period.

[0049] Since the acquired touch position areas have temporal area and position change characteristics, the embodiments of the present invention analyze the temporal area change characteristics of the touch position areas to obtain the change correlation, analyze the consistency of the structural changes in the constructed connectivity graph, adjust the similarity measure between the connectivity graph nodes, and obtain accurate clustering results.

[0050] First, based on the collected touch position area data, the temporal variation characteristics are analyzed to obtain the correlation of motion changes. Specifically, the time periods consisting of consecutive sampling moments with touch points and the time periods consisting of consecutive sampling moments without touch points on the early childhood education device are respectively obtained as touch time periods and blank time periods. The motion change correlation of touch point changes in two touch time periods adjacent to the blank time period is analyzed to determine the motion change correlation of the blank time period.

[0051] Because the collected touch position area data corresponds to multiple touch position area data when sampled at the sampling frequency, and the area of ​​the touch position area will change due to changes in the touch action trajectory, that is, the number of touch points in the area will change. When the touch action trajectory has a large bend, the area changes on the trajectory before and after the bend will have consistent characteristics, that is, they will increase or decrease together. When the touch is disconnected, there will be an abrupt area reduction or disappearance, which does not have consistent change. Therefore, the correlation degree of action changes can be obtained based on the temporal change characteristics of the touch point collection area.

[0052] Based on the acquired touch position area data, the number of touch points in the touch position area corresponding to the sampling moment is recorded; when the number of touch points corresponding to the touch action does not exist at the sampling moment, it is recorded as a blank moment, and multiple time segments with touch points are obtained and recorded as touch moments. The time period consisting of the continuous sampling moments with touch points and the time period consisting of the continuous sampling moments without touch points on the early childhood education device are obtained respectively as the touch time period and the blank time period. It should be noted that the moment within the touch time period is the touch moment, and the moment within the blank time period is the blank moment.

[0053] See also Figure 2 , performing motion change correlation analysis on the touch point changes in two touch time periods adjacent to the blank time period to determine the motion change correlation of the blank time period, specifically:

[0054] Step S201 : obtaining the touch position change degree at each sampling moment in two touch time periods adjacent to the blank time period.

[0055] Any sampling moment within the touch time period is taken as the target moment, the difference in the number of touch points between the target moment and the previous sampling moment is taken as the numerator, the time difference between the target moment and the previous sampling moment is taken as the denominator, and the fraction formed by the numerator and denominator is taken as the touch position change degree at the target moment.

[0056] Taking the i-th sampling moment as the target moment, the calculation formula for the touch position change degree at the corresponding target moment is:

[0057] ;in, is the touch position change at the i-th sampling moment; is the number of touch points at the i-th sampling moment; is the number of touch points at the i-1th sampling moment; is the i-th sampling moment; is the i-1th sampling moment; is the absolute value symbol.

[0058] In the embodiment of the present invention, the degree of change in the number of touch points at adjacent moments is used to reflect the degree of change in the touch position at different sampling moments.

[0059] Step S202 : determining the correlation between the touch position changes at different sampling moments according to the touch position changes at different sampling moments within the touch time period and the length of the touch time period.

[0060] The calculation formula of the interval change correlation is:

[0061]

[0062] in, is the correlation degree of the touch position interval change at the i-th sampling moment within the touch time period; is the normalization function; is the length of the touch time period at the i-th sampling moment; is the touch position change degree at the i-th sampling moment within the touch time period; is an exponential function with the natural constant e as the base; is the number of sampling moments in the touch time period of the i-th sampling moment; is the average value of the touch position change at all sampling moments within the touch time period.

[0063] The interval change correlation is calculated by combining the length of the touch time period at the sampling moment and the touch position change degree. It reflects the magnitude of the touch position change degree in the touch time period of the i-th sampling moment relative to the overall touch position change degree. The interval change correlation degree of the sampling moments is constructed by taking the average of the touch position change degrees of all sampling moments within the touch time period of the sampling moment. This improves the fault tolerance when calculating the interval change correlation degree and avoids the situation where when an abnormal situation occurs at a certain sampling moment, the interval change correlation degree corresponding to the sampling moment is constructed by using a single touch position change degree, which may cause large errors.

[0064] No. The correlation degree of interval changes at each sampling moment The larger the value is, the higher the accuracy of the user's touch position at the sampling moment is, and the more stable the changes in the multiple touch position areas in the corresponding touch time period are.

[0065] Step S203 : Determine the motion change correlation of the blank time period by combining the interval change correlation of the touch positions corresponding to the sampling moments belonging to the touch time period at the left and right end points of the blank time period.

[0066] The difference in the interval change correlation between the touch positions corresponding to the sampling moments of the touch time period at the left and right endpoints of the blank time period is calculated as the interval change correlation; the negative correlation normalized value of the interval change correlation is used as the motion change correlation of the blank time interval. In this embodiment of the present invention, the negative correlation normalization operation of the interval change correlation is implemented using an exponential function with a natural constant as the base and the negative value of the interval change correlation as the exponent.

[0067] The calculation formula for the action change correlation is:

[0068] ;in, is the action change correlation degree in the jth blank time period; is an exponential function with a natural constant as its base; is the interval change correlation degree of the touch position corresponding to the sampling moment belonging to the touch time period at the left end point of the j-th blank time period; is the interval change correlation degree of the touch position corresponding to the sampling moment belonging to the touch time period at the right end point of the j-th blank time period; is the absolute value symbol; is the interval variation.

[0069] Based on the correlation of the action changes in the blank time periods, the blank time periods are screened to obtain associated time periods. Specifically, the blank time periods whose corresponding interval change correlation is greater than a preset correlation threshold are selected as associated time periods. In this embodiment of the present invention, the preset correlation threshold is 0.68. In other embodiments, the value can be adjusted by the implementer based on actual conditions.

[0070] Obtain the correlation of action changes in all blank time periods and use the preset correlation threshold Perform blank time period screening. When the correlation of the action changes in the blank time period is greater than the preset correlation threshold When , it can be considered that the blank time period in which it is located has a high correlation with the corresponding touch time periods on both sides, and the blank time period here should not have a distinguishing degree, thereby obtaining all the remaining blank time periods to achieve the screening of blank time periods.

[0071] After the associated time period is obtained, the touch time period is updated based on the associated time period. Specifically, the associated time period and the adjacent touch time period are merged and updated to obtain an updated touch time period.

[0072] By collecting touch position area data and analyzing the timing change characteristics, we can obtain the correlation of action changes and screen out blank time periods in which the infrared touch panel sensor is blocked due to water or oil stains on the touch panel surface or the user's fingers, resulting in loss or instability of the touch signal and disconnection. These blank time periods are then updated and integrated with the touch time periods.

[0073] Step S300: construct multiple connectivity graphs based on the updated touch time period; determine the node distinction confidence of the nodes in the connectivity graph based on the structural distribution characteristics of the connectivity graph; use the node distinction confidence as a weight to weight the node similarity, and determine clusters based on the connectivity graph divisive clustering algorithm.

[0074] After obtaining the action change correlation, an accurate clustering result is obtained based on the connectivity graph splitting clustering algorithm and the connectivity graph structure distribution characteristics. Among them, the action change correlation obtained in step S200 is a touch action correlation analysis based on the relationship between the change characteristics of the number of touch points in the touch position area caused by the touch action. However, in the actual use of early childhood education equipment, due to the diversity of touch actions of the user group of young children, the obtained action change correlation is often difficult to cope with a variety of complex changes. The main reason is that the coherence of the touch action of young children is poor, and the fluctuation of the number of touch points in the touch position area on the change trajectory structure of the touch action is more complicated. Therefore, the centroid point of the touch position area within the updated touch time period can be used as a node of the connectivity graph to construct a connectivity graph, and further combined with the connectivity graph structure distribution characteristics to obtain accurate clustering analysis results.

[0075] The centroid of the touch position area corresponding to a single sampling moment in the updated touch time period is used as a node in the connectivity graph, the number of nodes in the action time period in the touch time period is used as the preset incremental clustering increment value, and the preset incremental clustering increment value is used as the number of nodes in the touch time period. According to the CABDDCG algorithm combined with the density and hierarchical partitioning method, multiple connectivity graphs are constructed. The CABDDCG algorithm combined with the density and hierarchical partitioning method is also a connectivity graph splitting clustering algorithm. Specifically, according to the obtained change correlation between the touch time period and the blank time period, the time is re-segmented to obtain multiple updated touch time periods, and the centroid of the touch position area corresponding to the single sampling moment in the updated touch time period is used as a node in the connectivity graph, thereby obtaining multiple connectivity graph nodes; incremental clustering is performed on multiple connectivity graph nodes. The specific method is: construct a two-dimensional sample space that is the same size or proportionally scaled as the infrared touch panel in the early childhood education equipment, and the coordinates of its nodes correspond to the coordinates of the centroid of the touch position area, and obtain the number of nodes in the touch time period. , the number of nodes in the action time period within the touch time period is used as the preset incremental clustering increment value, and multiple connected graphs are constructed with the preset incremental clustering increment value as the number of nodes in the touch time period. For example, if there are 3 updated touch time periods, incremental clustering is performed three times to obtain 3 connected graphs, corresponding to time periods 1, 12, and 123 respectively. It should be noted that according to the document Deng Jianshuang, Zheng Qilun, Peng Hong, and Deng Weiwei, "Clustering Algorithm Based on Dynamic Splitting of Connected Graphs", Journal of South China University of Technology (Natural Science Edition), January 2007, Vol. 35, No. 1, 1000-565X(2007)01-0118-05., a method for establishing a connected graph based on the CABDDCG algorithm combined with density and hierarchical partitioning methods is implemented in the embodiment of the present invention to construct multiple connected graphs with the preset incremental clustering increment value as the number of nodes in the touch time period. That is, the embodiment of the present invention is an improvement to the document "Clustering Algorithm Based on Dynamic Splitting of Connected Graphs".

[0076] For each connected graph with structural changes, its nodes can divide the connected graph into two parts. The higher the structural similarity of the node sets of the two connected graphs, the higher the credibility of the splitting threshold corresponding to this node.

[0077] The node discrimination confidence of the nodes in the connectivity graph is determined by combining the structural distribution characteristics of the connectivity graph.

[0078] The calculation formula for the node differentiation confidence is:

[0079] ;

[0080] ;

[0081] in, The node discrimination of the connectivity graph constructed for the i-th node in the r-th incremental clustering; Node discrimination confidence for the i-th node; is the correlation degree of the touch position interval change at the i-th sampling moment within the touch time period; The number of node connection edges of the first part of the connected graph constructed by incremental clustering for the i-th node; The number of nodes in the first part of the connected graph constructed by incremental clustering that is divided for the i-th node; The number of nodes in the second part of the connected graph constructed by incremental clustering divided for the i-th node; The number of nodes connecting the second part of the connected graph constructed by incremental clustering for the i-th node; R is the number of incremental clustering times; norm is the normalization function; exp is the exponential function with the natural constant e as the base; is the absolute value symbol.

[0082] By determining the structural similarity between two parts of the connected graph divided by different nodes, the node discrimination confidence of each node is obtained. The node discrimination obtained by different incremental clustering times is summed up and the sum of the node discrimination values ​​obtained multiple times is used as the node discrimination confidence of the node. This avoids the situation where the node discrimination obtained by a single incremental clustering time is used as the node discrimination confidence of the node, which is prone to large deviations.

[0083] Obtain the node discrimination confidence for each node. Each node has one or more node connections on the largest connected graph. Specifically, the node discrimination confidence can be used to adjust the node similarity between nodes. Specifically, the node discrimination confidence of each node is used as a weight to weight the node similarities. Clusters are determined based on the connected graph divisive clustering algorithm.

[0084] The calculation formula for weighted node similarity is:

[0085]

[0086] in, The weighted node similarity between the a-th node and the b-th node; Calculate the node distinction confidence between the a-th node and the k-th node in the b-th node; is the initial node similarity between the a-th node and the b-th node.

[0087] In the embodiment of the present invention, the initial node similarity between two nodes is the Euclidean distance between the two nodes. It should be noted that the node similarity between the two nodes is calculated here, so the maximum value of k is 2.

[0088] After obtaining the weighted node similarity, the subsequent connected graph splitting clustering algorithm is further completed to obtain accurate clustering results, that is, to determine the clusters. Each cluster contains multiple nodes, and the change trajectory of the touch location area corresponding to the node is the touch location area adjusted according to the requirements.

[0089] Step S400 : extracting sampling moments in the clusters that belong to the associated time period as interaction moments; interpolating and supplementing the touch position areas of the interaction moments to obtain final interaction trajectories formed by the touch position areas.

[0090] According to the obtained clustering results, the blank area in the touch position area corresponding to the cluster is extracted, and the blank area is a defect caused by broken touch, and linear interpolation processing is performed to obtain the final touch position area trajectory; wherein, the time period corresponding to the blank area in the touch position area corresponding to the cluster is also the screened associated time period, so according to the obtained clustering results, the blank area in the touch position area corresponding to the cluster is extracted, that is, the sampling time belonging to the associated time period in the cluster is extracted, the extracted sampling time is used as the interaction time, and linear interpolation processing is performed on the interaction time to realize interpolation and supplement of the touch position area at the interaction time, and then the final touch position area after interpolation and supplement is obtained, and the interaction trajectory of the early childhood education device is formed by the final touch position areas at different times.

[0091] Real-time human-computer interaction feedback, such as vibration feedback and sound feedback of the touchpad, can be performed based on the interaction trajectory of the final touch position area. The pressing force can also be estimated based on the change in the size of the touch area in the time sequence of the final touch position area trajectory to construct a more humane somatosensory feedback, thereby realizing human-computer interaction of intelligent early childhood education equipment.

[0092] In summary, the embodiment of the present invention relates to the field of interactive technology of teaching equipment. The method first obtains the touch points when the touch interaction is performed on the early childhood education device of the infrared touch panel, and the touch position area is formed by the touch points; respectively obtains the time period consisting of the continuous sampling moments of the touch points on the early childhood education device and the time period consisting of the continuous sampling moments of the touch points without the touch points as the touch time period and the blank time period; performs action change correlation analysis on the touch point changes in the two touch time periods adjacent to the blank time period, and determines the action change correlation of the blank time period; according to the action change correlation of the blank time period, the blank time period is screened to obtain the associated time period; based on the associated time period, the touch time period is updated; based on the updated touch time period, multiple connectivity graphs are constructed; in combination with the structural distribution characteristics of the connectivity graph, the node distinction confidence of the nodes in the connectivity graph is determined; the node similarity is weighted with the node distinction confidence as the weight, and the cluster cluster is determined based on the connectivity graph splitting clustering algorithm; the sampling moments belonging to the associated time period in the cluster cluster are extracted as the interaction moments; the touch position area is interpolated and supplemented for the interaction moments to obtain the final interaction trajectory formed by the touch position area. By extracting the changing characteristics of touch action positions, we can accurately analyze the correlation between the structural distribution of nodes to eliminate the problem of inaccurate analysis results caused by multiple touch operations and touch interruptions.

[0093] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0094] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A robot touch interaction control method, characterized in that: The method comprises the following steps: Acquire touch points generated when performing touch interaction on the early childhood education device of the infrared touch panel, and form a touch position area from the touch points; Respectively obtaining a time period consisting of continuous sampling moments in which a touch point exists on the early childhood education device and a time period consisting of continuous sampling moments in which no touch point exists as a touch time period and a blank time period; performing an action change correlation analysis on touch point changes in two touch time periods adjacent to the blank time period to determine an action change correlation of the blank time period; screening the blank time periods according to the action change correlation of the blank time period to obtain a correlated time period; and updating the touch time period based on the correlated time period; The method for obtaining the motion change correlation degree of the blank time period is as follows: obtaining the touch position change degree at each sampling moment in two touch time periods adjacent to the blank time period; determining the touch position interval change correlation degree at different sampling moments based on the touch position change degree at different sampling moments within the touch time period and the length of the touch time period; and determining the motion change correlation degree of the blank time period by combining the touch position interval change correlation degrees corresponding to the sampling moments belonging to the touch time period at the left and right endpoints of the blank time period. Based on the updated touch time period, multiple connectivity graphs are constructed; node distinction confidences of nodes in the connectivity graphs are determined based on the structural distribution characteristics of the connectivity graphs; node similarities are weighted using the node distinction confidences as weights, and clusters are determined based on a connectivity graph divisive clustering algorithm; The sampling moments belonging to the associated time period in the cluster are extracted as interaction moments; and the touch position areas are interpolated and supplemented for the interaction moments to obtain a final interaction trajectory formed by the touch position areas.

2. The robot touch interaction control method according to claim 1, characterized in that: The step of obtaining the touch position change degree at each sampling moment in two touch time periods adjacent to the blank time period includes: Any sampling moment within the touch time period is taken as the target moment, the difference in the number of touch points between the target moment and the previous sampling moment is taken as the numerator, the time difference between the target moment and the previous sampling moment is taken as the denominator, and the fraction formed by the numerator and denominator is taken as the touch position change degree at the target moment.

3. The robot touch interaction control method according to claim 1, characterized in that: The calculation formula of the interval change correlation is: ； in, is the interval change correlation degree of the touch position at the i-th sampling moment within the touch time period; is the normalization function; is the length of the touch time period at the i-th sampling moment; is the touch position change degree at the i-th sampling moment within the touch time period; is an exponential function with the natural constant e as the base; is the number of sampling moments in the touch time period of the i-th sampling moment; is the average value of the touch position change at all sampling moments within the touch time period.

4. The robot touch interaction control method according to claim 1, characterized in that: The step of determining the motion change correlation degree of the blank time period by combining the interval change correlation degrees of the touch positions corresponding to the sampling moments belonging to the touch time period at the left and right end points of the blank time period includes: The difference in interval change correlation between the touch positions corresponding to the sampling moments belonging to the touch time period at the left and right endpoints of the blank time period is calculated as the interval change degree; the negative correlation normalized value of the interval change degree is used as the action change correlation degree of the blank time interval.

5. The robot touch interaction control method according to claim 1, characterized in that: The step of constructing multiple connectivity graphs based on the updated touch time period includes: The centroid point of the touch position area corresponding to a single sampling moment within the updated touch time period is used as a node of the connectivity graph, the number of nodes in the action time period within the touch time period is used as the preset incremental clustering increment value, and the preset incremental clustering increment value is used as the number of nodes in the touch time period. Multiple connectivity graphs are constructed according to the connectivity graph splitting clustering algorithm.

6. The robot touch interaction control method according to claim 1, characterized in that: The calculation formula of the node differentiation confidence is: ； ； in, The node discrimination of the connectivity graph constructed for the i-th node in the r-th incremental clustering; Node discrimination confidence for the i-th node; is the interval change correlation degree of the touch position at the i-th sampling moment within the touch time period; The number of node connection edges of the first part of the connected graph constructed by incremental clustering for the i-th node; The number of nodes in the first part of the connected graph constructed by incremental clustering that is divided for the i-th node; The number of nodes in the second part of the connected graph constructed by incremental clustering divided for the i-th node; The number of nodes connecting the second part of the connected graph constructed by incremental clustering for the i-th node; R is the number of incremental clustering; norm is the normalization function; exp is the exponential function with the natural constant e as the base; is the absolute value symbol.

7. The robot touch interaction control method according to claim 1, characterized in that: The step of weighting the node similarity by using the node distinction confidence as a weight includes: The calculation formula for weighted node similarity is: ； in, The weighted node similarity between the a-th node and the b-th node; Calculate the node distinction confidence between the a-th node and the k-th node in the b-th node; is the initial node similarity between the a-th node and the b-th node.

8. The robot touch interaction control method according to claim 1, characterized in that: The updating of the touch time period based on the associated time period includes: The associated time period and the adjacent touch time period are merged and updated to obtain an updated touch time period.

9. The robot touch interaction control method according to claim 1, characterized in that: The step of screening the blank time periods according to the action change correlation of the blank time periods to obtain the correlated time periods includes: The blank time period whose corresponding interval change correlation degree is greater than the preset correlation degree threshold is used as the correlation time period.

Images