Self-adaptive game difficulty design method based on dynamic time warping

Abstract

The invention discloses an adaptive game difficulty design method, device and equipment based on dynamic time warping and a computer readable storage medium, and the method comprises the steps: obtaining a player skeleton point sequence corresponding to real-time actions of a player, and carrying out dynamic time warping algorithm comparison with a standard action template to generate a real-time action score; performing statistical analysis on the real-time action score by applying a preset sliding window to generate score trend data; judging the score trend data by applying a preset difficulty adjustment rule to generate a preliminary game difficulty adjustment instruction; generating a growth curve model representing the long-term ability change of the player based on the stored time sequence of the historical action scores of the player; based on the growth curve model, difficulty adjustment sensitivity parameters are determined; the initial game difficulty adjusting instruction and the difficulty adjusting sensitivity parameter are combined to generate the final game difficulty parameter, and the method has the advantage of being high in difficulty adjusting sensitivity.

Description

Technical Field

[0001] The present invention relates to the technical field of game difficulty design, and in particular to a method, device, equipment and computer-readable storage medium for adaptive game difficulty design based on dynamic time warping. Background Art

[0002] Dynamic difficulty adjustment is a key technology in modern video game design, aiming to provide optimal immersion and user experience by matching the game's challenge to the player's real-time abilities. Effective dynamic difficulty adjustment in games based on somatosensory or motion recognition requires an objective and accurate quantitative assessment of the player's physical movement completion. This process relies on using pose estimation algorithms to capture the player's real-time movements as a time series of skeletal key points for analysis.

[0003] However, existing game difficulty adjustment mechanisms have two main flaws. The first is static difficulty grading, which involves presetting fixed difficulty levels such as "easy," "normal," and "hard" before the game begins. The drawback of this method is that it is completely unable to adapt in real time to the player's actual performance during the game. The second is statistical adjustments based on historical data, such as adjusting the parameters of subsequent levels based on the player's average score or clearance time. Although this method has a certain degree of dynamism, its adjustment basis is still relatively rough. It ignores the refined evaluation of the quality of the player's actions themselves, responds slowly to immediate changes in player performance, and fails to consider the personalized growth curves of different players, resulting in its lack of adaptability.

[0004] In summary, existing game difficulty design methods either lack real-time performance or provide insufficient basis for evaluating player performance, failing to quantify action completion. They struggle to quickly respond to short-term fluctuations in player status and fail to fully consider long-term growth trends, resulting in a gaming experience that lacks true personalization and adaptability. Therefore, there is an urgent need for a new adaptive difficulty design method that can dynamically, responsively, and individually adjust game difficulty. Summary of the Invention

[0005] The embodiments of the present application aim to provide an adaptive game difficulty design method based on dynamic time regularization, so as to achieve dynamic, sensitive and personalized adjustment of game difficulty.

[0006] To achieve the above objectives, the present invention provides an adaptive game difficulty design method based on dynamic time warping, comprising:

[0007] Obtain the player's skeleton point sequence corresponding to the player's real-time action, and compare it with the standard action template through the dynamic time warping algorithm to generate a real-time action score that represents the player's instantaneous action completion degree;

[0008] Applying a preset sliding window to perform statistical analysis on the real-time action scores to generate score trend data representing the player's short-term performance;

[0009] Applying a preset difficulty adjustment rule to judge the score trend data to generate a preliminary game difficulty adjustment instruction;

[0010] Applying a time series prediction model to the stored time series of players' historical action scores to generate a growth curve model that represents the long-term changes in players' abilities;

[0011] Determining a difficulty adjustment sensitivity parameter for adjusting the intensity of the preliminary game difficulty adjustment instruction based on the growth curve model;

[0012] The preliminary game difficulty adjustment instruction and the difficulty adjustment sensitivity parameter are combined to generate a final game difficulty parameter, and the final game difficulty parameter is applied to the game engine.

[0013] In one embodiment, a player's skeleton point sequence corresponding to the player's real-time action is obtained and compared with a standard action template using a dynamic time warping algorithm to generate a real-time action score representing the instantaneous completion of the player's action, including:

[0014] Calculate the position difference and speed difference between the player's skeleton point sequence and the corresponding frame in the standard action template;

[0015] Performing weighted summation on the position difference value and the speed difference value to obtain an inter-frame cost;

[0016] Based on the inter-frame cost, the real-time action score is calculated using a dynamic time warping algorithm.

[0017] In one embodiment, a preset sliding window is applied to perform statistical analysis on the real-time action scores to generate score trend data representing the player's short-term performance, including:

[0018] Setting the size and moving step of the sliding window;

[0019] Each time the sliding window moves, calculating the arithmetic mean and variance of all real-time action scores within the window;

[0020] According to the arithmetic mean and the variance, score trend data containing mean and variance information is jointly constructed.

[0021] In one embodiment, applying a preset difficulty adjustment rule to the score trend data to generate a preliminary game difficulty adjustment instruction includes:

[0022] When the average value in the score trend data increases and the variance is lower than a first preset threshold, generating a preliminary game difficulty adjustment instruction to increase the difficulty;

[0023] When the average value of the score trend data decreases or the variance is higher than a second preset threshold, a preliminary game difficulty adjustment instruction for reducing the difficulty is generated.

[0024] In one embodiment, the time series prediction model is an exponential smoothing model;

[0025] The time series prediction model is applied to the stored player's historical action scores to generate a growth curve model that represents the player's long-term ability changes, including:

[0026] Set the smoothing coefficient α of the exponential smoothing method, where 0<α<1;

[0027] Apply the following recursive formula to iteratively process the time series of the player's historical action scores to calculate the smoothed score value S_t:

[0028] S_t=α·y_t+(1-α)·S_t-1,

[0029] Among them, y_t is the real-time action score at time t, and S_t-1 is the smoothed score value at time t-1;

[0030] The score value is used as the growth curve model.

[0031] In one embodiment, determining a difficulty adjustment sensitivity parameter for adjusting the intensity of the preliminary game difficulty adjustment instruction based on the growth curve model includes:

[0032] Extracting a gradient value representing the player's ability improvement rate from the growth curve model;

[0033] According to the gradient value, the corresponding difficulty adjustment sensitivity parameter is searched and determined in a preset gradient-sensitivity mapping table.

[0034] In one embodiment, the final game difficulty parameter includes at least one of game level complexity, non-player character behavior pattern, or the number of obstacles in the scene.

[0035] To achieve the above objectives, the present application also proposes an adaptive game difficulty design device based on dynamic time warping, comprising:

[0036] The real-time scoring module is used to obtain the player's skeleton point sequence corresponding to the player's real-time action and compare it with the standard action template through the dynamic time warping algorithm to generate a real-time action score that represents the player's instantaneous action completion degree;

[0037] a trend analysis module, configured to perform statistical analysis on the real-time action scores using a preset sliding window to generate score trend data representing a player's short-term performance;

[0038] A preliminary adjustment module, configured to apply a preset difficulty adjustment rule to the score trend data to generate a preliminary game difficulty adjustment instruction;

[0039] A growth modeling module is used to process the stored time series of players' historical action scores using a time series prediction model to generate a growth curve model that represents the player's long-term ability changes;

[0040] a sensitivity calibration module, configured to determine, based on the growth curve model, a difficulty adjustment sensitivity parameter for adjusting the intensity of the preliminary game difficulty adjustment instruction;

[0041] A parameter generation module is used to combine the preliminary game difficulty adjustment instruction and the difficulty adjustment sensitivity parameter to generate a final game difficulty parameter, and apply the final game difficulty parameter to the game engine.

[0042] To achieve the above-mentioned purpose, an embodiment of the present application also proposes an adaptive game difficulty design device based on dynamic time regularization, including a memory, a processor, and an adaptive game difficulty design program based on dynamic time regularization stored in the memory and runnable on the processor. When the processor executes the adaptive game difficulty design program based on dynamic time regularization, it implements the adaptive game difficulty design method based on dynamic time regularization as described in any one of the above items.

[0043] To achieve the above-mentioned purpose, an embodiment of the present application also proposes a computer-readable storage medium, on which an adaptive game difficulty design program based on dynamic time regularization is stored. When the adaptive game difficulty design program based on dynamic time regularization is executed by a processor, the adaptive game difficulty design method based on dynamic time regularization as described in any one of the above items is implemented.

[0044] From the above embodiments, it can be seen that the adaptive game difficulty design method based on dynamic time warping of the present application has the following beneficial effects:

[0045] 1. Achieved accurate, two-level dynamic evaluation of player performance

[0046] This application significantly improves the accuracy of player ability assessment through a two-layer dynamic analysis architecture. On the one hand, by applying a sliding window to statistically analyze a player's recent real-time scores, we can accurately determine their short-term competitive state. On the other hand, by applying a time series prediction model to a player's entire historical score, we can generate a growth curve model that filters out accidental fluctuations and better represents their true long-term ability.

[0047] 2. Provides a truly personalized and adaptive dynamic difficulty experience

[0048] This application innovatively combines a growth curve model that reflects a player's long-term growth trends with adjustment instructions that reflect their short-term performance. It dynamically calibrates the "sensitivity" of difficulty adjustments based on a player's long-term growth rate, allowing difficulty changes to match each player's unique learning curve, providing players of different skill levels with challenges that best suit their pace. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. 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 the structures shown in these drawings without paying any creative work.

[0050] Figure 1 This is a module structure diagram of an embodiment of a device for adaptive game difficulty design based on dynamic time warping according to the present invention;

[0051] Figure 2 2. Schematic diagram of a flow chart of an embodiment of a method for adaptive game difficulty design based on dynamic time warping of the present invention.

[0052] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0053] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0054] To better understand the above technical solutions, exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. Instead, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0055] It should be noted that in the claims, any reference signs placed between brackets shall not be construed as limiting the claims. The presence of "comprising" in the text does not exclude the presence of components or steps not listed in the claims. The quantifier "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The present invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In a unit claim that lists several means, several of these means may be embodied by the same hardware item. The use of "first", "second", and "third" etc. does not indicate any order and these words may be interpreted as names.

[0056] like Figure 1 As shown, Figure 1 It is a structural diagram of server 1 (also called adaptive game difficulty design device based on dynamic time regularization) of the hardware operating environment involved in the embodiment of the present invention.

[0057] The server of the embodiment of the present invention is a device with display function such as "Internet of Things devices", smart air conditioners, smart lights, smart power supplies with networking functions, AR / VR devices with networking functions, smart speakers, self-driving cars, PCs, smart phones, tablet computers, e-book readers, portable computers, etc.

[0058] like Figure 1 As shown, the server 1 includes: a memory 11 , a processor 12 and a network interface 13 .

[0059] The memory 11 includes at least one type of readable storage medium, including a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., an SD or DX memory), a magnetic memory, a magnetic disk, an optical disk, etc. In some embodiments, the memory 11 may be an internal storage unit of the server 1, such as a hard disk of the server 1. In other embodiments, the memory 11 may also be an external storage device of the server 1, such as a plug-in hard disk equipped on the server 1, a smart media card (SMC), a secure digital (SD) card, a flash card, etc.

[0060] Furthermore, the memory 11 may include both an internal storage unit of the server 1 and an external storage device. The memory 11 may be used not only to store application software installed on the server 1 and various data, such as the code of the adaptive game difficulty design program 10 based on dynamic time warping, but also to temporarily store data that has been output or is about to be output.

[0061] In some embodiments, the processor 12 can be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip, used to run the program code or process data stored in the memory 11, such as executing the adaptive game difficulty design program 10 based on dynamic time warping.

[0062] The network interface 13 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface), and is generally used to establish a communication connection between the server 1 and other electronic devices.

[0063] The network may be the Internet, a cloud network, a wireless fidelity (Wi-Fi) network, a personal area network (PAN), a local area network (LAN), and / or a metropolitan area network (MAN). Various devices in the network environment may be configured to connect to the communication network according to various wired and wireless communication protocols. Examples of such wired and wireless communication protocols may include, but are not limited to, at least one of the following: Transmission Control Protocol and Internet Protocol (TCP / IP), User Datagram Protocol (UDP), Hypertext Transfer Protocol (HTTP), File Transfer Protocol (FTP), ZigBee, EDGE, IEEE 802.11, Light Fidelity (Li-Fi), 802.16, IEEE 802.11s, IEEE 802.11g, multi-hop communication, wireless access point (AP), device-to-device communication, cellular communication protocol, and / or Bluetooth communication protocol, or a combination thereof.

[0064] Optionally, the server may further include a user interface, which may include a display and an input unit such as a keyboard. The optional user interface may also include a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display, which may also be referred to as a display screen or display unit, is used to display information processed in the server 1 and to display a visual user interface.

[0065] Figure 1 Only the server 1 having components 11-13 and the adaptive game difficulty design program 10 based on dynamic time warping is shown. It can be understood by those skilled in the art that Figure 1 The structure shown does not constitute a limitation on the server 1 , and the server 1 may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0066] In this embodiment, the processor 12 may be configured to call the adaptive game difficulty design program based on dynamic time warping stored in the memory 11 and perform the following operations:

[0067] Obtain the player's skeleton point sequence corresponding to the player's real-time action, and compare it with the standard action template through the dynamic time warping algorithm to generate a real-time action score that represents the player's instantaneous action completion degree;

[0068] Applying a preset sliding window to perform statistical analysis on the real-time action scores to generate score trend data representing the player's short-term performance;

[0069] Applying a preset difficulty adjustment rule to judge the score trend data to generate a preliminary game difficulty adjustment instruction;

[0070] Applying a time series prediction model to the stored time series of players' historical action scores to generate a growth curve model that represents the long-term changes in players' abilities;

[0071] Determining a difficulty adjustment sensitivity parameter for adjusting the intensity of the preliminary game difficulty adjustment instruction based on the growth curve model;

[0072] The preliminary game difficulty adjustment instruction and the difficulty adjustment sensitivity parameter are combined to generate a final game difficulty parameter, and the final game difficulty parameter is applied to the game engine.

[0073] Based on the hardware architecture of the aforementioned device for adaptive game difficulty design based on dynamic time warping, an embodiment of the present invention's method for adaptive game difficulty design based on dynamic time warping is proposed. This method aims to provide an adaptive difficulty design method that enables dynamic, sensitive, and personalized adjustment of game difficulty.

[0074] Reference Figure 2 , Figure 2 This is an embodiment of the adaptive game difficulty design method based on dynamic time warping of the present invention, and the adaptive game difficulty design method based on dynamic time warping includes the following steps:

[0075] S10. Obtain the player's skeleton point sequence corresponding to the player's real-time action, and compare it with the standard action template using a dynamic time warping algorithm to generate a real-time action score that represents the player's instantaneous action completion degree.

[0076] Specifically, the standard action template is a pre-stored time series data, which represents the ideal and standard execution process of a specific action that a player should complete in the game.

[0077] The real-time action score is a quantitative value, usually normalized to a range of 0 to 100, used to represent the degree to which a player's body movements conform to the standard action template at a certain moment or within a very short time window.

[0078] Specifically, the system first needs to obtain the player's real-time skeleton point sequence. In a specific embodiment, the acquisition process includes: first, a video stream containing the player's movements is collected in real time through a visual sensor (for example, a camera). Then, the system uses a preset, lightweight posture estimation model (for example, the MediaPipe Pose model) to process the video stream frame by frame. The model can identify and extract the three-dimensional coordinates of all key skeletal joints of the player's body from each frame of the image. Finally, the system attaches an accurate timestamp to the coordinate data extracted for each frame, and combines these continuous posture data with timestamps to form the player's skeleton point sequence.

[0079] After obtaining the player's skeleton point sequence, the system then compares it with the pre-stored standard action template using a dynamic time warping algorithm to generate the real-time action score. In a specific embodiment, the comparison process can be implemented by following the steps S11 to S13:

[0080] First, in step S11, when comparing the player's skeleton point sequence with any pair of corresponding frames in the standard template, the system will calculate two different difference values in parallel. The first is the "position difference value", which is usually obtained by calculating the Euclidean distance between the coordinate vectors of all corresponding joint points in the two frames. It reflects the accuracy of the player's posture in space. The second is the "speed difference value", which is obtained by calculating the difference in the speed or speed vector of the corresponding joint points in the two frames. It reflects the speed and rhythm of the player's movements. Then, in step S12, the system will perform a weighted summation on the position difference value and speed difference value calculated in the previous step to obtain a comprehensive "inter-frame cost". This weighted summation process gives different dimensional difference values different importance.

[0081] Finally, in step S13, the system uses this comprehensive inter-frame cost as a basis to calculate the final real-time action score through a DTW algorithm process that includes cost matrix construction and normalization. The process specifically includes:

[0082] The first step is to construct a local cost matrix. The system initializes a two-dimensional matrix D of size N x M, where N is the length of the player's skeletal point sequence and M is the length of the standard action template. The system iterates over all frames and calculates the inter-frame cost between the i-th frame of the player sequence and the j-th frame of the template sequence, filling the element D(i, j) in the i-th row and j-th column of matrix D with the inter-frame cost calculated between the i-th frame of the player sequence and the j-th frame of the template sequence.

[0083] The second step is to construct a cumulative cost matrix. After obtaining the local cost matrix D, the system constructs a cumulative cost matrix γ of the same size. The elements γ(i,j) of this matrix represent the minimum cumulative cost of the optimal alignment path from the starting point (1,1) to any point (i,j) in the matrix. The elements of this matrix are filled in using the following dynamic programming recursive formula:

[0084] γ(i,j)=D(i,j)+min{γ(i-1,j),γ(i-1,j-1),γ(i,j-1)}, where the boundary condition is γ(1,1)=D(1,1).

[0085] The third step is to determine the optimal path and minimum cumulative cost. After completing the cumulative cost matrix γ, the cumulative cost value γ(N,M) at the end point (N,M) of the matrix is the minimum cumulative cost between the two sequences. Starting from this end point, the system can use backtracking (i.e., reversely selecting the adjacent cell with the minimum cumulative cost) to find a path from the end point (N,M) back to the starting point (1,1). This path is the optimal regularized path.

[0086] The fourth step is to normalize and obtain the real-time action score. The system processes the minimum cumulative cost obtained in the previous step through a preset normalization function to obtain the final real-time action score. For example, the normalization function can be:

[0087] Real-time action score = 100*max(0,1-(minimum cumulative cost / preset maximum possible cost)).

[0088] For example, in step S12, the weighted summation formula for calculating the inter-frame cost can be specifically set as:

[0089] Inter-frame cost = 0.7 × position difference + 0.3 × speed difference;

[0090] This formula indicates that in this game's scoring system, posture accuracy (weighted 0.7) is more important than movement speed (weighted 0.3). For example, in a given frame, the position difference between the player's movement and the standard template is 10, and the speed difference is 20. The inter-frame cost between this pair of frames is: 0.7 * 10 + 0.3 * 20 = 7 + 6 = 13. The system uses this value of 13 as the basis for subsequent DTW calculations, rather than using 10 or 20 separately.

[0091] As you can understand, the calculation of inter-frame costs incorporates both position and velocity differences, allowing for a comprehensive assessment of player performance from both the spatial and temporal dimensions of the action. This results in a more comprehensive and accurate evaluation than traditional methods that rely solely on positional information. Furthermore, by employing a weighted summation of the differences across different dimensions, the evaluation algorithm's emphasis can be flexibly configured to meet the specific needs of game design. For example, some games may prioritize posture accuracy, while others may prioritize rhythm control, thus enhancing the flexibility and rationality of the scoring mechanism.

[0092] S20: Apply a preset sliding window to perform statistical analysis on the real-time action score to generate score trend data representing the player's short-term performance.

[0093] Specifically, a sliding window is a data processing mechanism that intercepts and analyzes a fixed-size, recent subset of continuous data from a continuously incoming data stream (such as a series of real-time action scores). As new data arrives, the window "slides" forward to always maintain a focus on recent data.

[0094] Score trend data is a structured data object that contains one or more statistical indicators. For example, it can contain mean and variance, which are used to jointly characterize the average performance level of players in a short time window and the stability of their performance.

[0095] In a specific embodiment, the above-mentioned process of applying the sliding window to perform statistical analysis can be implemented by following the steps S21 to S23:

[0096] First, in step S21, the system sets two key parameters of the sliding window: the size of the window and the step size of the movement. The window size defines the time range or number of scores covered by the analysis (for example, analyzing the most recent 10 scores), while the step size defines the frequency with which the system updates the trend analysis (for example, performing an analysis every time 5 new scores are obtained). Then, in step S22, every time the sliding window moves to a new position, the system calculates two core statistical indicators for all real-time action scores contained in the current window: the arithmetic mean (mean μ) and the variance (σ 2 The mean reflects the player's average performance level over a short period of time, while the variance reflects the stability of their performance; a larger variance indicates more unstable performance. Finally, in step S23, the system encapsulates the calculated arithmetic mean and variance into a structured score trend data object for use in subsequent steps.

[0097] For example, assume that the window size set in step S21 is 10 scores. At a certain moment, the 10 consecutive real-time action scores captured in the sliding window are: [85, 88, 90, 92, 89, 93, 95, 91, 94, 96]. In step S22, the system first calculates the arithmetic mean μ of these 10 scores, that is, μ = (85 + ... + 96) / 10 = 91.3. Then, based on this mean, the system calculates the variance σ using the formula 2 =(1 / 10)*Σ((score_i-91.3) 2 ), get a smaller variance value, such as σ 2 =9.81. In step S23, the system generates the following score trend data for this round of analysis: {mean: 91.3, variance: 9.81}. This data indicates that the player has recently performed well (high average score) and is stable (small variance), providing a basis for determining whether to increase the game difficulty in step S30.

[0098] It's understandable that applying a sliding window to statistically analyze a series of continuous real-time scores can smooth out the random fluctuations of individual scores, thereby more stably and reliably reflecting a player's true performance level and status trends over the short term than relying solely on instantaneous scores. Furthermore, by simultaneously calculating two different statistical metrics—mean (reflecting performance level) and variance (reflecting performance stability)—this provides a richer basis for subsequent difficulty adjustment rules, enabling more refined intelligent decisions that can distinguish between "stably good / bad" and "unstable fluctuations."

[0099] S30, applying a preset difficulty adjustment rule to judge the score trend data to generate a preliminary game difficulty adjustment instruction;

[0100] Specifically, difficulty adjustment rules are a set of pre-defined, if-then logical conditions that map input score trend data to specific difficulty adjustment actions. The initial game difficulty adjustment instruction, generated by this step, is an intermediate instruction indicating whether the game difficulty should be "increased," "decreased," or "remained unchanged." This instruction will subsequently undergo sensitivity calibration.

[0101] In a specific embodiment, the above-mentioned process of applying the difficulty adjustment rules to make judgments is implemented through a set of conditional judgment logic defined in steps S31 and S32:

[0102] On the one hand, in step S31, the system determines whether the average value in the score trend data shows an upward trend and whether its variance is below a first preset threshold value, which is used to indicate "high stability." Only when these two conditions are simultaneously met will the system determine that the player is currently in a good state of "stable progress" and, therefore, generate a preliminary game difficulty adjustment instruction to "increase difficulty." On the other hand, in step S32, the system determines whether the average value in the score trend data shows a downward trend or whether its variance is above a second preset threshold value, which is used to indicate "excessive state fluctuation." As long as either of these two conditions is met, the system will determine that the player may be currently experiencing difficulties or an unstable state, and, therefore, generate a preliminary game difficulty adjustment instruction to "reduce difficulty." If all of the above conditions are not met (for example, the average score is stable but not high enough), the system can generate an instruction to "maintain difficulty unchanged."

[0103] For example, continuing the example of the previous step, the system obtains a set of score trend data of {mean: 91.3, variance: 9.81}. Assume that the first preset variance threshold set by the system is 10.0. Since the current variance of 9.81 is lower than the threshold, and assuming that the system has determined that the current mean is on an upward trend by comparing the mean of the previous round, all the conditions of step S31 are met at this time, and the system will generate an instruction to "increase difficulty". As another example, suppose that the score trend data obtained in another round is {mean: 65.0, variance: 150.0}. Although its mean may not have decreased, since its variance of 150.0 is much higher than the set second preset variance threshold (for example, 25.0), the "variance is too high" condition in step S32 is met at this time, and the system will also generate a "reduce difficulty" instruction to help players whose status fluctuates too much to stabilize.

[0104] It's understandable that by comprehensively evaluating both the "average" (which reflects a player's performance level) and the "variance" (which reflects the stability of their performance), more refined and intelligent decisions can be made than relying solely on a single average score, effectively distinguishing between the two distinct player states of "consistently excellent" and "unstable excellent." Furthermore, by setting up different, asymmetric judgment logic for increasing and decreasing difficulty (increasing requires both conditions to be met, while decreasing only requires one), the entire difficulty adjustment system is more cautious when increasing difficulty, while being more proactive and responsive when helping players who are struggling, providing players with better frustration management and a flow experience.

[0105] S40. Apply a time series prediction model to the stored time series of the player's historical action scores to generate a growth curve model that represents the player's long-term ability changes, wherein the time series prediction model is an exponential smoothing model.

[0106] Specifically, the time series of a player's historical action scores is a data set that records all or part of the player's previous real-time action scores (for example, sampled at a specific period) since the player started the game and is arranged in chronological order.

[0107] A time series forecasting model is a mathematical or algorithmic model that can analyze historical time series data and learn its inherent patterns to predict future data trends.

[0108] The growth curve model is the output generated by the time series prediction model after processing historical scores. It is a relatively smooth curve that can represent the average ability level of players and their changing trends over a longer time span, effectively filtering out short-term, accidental score fluctuations.

[0109] Specifically, this step no longer focuses on the player's instantaneous or short-term performance. Instead, it applies a time series prediction model to the time series of all historical action scores accumulated by the player to generate a growth curve model that can characterize the player's long-term ability changes. The core purpose of this step is to model the player's overall ability level and its growth rate from a macro and longer time dimension. This enables the system to determine whether the player is in the "newbie rapid growth period", "platform bottleneck period" or "experienced stable period", thereby providing a more strategic and personalized macro guidance for subsequent difficulty adjustment strategies.

[0110] In a specific embodiment, the process of applying the exponential smoothing model to generate the growth curve model can be implemented by following the steps S41 to S43:

[0111] First, in step S41, the system sets a key parameter required for exponential smoothing: the smoothing coefficient α. This coefficient, which ranges from 0 to 1, determines the model's "memory" of new data. A larger α value means the model prioritizes recent player performance and is more sensitive to new scores; a smaller α value means the model prioritizes long-term historical averages, resulting in a smoother curve and less sensitive to short-term fluctuations. Next, in step S42, the system applies the classic exponential smoothing recursive formula: S_t = α·y_t + (1-α)·S_t-1, to iteratively process the entire time series of a player's historical action scores. Here, y_t is the player's actual, real-time action score at time t, and S_t-1 is the smoothed score calculated at the previous time t-1. Using this formula, each smoothed score at a new time is a weighted average of the current actual score and the previous smoothed score. Finally, in step S43, the system combines this series of iteratively calculated, smoothed scores, $S_t$, as the final output of the growth curve model.

[0112] For example, assuming the smoothing coefficient α is set to 0.1, at the beginning of the game, the player's score y_1 at time t=1 is 50 points, and the initial smoothing value S_1 at this time is also 50.

[0113] At time t=2, the player has improved and the score y_2 is 60 points, so the new smoothed value S_2=0.1×60+(1-0.1)×50=6+45=51.

[0114] At time t=3, the player performs exceptionally well and scores y_3 up to 90 points, but the new smoothed value S_3=0.1×90+(1-0.1)×51=9+45.9=54.9.

[0115] From this example, we can see that although the player has a very high instantaneous score at t=3, due to the small value of α, the growth curve model (composed of S1, S2, S3...) still maintains a steady, slowly rising trend, effectively filtering out this accidental "outburst", thereby more truly reflecting the objective fact that the player, as a novice, has an average ability that is gradually improving.

[0116] As can be understood, by applying an exponential smoothing model to smooth a player's historical score series, which contains a large number of short-term fluctuations, it can generate a growth curve model that effectively filters out randomness and noise interference and better represents the player's true long-term ability level. Furthermore, because this model can be continuously updated online through a simple recursive formula, the computational cost is extremely low. This allows this method to provide a stable, reliable, and real-time evolving macro-decision-making basis for subsequent personalized difficulty adjustment strategies without consuming large amounts of computing resources.

[0117] S50: Determine a difficulty adjustment sensitivity parameter for adjusting the intensity of the preliminary game difficulty adjustment instruction based on the growth curve model;

[0118] Specifically, the difficulty adjustment sensitivity parameter is a factor used to adjust or scale the "strength" of the preliminary game difficulty adjustment instructions generated by short-term performance analysis. A higher sensitivity parameter means that the game difficulty will adjust significantly and quickly based on short-term performance; a lower parameter means that the adjustment will be more gradual and conservative.

[0119] Specifically, this step uses the growth curve model generated in the previous step to determine the difficulty adjustment sensitivity parameter used to adjust the intensity of the initial game difficulty adjustment commands. Its core purpose is to dynamically determine the system's "reaction speed" and "reaction intensity" to short-term fluctuations in a player's state based on the player's long-term ability trends. This allows difficulty adjustments to adapt not only to the player's immediate performance, but also to their different learning stages (such as the novice stage or the plateau stage), achieving true personalization.

[0120] In a specific embodiment, the process of determining the difficulty adjustment sensitivity parameter is implemented through steps S51 and S52:

[0121] In step S51, the system extracts a gradient value that can characterize the player's recent ability improvement rate from the growth curve model generated in S40. The extraction process specifically includes: first, the system selects the data within the most recent preset time window (for example, the smoothed value of the last 100 scores) from a series of smoothed score values representing the growth curve model. Then, the difference between the last smoothed score value and the first smoothed score value in the time window is calculated. Finally, this difference is divided by the actual game time corresponding to the time window, and the quotient obtained is the gradient value. Mathematically, this gradient value is the average slope of the growth curve in this window, and its physical meaning is the average ability improvement (score growth) of the player in unit time.

[0122] After calculating the gradient value that quantifies the player's progress, the system then searches a preset "gradient-sensitivity mapping table" based on the gradient value in step S52 to determine the current difficulty adjustment sensitivity parameter. This mapping table stores the correspondence between gradient value intervals and sensitivity parameters.

[0123] For example, a gradient-sensitivity map can be designed as:

[0124] When the gradient value is > 0.5 (representing a novice player whose ability improves rapidly), the corresponding sensitivity parameter is 1.5 (high sensitivity).

[0125] When 0.1<Gradient value≤0.5 (representing an average player, steady progress), the corresponding sensitivity parameter is 1.0 (standard sensitivity).

[0126] When the gradient value is ≤0.1 (representing an experienced player whose ability tends to be stable or reaches a bottleneck), the corresponding sensitivity parameter is 0.7 (low sensitivity).

[0127] Suppose the system calculates a player's recent growth curve gradient value of 0.8 in S51. In S52, the system queries the mapping table. Since 0.8 > 0.5, the system determines that the current difficulty adjustment sensitivity parameter should be 1.5. This parameter will be used in the next step to amplify the preliminary difficulty adjustment instructions generated by the short-term performance analysis.

[0128] As can be understood, calculating the gradient value of the growth curve model objectively quantifies the player's long-term ability improvement rate. Therefore, it provides a reliable mathematical basis for the system to determine the different growth stages of players (such as the novice stage and the plateau stage). Furthermore, by mapping this quantified gradient value into a sensitivity parameter used to adjust the intensity of difficulty adjustments, the adjustment strategy of the entire adaptive difficulty system can be matched to the player's personalized learning curve, providing more timely challenges for players who are improving rapidly and a more stable environment for players who are in the plateau stage, thereby greatly enhancing the personalized experience and long-term appeal of the game.

[0129] S60: Combining the preliminary game difficulty adjustment instruction and the difficulty adjustment sensitivity parameter to generate a final game difficulty parameter, and applying the final game difficulty parameter to the game engine. The final game difficulty parameter includes at least one of game level complexity, non-player character behavior patterns, or the number of obstacles in the scene.

[0130] Specifically, the final game difficulty parameter is a set of quantitative values that can be directly read and applied by the game engine to change the specific challenging elements in the game.

[0131] Specifically, in step S60, the system combines the preliminary adjustment instructions derived from the short-term performance analysis with the sensitivity parameters derived from the long-term growth curve analysis. The core purpose of this step is to modulate a directional, tactical adjustment decision (i.e., the "increase" or "decrease" difficulty instruction output from S30) through a strategic coefficient that reflects the player's long-term ability (i.e., the sensitivity parameter output from S50), thereby generating a precise, quantified final instruction that can be directly executed by the game engine.

[0132] In one specific embodiment, the above-described combination and generation process is achieved by using the sensitivity parameter as a multiplier for the adjustment step size. First, the system presets a base difficulty adjustment step size. For example, each adjustment results in a base change of 0.1 m / s in the movement speed of non-player characters (NPCs). When a preliminary "increase difficulty" instruction is generated at S30, the system multiplies this base step size by the difficulty adjustment sensitivity parameter determined at S50 to obtain the final actual adjustment amount. This actual adjustment amount is added to the current game parameters to generate the final game difficulty parameters, which are then applied by the game engine. As defined in this embodiment, these final game difficulty parameters can be specifically reflected in various aspects, such as increasing the complexity of the game level (e.g., generating a more complex maze), changing the behavior of non-player characters (NPCs) (e.g., making them attack more frequently or move faster), or increasing the number of obstacles in the scene.

[0133] For example, for a novice player with a growth curve gradient of 0.8, the system determines a sensitivity parameter of 1.5. Assume that the initial instruction generated by S30 is "increase difficulty" and the basic speed adjustment step is 0.1 m / s. Then, the final actual speed adjustment is 0.1*1.5=0.15 m / s. The game engine will increase the NPC's current speed by 0.15 m / s. In contrast, for an experienced player with a gradient of 0.1, the system determines a sensitivity parameter of 0.7. When the same "increase difficulty" instruction is received, the actual speed adjustment is 0.1*0.7=0.07 m / s. It can be seen that for novices, the difficulty increase is more radical to match their rapid progress; while for veterans, it is more gentle and conservative to avoid excessive frustration during their ability bottleneck period.

[0134] It's understandable that by combining sensitivity parameters reflecting a player's long-term growth trends with preliminary adjustment instructions reflecting their short-term performance, a final difficulty adjustment decision can be generated that both responds to the player's immediate state and fully considers their overall ability level. Furthermore, this mechanism, combining long-term strategic calibration with short-term tactical adjustments, makes the game's difficulty changes highly personalized for each player, providing players at different learning stages with challenges that best suit their pace, thereby maximizing the game's playability and long-term user retention.

[0135] Based on the above embodiments, the adaptive game difficulty design method based on dynamic time warping of the present application has at least the following beneficial effects:

[0136] 1. Achieved accurate, two-level dynamic evaluation of player performance

[0137] This application significantly improves the accuracy of player ability assessment through a two-layer dynamic analysis architecture. On the one hand, by applying a sliding window to statistically analyze a player's recent real-time scores, we can accurately determine their short-term competitive state. On the other hand, by applying a time series prediction model to a player's entire historical score, we can generate a growth curve model that filters out accidental fluctuations and better represents their true long-term ability.

[0138] 2. Provides a truly personalized and adaptive dynamic difficulty experience

[0139] This application innovatively combines a growth curve model that reflects a player's long-term growth trends with adjustment instructions that reflect their short-term performance. It dynamically calibrates the "sensitivity" of difficulty adjustments based on a player's long-term growth rate, allowing difficulty changes to match each player's unique learning curve, providing players of different skill levels with challenges that best suit their pace.

[0140] The embodiment of the present invention further provides an adaptive game difficulty design device based on dynamic time warping, the adaptive game difficulty design device based on dynamic time warping comprising:

[0141] The real-time scoring module is used to obtain the player's skeleton point sequence corresponding to the player's real-time action and compare it with the standard action template through the dynamic time warping algorithm to generate a real-time action score that represents the player's instantaneous action completion degree;

[0142] a trend analysis module, configured to perform statistical analysis on the real-time action scores using a preset sliding window to generate score trend data representing a player's short-term performance;

[0143] A preliminary adjustment module, configured to apply a preset difficulty adjustment rule to the score trend data to generate a preliminary game difficulty adjustment instruction;

[0144] A growth modeling module is used to process the stored time series of players' historical action scores using a time series prediction model to generate a growth curve model that represents the player's long-term ability changes;

[0145] a sensitivity calibration module, configured to determine, based on the growth curve model, a difficulty adjustment sensitivity parameter for adjusting the intensity of the preliminary game difficulty adjustment instruction;

[0146] A parameter generation module is used to combine the preliminary game difficulty adjustment instruction and the difficulty adjustment sensitivity parameter to generate a final game difficulty parameter, and apply the final game difficulty parameter to the game engine.

[0147] Among them, the steps for implementing each functional module of the adaptive game difficulty design device based on dynamic time regularization can refer to the various embodiments of the adaptive game difficulty design method based on dynamic time regularization of the present invention, and will not be repeated here.

[0148] In addition, an embodiment of the present invention further provides a computer-readable storage medium. The computer-readable storage medium can be any one of a hard disk, a multimedia card, an SD card, a flash memory card, an SMC, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable compact disk read-only memory (CD-ROM), a USB memory, or any combination thereof. The computer-readable storage medium includes an adaptive game difficulty design program 10 based on dynamic time warping. The specific implementation of the computer-readable storage medium of the present invention is substantially the same as the specific implementation of the adaptive game difficulty design method based on dynamic time warping and the server 1 described above, and will not be repeated here.

[0149] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0150] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0151] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0152] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0153] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0154] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A method for adaptive game difficulty design based on dynamic time warping, characterized in that: include: Obtain the player's skeleton point sequence corresponding to the player's real-time action, and compare it with the standard action template through the dynamic time warping algorithm to generate a real-time action score that represents the player's instantaneous action completion degree; Applying a preset sliding window to perform statistical analysis on the real-time action scores to generate score trend data representing the player's short-term performance; Applying a preset difficulty adjustment rule to judge the score trend data to generate a preliminary game difficulty adjustment instruction; Applying a time series prediction model to the stored time series of players' historical action scores to generate a growth curve model that represents the long-term changes in players' abilities; Determining a difficulty adjustment sensitivity parameter for adjusting the intensity of the preliminary game difficulty adjustment instruction based on the growth curve model; The preliminary game difficulty adjustment instruction and the difficulty adjustment sensitivity parameter are combined to generate a final game difficulty parameter, and the final game difficulty parameter is applied to the game engine.

2. The adaptive game difficulty design method based on dynamic time warping according to claim 1, characterized in that: Obtain the player's skeleton point sequence corresponding to the player's real-time action and compare it with the standard action template through the dynamic time warping algorithm to generate a real-time action score that represents the player's instantaneous action completion, including: Calculate the position difference and speed difference between the player's skeleton point sequence and the corresponding frame in the standard action template; Performing weighted summation on the position difference value and the speed difference value to obtain an inter-frame cost; Based on the inter-frame cost, the real-time action score is calculated using a dynamic time warping algorithm.

3. The adaptive game difficulty design method based on dynamic time warping according to claim 1, characterized in that: Performing statistical analysis on the real-time action scores using a preset sliding window to generate score trend data representing the player's short-term performance, including: Setting the size and moving step of the sliding window; Each time the sliding window moves, calculating the arithmetic mean and variance of all real-time action scores within the window; According to the arithmetic mean and the variance, score trend data including mean and variance information is jointly constructed.

4. The adaptive game difficulty design method based on dynamic time warping according to claim 1, characterized in that: Applying a preset difficulty adjustment rule to judge the score trend data to generate a preliminary game difficulty adjustment instruction, including: When the average value in the score trend data increases and the variance is lower than a first preset threshold, generating a preliminary game difficulty adjustment instruction to increase the difficulty; When the average value of the score trend data decreases or the variance is higher than a second preset threshold, a preliminary game difficulty adjustment instruction for reducing the difficulty is generated.

5. The adaptive game difficulty design method based on dynamic time warping according to claim 1, characterized in that: The time series prediction model is an exponential smoothing model; The time series prediction model is applied to the stored player's historical action scores to generate a growth curve model that represents the player's long-term ability changes, including: Set the smoothing coefficient α of the exponential smoothing method, where 0<α<1; Apply the following recursive formula to iteratively process the time series of the player's historical action scores to calculate the smoothed score value S_t: S_t=α·y_t+(1-α)·S_t-1, Where y_t is the real-time action score at time t, and S_t-1 is the smoothed score at time t-1; The score value is used as the growth curve model.

6. The adaptive game difficulty design method based on dynamic time warping according to claim 1, characterized in that: Determining a difficulty adjustment sensitivity parameter for adjusting the intensity of the preliminary game difficulty adjustment instruction based on the growth curve model includes: Extracting a gradient value representing the player's ability improvement rate from the growth curve model; According to the gradient value, the corresponding difficulty adjustment sensitivity parameter is searched and determined in a preset gradient-sensitivity mapping table.

7. The adaptive game difficulty design method based on dynamic time warping according to claim 1, characterized in that: The final game difficulty parameter includes at least one of game level complexity, non-player character behavior pattern, or the number of obstacles in the scene.

8. An adaptive game difficulty design device based on dynamic time warping, characterized in that: include: The real-time scoring module is used to obtain the player's skeleton point sequence corresponding to the player's real-time action and compare it with the standard action template through the dynamic time warping algorithm to generate a real-time action score that represents the player's instantaneous action completion degree; a trend analysis module, configured to perform statistical analysis on the real-time action scores using a preset sliding window to generate score trend data representing a player's short-term performance; A preliminary adjustment module, configured to apply a preset difficulty adjustment rule to the score trend data to generate a preliminary game difficulty adjustment instruction; A growth modeling module is used to process the stored time series of players' historical action scores using a time series prediction model to generate a growth curve model that represents the player's long-term ability changes; a sensitivity calibration module, configured to determine, based on the growth curve model, a difficulty adjustment sensitivity parameter for adjusting the intensity of the preliminary game difficulty adjustment instruction; The parameter generation module is used to combine the preliminary game difficulty adjustment instruction and the difficulty adjustment sensitivity parameter to generate a final game difficulty parameter, and apply the final game difficulty parameter to the game engine.

9. An adaptive game difficulty design device based on dynamic time warping, characterized in that: It includes a memory, a processor, and an adaptive game difficulty design program based on dynamic time regularization stored in the memory and executable on the processor. When the processor executes the adaptive game difficulty design program based on dynamic time regularization, the adaptive game difficulty design method based on dynamic time regularization as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores an adaptive game difficulty design program based on dynamic time regularization. When the adaptive game difficulty design program based on dynamic time regularization is executed by a processor, the adaptive game difficulty design method based on dynamic time regularization as described in any one of claims 1-7 is implemented.

Images