A toy robot humanization interaction feedback system with emotional inertia mechanism

Abstract

The application relates to the technical field of human-computer interaction, and discloses a toy robot personification interaction feedback system with an emotional inertia mechanism, and specifically comprises the following steps: S1, the system collects interaction signals in real time through a perception layer, and extracts emotional features by using a deep learning model; S2, the toy robot performs transition by using a mathematical model with inertia to eliminate action jumps caused by sensor noise; and S3, state parameters calculated in S2 are mapped into physical action flow of the toy robot, so that LED feedback and action execution are realized. The application applies a BERT+BiLSTM model (HMT-BB) to a method for synchronously predicting conversation emotion and intensity in the toy robot, uses an emotional inertia operator to realize a closed-loop control algorithm for smooth transition of a psychological state in a V-A space, and dynamically adjusts a fusion weight according to the reliability of sensor signals (such as environmental noise and illumination).

Description

Technical Field

[0001] This invention relates to the field of human-computer interaction technology, specifically to a toy robot anthropomorphic interactive feedback system with an emotional inertia mechanism. Background Technology

[0002] Most current toy robots (such as smart dolls and robotic dogs) use discrete trigger logic: that is, after clicking a sensor or recognizing a specific word, the corresponding recording is played or a preset action is executed immediately. Some high-end toys have begun to introduce continuous emotion prediction (VA space) and basic speech recognition, and even tried to use multi-task deep learning models (such as HMT-BB) to identify the intensity of emotions in children's conversations.

[0003] The shortcomings of existing technology:

[0004] Ⅰ: Mechanical vibration (a common problem in existing toys): Because toys usually use low-cost sensors and servo motors, there is a lot of signal interference, which causes the robot's expression to frequently jump between "laughing" and "dull" without transition, lacking a sense of life.

[0005] II: Monotonous and rigid feedback: Unable to understand the long-term logic of the conversation, and unable to express emotions with a "slow start" or "lingering effect" like a real pet.

[0006] III: Conflict between computing efficiency and privacy: Toys are often sensitive to chip computing power, and parents are extremely concerned about children's voice privacy.

[0007] Therefore, it is necessary to design a toy robot anthropomorphic interactive feedback system with an emotional inertia mechanism. Summary of the Invention

[0008] The purpose of this invention is to provide a toy robot anthropomorphic interactive feedback system with an emotional inertia mechanism to solve the problems in the prior art.

[0009] The objective of this invention can be achieved through the following technical solutions:

[0010] A toy robot anthropomorphic interactive feedback system with an emotional inertia mechanism specifically includes the following steps:

[0011] S1. The system collects interaction signals in real time through the perception layer and extracts emotional features using a deep learning model.

[0012] S2. The toy robot uses a mathematical model with "inertia" to transition, in order to eliminate motion jumps caused by sensor noise;

[0013] S3. The state parameters calculated from S2 are mapped to the physical motion flow of the toy robot to realize LED feedback and motion execution.

[0014] Furthermore, the emotional features in S1 include semantic feature extraction (toy semantic brain):

[0015] The system inputs the collected dialogue text into the HMT-BB hybrid multi-task model. First, it uses a pre-trained BERT layer to understand the context of the text through a bidirectional attention mechanism. Then, it uses a BiLSTM layer to capture the sequence dependencies between words and accurately extract the trend of emotion flowing with the progress of the discourse. Finally, it outputs an emotion intensity vector containing valence and arousal.

[0016] Furthermore, the emotional features in S1 include visual feature extraction (expression capture):

[0017] For toy-grade low-pixel cameras, a lightweight convolutional neural network is used to extract facial key points. Referring to the continuous emotion prediction method, the system outputs instantaneous visual emotion coordinates.

[0018] Furthermore, the emotional features in S1 include multimodal weighted fusion:

[0019] The system calculates the comprehensive instantaneous emotional target point using the following formula:

[0021] Furthermore, the evolution formula used for the transition in S2 is as follows:

[0022] In the above formula, (historical mood) refers to the robot's internal psychological state at the previous moment. It acts as an "anchor" to firmly hold onto emotions and achieve the memory effect.

[0023] (Instantaneous stimulus): The emotional signal given by the user in this second;

[0024] (Emotional inertia operator / damping coefficient): determines the toy's reaction speed.

[0025] Furthermore, S2 has an adaptive mechanism, including: if the HMT-BB model detects a drastic shift in the user's emotion (such as from "Joy" to "Anger") and the confidence level is high, the system will temporarily increase the value to make a sensitive response; otherwise, the smaller value will filter out about 70% of the sensor's environmental noise interference.

[0026] Furthermore, the LED eye array feedback in S3 includes:

[0027] The valence is mapped to the color temperature of the LED color, and the arousal is mapped to the flashing frequency of the breathing light.

[0028] Furthermore, the silk-sliding operation in S3 includes:

[0029] This algorithm ensures that the servo moves from a "dejected look" to a "look up" in a continuous, smooth trajectory. Even when using inexpensive toy-grade servos, the continuous coordinate guidance can prevent instantaneous jamming or jitter.

[0030] The beneficial effects of this invention are:

[0031] 1. This invention describes a method for synchronously predicting dialogue emotion and intensity using the BERT+BiLSTM model (HMT-BB) in toy robots, a closed-loop control algorithm for smooth transition of psychological states in the VA space using an "emotional inertia operator", and a method for dynamically adjusting fusion weights based on the reliability of sensor signals (such as environmental noise and illumination).

[0032] 2. The fusion node of this invention merges the semantic coordinates output by the HMT-BB model with the coordinates extracted by the sensor to form an instantaneous target. The core operator module obtains the historical state returned by the time delay unit and performs linear weighted evolution according to the inertia coefficient adjusted in real time, thereby generating a smooth psychological state at the current moment. This design ensures that the toy robot can evolve along a continuous trajectory when performing facial expressions or body movements, avoiding instantaneous jumps caused by sensor noise. Attached Figure Description

[0033] The invention will now be further described with reference to the accompanying drawings.

[0034] Figure 1 This is a schematic diagram of the anthropomorphic interactive feedback system for toy robots of the present invention. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] Example 1:

[0037] A toy robot anthropomorphic interactive feedback system with an emotional inertia mechanism, combined with Figure 1 As shown, the anthropomorphic interactive feedback system for toy robots is based on the anthropomorphic evolutionary control logic of toy robots according to "emotional inertia". It follows a complete closed loop of "multi-dimensional perception - deep understanding - temporal buffer - anthropomorphic mapping" and aims to upgrade toy robots from simple "electronic switch feedback" to intelligent life forms with "psychological coherence".

[0038] Specifically, the steps include the following:

[0039] S1, Multimodal "Emotion Capture" and Intensity Prediction

[0040] The system collects interaction signals in real time through the perception layer and extracts emotional features using a deep learning model;

[0041] The emotional characteristics include the following:

[0042] Ⅰ Semantic Feature Extraction (Toy Semantic Brain): The system inputs the collected dialogue text into the HMT-BB hybrid multi-task model. First, it uses a pre-trained BERT layer to understand the context of the text through a bidirectional attention mechanism; then, it uses a BiLSTM layer to capture the sequence dependencies between words and accurately extract the flow trend of emotion as the discourse progresses; finally, it outputs an emotion intensity vector containing valence and arousal.

[0043] II. Visual Feature Extraction (Face Capture): For toy-grade low-resolution cameras, a lightweight convolutional neural network is used to extract facial key points. Referring to continuous emotion prediction methods, the system outputs instantaneous visual emotion coordinates.

[0044] III. Multimodal Weighted Fusion: The system calculates the comprehensive instantaneous sentiment target point using the following formula:

[0046] S2, Core Algorithm—"Emotional Inertia" Buffer (The Heart of the Algorithm)

[0047] The core of this technical solution is that the toy robot does not jump directly to the instantaneous target point, but transitions through a mathematical model with "inertia" to eliminate motion jumps caused by sensor noise.

[0048] The evolution formula used in the transition process is as follows:

[0049] In the above formula, (historical mood) refers to the robot's internal psychological state at the previous moment. It acts as an "anchor" to firmly hold onto emotions and achieve the memory effect.

[0050] (Instantaneous stimulus): The emotional signal given by the user in this second.

[0051] (Emotional inertia operator / damping coefficient): determines the toy's reaction speed.

[0052] It also features an adaptive mechanism, including: if the HMT-BB model detects a sharp shift in the user's emotion (such as from "Joy" to "Anger") and the confidence level is high, the system will temporarily increase the level to make a more sensitive response; conversely, the lower level will filter out about 70% of the sensor's environmental noise interference.

[0053] S3, Anthropomorphic "Motion Trajectory" Mapping (Output)

[0054] The smooth state calculated in S2 is mapped to the physical motion flow of the toy robot, enabling LED feedback and motion execution.

[0055] I. LED Eye Array Feedback: Mapping Valence to the color temperature of the LED color and Arousal to the flashing frequency of the breathing light.

[0056] II. Smooth Execution: This algorithm ensures that the servo's transition from a "dejected" to a "looking up" motion is a continuous, smooth trajectory. Even with inexpensive, toy-grade servos, continuous coordinate guidance can prevent momentary jerking or vibration.

[0057] The emotional inertia buffer logic includes: the fusion node combines the semantic coordinates output by the HMT-BB model with the coordinates extracted by the sensors to form an instantaneous target. The core operator module obtains the historical state returned by the time delay unit and performs linear weighted evolution based on the real-time adjusted inertia coefficient, thereby generating a smooth psychological state at the current moment. This design ensures that the toy robot can evolve along a continuous trajectory when performing facial expressions or body movements, avoiding instantaneous jumps caused by sensor noise.

[0058] Example 2:

[0059] Based on Example 1, the BiLSTM in the HMT-BB engine is replaced with a GRU or Transformer structure to optimize computation speed.

[0060] Among them, the emotional inertia algorithm (smoothing logic) is replaced by:

[0061] Ⅰ Alternative filtering algorithm:

[0062] The core formula in the embodiment is essentially a first-order low-pass filter logic.

[0063] Alternative solutions include using a Kalman filter or a second-order Butterworth filter to handle the evolution of the sentiment coordinates and obtain a smoother dynamic curve.

[0064] II. Nonlinear attenuation:

[0065] The inertial operator can be replaced by a nonlinear decay function based on an exponential or logarithmic distribution to simulate the asymmetric psychological characteristics of emotions under extreme stimulation, which are characterized by "rapid outbreak and slow dissipation".

[0066] Execution method: In addition to physical actions, this system can also be used for the emotional control of virtual avatars on the screen.

[0067] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A toy robot anthropomorphic interactive feedback system with an emotional inertia mechanism, characterized in that, Specifically, the steps include the following: S1. The system collects interaction signals in real time through the perception layer and extracts emotional features using a deep learning model. S2. The toy robot uses a mathematical model with "inertia" to transition, in order to eliminate motion jumps caused by sensor noise; S3. The state parameters calculated from S2 are mapped to the physical motion flow of the toy robot to realize LED feedback and motion execution.

2. The anthropomorphic interactive feedback system for a toy robot with an emotional inertia mechanism according to claim 1, characterized in that, The emotional features in S1 include semantic feature extraction (toy semantic brain): The system inputs the collected dialogue text into the HMT-BB hybrid multi-task model, and first uses a bidirectional attention mechanism through a pre-trained BERT layer to understand the context of the text. Subsequently, a BiLSTM layer is used to capture the sequence dependencies between words, accurately extracting the flow trend of emotion as the discourse progresses; the final output is an emotion intensity vector containing valence and arousal. .

3. The anthropomorphic interactive feedback system for a toy robot with an emotional inertia mechanism according to claim 2, characterized in that, The emotional features in S1 include visual feature extraction (expression capture): For toy-grade low-pixel cameras, a lightweight convolutional neural network is used to extract facial key points. Referring to continuous emotion prediction methods, the system outputs instantaneous visual emotion coordinates. .

4. The anthropomorphic interactive feedback system for a toy robot with an emotional inertia mechanism according to claim 3, characterized in that, The emotional features in S1 include multimodal weighted fusion: The system calculates the comprehensive instantaneous emotional target point using the following formula: 。 5. The anthropomorphic interactive feedback system for a toy robot with an emotional inertia mechanism according to claim 1, characterized in that, The evolution formula used for the transition in S2 is as follows:

6. In the above formula, (Historical state of mind): The robot's internal psychological state at the previous moment, which acts as an "anchor" to firmly hold onto emotions and achieve a memory effect; (Instantaneous stimulus): The emotional signal given by the user in this second; (Emotional inertia operator / damping coefficient): determines the toy's reaction speed.

7. The anthropomorphic interactive feedback system for a toy robot with an emotional inertia mechanism according to claim 5, characterized in that, The S2 includes an adaptive mechanism, which includes: if the HMT-BB model detects a drastic shift in the user's emotion (e.g., from "Joy" to "Anger"), and the confidence level is high, the system will temporarily increase the [something]. To make a sensitive response; conversely, small It will filter out about 70% of the sensor's environmental noise interference.

8. The anthropomorphic interactive feedback system for a toy robot with an emotional inertia mechanism according to claim 1, characterized in that, The LED eye array feedback in S3 includes: The valence is mapped to the color temperature of the LED color, and the arousal is mapped to the flashing frequency of the breathing light.

9. The anthropomorphic interactive feedback system for a toy robot with an emotional inertia mechanism according to claim 1, characterized in that, The silk-sliding operation in S3 includes: This algorithm ensures that the servo's transition from a "dejected" to a "looking up" motion is a continuous, smooth trajectory, even when using inexpensive, toy-grade servos. Continuous coordinate guidance can also avoid instantaneous lag or jitter.

Images