An electronic pet emotion simulation system and method based on multi-sensor feedback
By simulating the emotions of electronic pets through a multi-sensor feedback system and a nonlinear dynamic system, the continuity and inertia of emotions are achieved, which enhances the realism and personalization of electronic pet interactions and solves the problems of monotonous reactions and uniform personalities in existing technologies.
Patent Information
- Application Number
- CN202610656123.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-05-13
- Publication Date
- 2026-07-31
AI Technical Summary
The current electronic pets have a limited number of emotional states and fixed state transition logic, resulting in a strong mechanical feel. They cannot simulate the spontaneous fluctuations and intrinsic motivations of living organisms' emotions. Furthermore, electronic pets of the same model have the same personality, lacking personalized features, which reduces the immersiveness and fun of the interaction.
Employing a multi-sensor feedback system, combined with a nonlinear dynamic system and an internal driving model, the system receives sensor signals through a multimodal perception interface layer. It utilizes a core emotion generation engine and a personality parameter module to simulate the emotional state of electronic pets, achieving the continuity, inertia, and spontaneous fluctuation of emotions. Furthermore, it showcases unique personalities through a behavior expression mapping module.
It enhances the realism and immersion of electronic pet interaction, enabling electronic pets to exhibit unique emotional states and proactive interactive behaviors, establishing personalized emotional connections, and solving the problems of monotonous reactions and uniform personalities in existing electronic pet technologies.
Smart Images
Figure CN122491339A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of computer technology of artificial intelligence, and more specifically to an electronic pet emotion simulation system and method based on multi-sensor feedback. Background Technology
[0002] With the rapid development of robotics, artificial intelligence, and virtual reality technologies, electronic pets, as alternatives to or supplements to traditional pets, have gradually entered people's lives. Representative electronic pet products, such as Sony's AIBO robot dog, Foxconn's intelligent robot cat, and various mobile virtual pet apps, have all gained widespread market recognition. Users' expectations for electronic pets have risen from simple command response or entertainment functions to demanding more realistic, rich, and lifelike emotional interaction capabilities to build deeper emotional connections.
[0003] Several technical solutions for processing the emotions of electronic pets have emerged in the prior art. For example, Chinese Patent: 202411115983.3, "An AI-based Pet Emotion Recognition System," discloses a solution that collects multimodal data from real pets through a sensor network, such as video, sound, body temperature, and heart rate. It then uses sparse coding and dictionary learning to extract key features and utilizes capsule networks and generative adversarial networks to identify the pet's emotional state. Simultaneously, it tracks emotional change trends through graph neural networks. This solution has achieved effective technical results in recognizing the emotions of real biological pets.
[0004] However, in the field of electronic pet technology, the above-mentioned solutions and other existing technologies still have significant shortcomings in the technology of simulating the emotions of electronic pets:
[0005] Shortcoming 1: Existing electronic pets typically have 4-8 emotional states, and the state transition logic is fixed and predictable. Users can fully grasp their reaction patterns after a short interaction, resulting in a strong sense of mechanicalness and toy-likeness. The emotional model of electronic pets is too simple and the reactions are mechanical, which seriously reduces the immersion and fun of electronic pet interaction.
[0006] Second shortcoming: The reaction mode of existing electronic pets is usually passive response, which cannot simulate the emotions of living organisms. They are not only affected by external stimuli, but also driven by internal forces such as hunger, fatigue, and social needs. This makes the emotions of electronic pets completely still or simply zero when there is no interaction, and they do not have the spontaneous fluctuations and mood continuity of living organisms.
[0007] Third shortcoming: Electronic pets of the same model or type have completely identical emotional response patterns. Users cannot create a unique electronic pet with a distinctive personality based on their own preferences or the history of their interaction with the pet, which limits the product's value as a personalized emotional companion.
[0008] This design scheme constructs a computational model inspired by the psychological and emotional dimensions, enabling electronic pets to autonomously and continuously generate rich and unique emotional states based on real-time interactive information from multiple sensors, combined with their internal needs and personality. These emotions are then mapped onto anthropomorphic multimodal behaviors, which not only greatly enhances the realism and immersion of the electronic pet's interaction but also makes the electronic pet more like a real living being. Summary of the Invention
[0009] This invention discloses an electronic pet emotion simulation system and method based on multi-sensor feedback, the main purpose of which is to overcome the above-mentioned shortcomings and disadvantages of the existing technology.
[0010] The technical solution adopted in this invention is as follows:
[0011] An electronic pet emotion simulation system based on multi-sensor feedback is disclosed. The simulation system includes: a multimodal perception interface layer, a core emotion generation engine, a personality parameter configuration module, and a behavior expression mapping module. The multimodal perception interface layer is used to receive and preprocess raw signals from various sensors on the electronic pet in real time and output standardized emotional event vectors. The core emotion generation engine is a computational module based on a nonlinear dynamic system and an internal driving model. It internally maintains an internal driving state vector and a core emotion state vector, and dynamically evolves the core emotion state vector according to the emotional event vector, the personality parameter matrix, and random noise. The personality parameter configuration module stores a set of adjustable parameters to construct the parameters in the personality parameter matrices A, B, and C and the natural decay function to simulate different electronic pet personalities. The behavior expression mapping module is used to convert the core emotion state vector into output behavioral commands for the electronic pet.
[0012] Furthermore, the plurality of sensors includes at least two of the following: a visual sensor, an auditory sensor, a tactile sensor, an inertial measurement unit, and an environmental sensor.
[0013] Furthermore, the natural decay function causes the energy value in the intrinsic drive state variables to decay linearly over time, while comfort and social satisfaction decay exponentially over time, and the decay rate is controlled by the personality parameters in the personality parameter configuration module.
[0014] Furthermore, the behavior expression mapping module includes a configurable mapping rule library, which maps the core emotional state vector to different regions in the multidimensional emotional space as different behavioral combinations, and the mapping rule library can be replaced or updated according to the hardware form of the electronic pet.
[0015] Furthermore, the parameters in the personalized parameter configuration module can be learned and adjusted online based on the user's interaction history, enabling the electronic pet's personality to exhibit adaptive changes.
[0016] A method for simulating the emotions of electronic pets based on multi-sensor feedback, the simulation method comprising the following specific steps:
[0017] Step 1: Receive raw signals from various sensors on the electronic pet in real time through the multimodal perception interface layer, and preprocess the raw signals to quantize them into a standardized emotional event vector E(t).
[0018] Step 2: Construct a core emotion generation engine based on a nonlinear dynamic system and an intrinsic driving model. The core emotion generation engine internally maintains: an intrinsic driving force state variable D(t) and a core emotion state vector M(t). The intrinsic driving force state variable includes energy value, comfort level and social satisfaction. The core emotion state vector is used to represent the instantaneous emotions of the electronic pet in a continuous multidimensional emotional space.
[0019] Step 3: Dynamically update the internal drive state variable D(t) according to the differential equation dD / dt=f(D)+g(E), where f(D) is the natural decay function and g(E) is the influence function of the emotional event vector on the internal drive;
[0020] Step 4: Based on the updated intrinsic motivation state variable D(t), the current core emotion state vector M(t), the emotional event vector E(t), and the random walk noise term N(t), the core emotion evolution equation is used.
[0021] dM / dt=AM(t)+BD(t)+CE(t)+N(t)
[0022] Calculate the core emotional state vector M(t+Δt) for the next moment, where A, B, C, and D are personality parameter matrices, representing emotional inertia, drive weight, and sensitivity to external stimuli, respectively.
[0023] Step 5: Through the behavior expression mapping module, the updated core situation state vector M(t+Δt) is mapped to one or more output behaviors of the electronic pet. The output behaviors include the electronic pet's body movements, synthesized sounds, and displayed facial expressions.
[0024] Furthermore, the continuous multidimensional emotional space in step two is the PAD three-dimensional emotional space, where P represents pleasure, A represents agent activation, D represents agent dominance, and the core emotional state vector M(t) is a point in the PAD three-dimensional emotional space.
[0025] Furthermore, the preprocessing of the original signal in step one includes filtering, feature extraction, and normalization, so that sensor signals from different sources are quantized into emotional event vector components with a uniform numerical range.
[0026] Furthermore, the random walk noise term N(t) in step four follows a Gaussian distribution, and its variance is controlled by the emotion stability parameter in the personality parameter configuration module.
[0027] Furthermore, the simulation method also includes step six: repeating steps one to five, so that the emotional state of the electronic pet continues to evolve under the combined effect of internal driving force and external interaction, forming a closed-loop interaction.
[0028] As can be seen from the above description and explanation of the present invention, compared with the prior art, the advantages of the present invention are as follows:
[0029] Advantage 1: This invention simulates the continuity, inertia, spontaneous fluctuations, and unpredictability of electronic pet emotions through a nonlinear dynamic system based on differential equations. This makes the electronic pet's reactions far exceed the existing single reaction rules, greatly enhancing the realism and immersion of electronic pet interaction.
[0030] Advantage Two: This invention employs a multi-dimensional continuous emotional space, enabling electronic pets to express different emotional states, which are then displayed through a behavior mapping module, resulting in expressiveness far exceeding existing technologies. Simultaneously, through a personalized parameter configuration module, electronic pets possess unique personalities and growth paths. This diversity establishes an emotional connection with the user, solving the problem of existing electronic pets being all the same and enhancing the product's value.
[0031] Advantage 3: By introducing intrinsic motivation variables, this invention enables electronic pets to have needs and motivations, allowing them to actively seek interaction, thus changing the traditional electronic pet's passive recognition and response. Attached Figure Description
[0032] Figure 1 This is a schematic diagram of the system structure of the present invention.
[0033] The system consists of: a multimodal perception interface layer 1, a core emotion generation engine 2, a personality parameter configuration module 3, and a behavior expression mapping module 4. Detailed Implementation
[0034] The specific embodiments of the present invention will be further described and illustrated below with reference to the accompanying drawings.
[0035] like Figure 1As shown, an electronic pet emotion simulation system based on multi-sensor feedback is disclosed. The simulation system includes: a multimodal perception interface layer 1, a core emotion generation engine 2, a personality parameter configuration module 3, and a behavior expression mapping module 4. The multimodal perception interface layer 1 is used to receive and preprocess raw signals from various sensors mounted on the electronic pet in real time, and output standardized emotional event vectors. In this embodiment, the various sensors of the multimodal perception interface layer 1 include at least two of the following: visual sensors, auditory sensors, tactile sensors, inertial measurement units, and environmental sensors. The multimodal perception interface layer 1 polls the above-mentioned sensors at a fixed sampling frequency, filters, extracts features, normalizes, and packages the raw signals into standardized emotional event vectors, which are then sent to the core emotion generation engine 2.
[0036] The core emotion generation engine 2 is a computational module based on a nonlinear dynamic system and an internal driving model. Internally, it maintains an internal driving state vector and a core emotion state vector, and dynamically evolves the core emotion state vector based on the emotional event vector, the personality parameter matrix, and random noise. Simultaneously, the core emotion generation engine 2 updates in real time according to the differential equation model proposed by the emotion simulation method, performing an iterative calculation periodically.
[0037] The personality parameter configuration module 3 stores a set of adjustable parameters used to construct the personality parameter matrices A, B, and C, as well as the parameters in the natural decay function, to simulate different electronic pet personalities. The natural decay function causes the energy value in the drive state variables to decay linearly over time, while comfort and social satisfaction decay exponentially over time, with the decay rate controlled by the personality parameters in the personality parameter configuration module. Furthermore, the parameters in the personality parameter configuration module can be learned and adjusted online based on the user's interaction history, allowing the electronic pet's personality to exhibit adaptive changes.
[0038] The behavior expression mapping module 4 is used to convert the core emotional state vector into output behavior commands for the electronic pet. The behavior expression mapping module includes a configurable mapping rule library, which maps the core emotional state vector to different regions in the multidimensional emotional space as different behavior combinations. The mapping rule library can be replaced or updated according to the hardware form of the electronic pet.
[0039] A method for simulating the emotions of electronic pets based on multi-sensor feedback, the simulation method comprising the following specific steps:
[0040] Step 1: Receive raw signals from various sensors on the electronic pet in real time through the multimodal perception interface layer, and preprocess the raw signals to quantize them into a standardized emotional event vector E(t). The preprocessing of the raw signals includes filtering, feature extraction, and normalization, so that the sensor signals from different sources are quantized into emotional event vector components with a uniform numerical range.
[0041] Step 2: Construct a core emotion generation engine based on a nonlinear dynamic system and an intrinsic driving model. The core emotion generation engine internally maintains an intrinsic driving state variable D(t) and a core emotion state vector M(t). The intrinsic driving state variable includes energy value, comfort level, and social satisfaction. The core emotion state vector is used to represent the instantaneous emotion of the electronic pet in a continuous multidimensional emotional space. This continuous multidimensional emotional space is the PAD three-dimensional emotional space, where P represents pleasure, A represents agent activation, and D represents agent dominance. The core emotion state vector M(t) is a point in the PAD three-dimensional emotional space.
[0042] Step 3: Dynamically update the internal drive state variable D(t) according to the differential equation dD / dt=f(D)+g(E), where f(D) is the natural decay function and g(E) is the influence function of the emotional event vector on the internal drive;
[0043] Step 4: Based on the updated intrinsic motivation state variable D(t), the current core emotion state vector M(t), the emotional event vector E(t), and the random walk noise term N(t), the core emotion evolution equation is used.
[0044] dM / dt=AM(t)+BD(t)+CE(t)+N(t)
[0045] Calculate the core emotional state vector M(t+Δt) for the next moment, where A, B, and C are personality parameter matrices, representing emotional inertia, drive weight, and sensitivity to external stimuli, respectively; the random walk noise term N(t) follows a Gaussian distribution, and its variance is controlled by the emotional stability parameter in the personality parameter configuration module.
[0046] Step 5: Through the behavior expression mapping module, the updated core situation state vector M(t+Δt) is mapped to one or more output behaviors of the electronic pet. The output behaviors include the electronic pet's body movements, synthesized sounds, and displayed facial expressions.
[0047] Step Six: Repeat steps one through five to allow the electronic pet's emotional state to continuously evolve under the combined influence of internal driving forces and external interactions, forming a closed-loop interaction.
[0048] This invention simulates the continuity, inertia, spontaneous fluctuations, and unpredictability of electronic pet emotions using a nonlinear dynamic system based on differential equations. This allows the electronic pet's reactions to far exceed existing single-response rules, greatly enhancing the realism and immersion of electronic pet interaction. Simultaneously, this invention employs a multi-dimensional continuous emotion space, enabling the electronic pet to express different emotional states, which are then displayed through a behavior mapping module, resulting in expressiveness far surpassing existing technologies. Furthermore, through a personalized parameter configuration module, the electronic pet possesses a unique personality and growth path. This diversity establishes an emotional connection with the user, solving the problem of existing electronic pets being all the same and enhancing the product's value.
[0049] The above are merely specific embodiments of the present invention, but the design concept of the present invention is not limited thereto. Any non-substantial improvements made to the present invention using this concept should be considered as infringing upon the protection scope of the present invention.
Claims
1. An electronic pet emotion simulation system based on multi-sensor feedback, characterized in that: The simulation system includes a multimodal perception interface layer, a core emotion generation engine, a personality parameter configuration module, and a behavior expression mapping module. The multimodal perception interface layer receives and preprocesses raw signals from various sensors mounted on the electronic pet in real time and outputs standardized emotional event vectors. The core emotion generation engine is a computational module based on a nonlinear dynamic system and an internal driving model. It internally maintains an internal driving state vector and a core emotion state vector, and dynamically evolves the core emotion state vector according to the emotional event vector, the personality parameter matrix, and random noise. The personality parameter configuration module stores a set of adjustable parameters to form the parameters in the personality parameter matrices A, B, and C, as well as the natural decay function, to simulate different electronic pet personalities. The behavior expression mapping module converts the core emotion state vector into output behavioral commands for the electronic pet.
2. The electronic pet emotion simulation system based on multi-sensor feedback according to claim 1, characterized in that: The plurality of sensors includes at least two of the following: a visual sensor, an auditory sensor, a tactile sensor, an inertial measurement unit, and an environmental sensor.
3. The electronic pet emotion simulation system based on multi-sensor feedback according to claim 1, characterized in that: The natural decay function causes the energy value in the drive state variables to decay linearly over time, while comfort and social satisfaction decay exponentially over time, and the decay rate is controlled by the personality parameters in the personality parameter configuration module.
4. The electronic pet emotion simulation system based on multi-sensor feedback according to claim 1, characterized in that: The behavior expression mapping module includes a configurable mapping rule library, which maps the core emotional state vector to different regions in the multidimensional emotional space as different behavioral combinations, and the mapping rule library can be replaced or updated according to the hardware form of the electronic pet.
5. The electronic pet emotion simulation system based on multi-sensor feedback according to claim 1, characterized in that: The parameters in the personalized parameter configuration module can be learned and adjusted online based on the user's interaction history, allowing the electronic pet's personality to undergo adaptive changes.
6. A method for simulating the emotions of electronic pets based on multi-sensor feedback, characterized in that: The simulation method includes the following specific steps: Step 1: Receive raw signals from various sensors on the electronic pet in real time through the multimodal perception interface layer, and preprocess the raw signals to quantize them into a standardized emotional event vector E(t). Step 2: Construct a core emotion generation engine based on a nonlinear dynamic system and an intrinsic driving model. The core emotion generation engine internally maintains: an intrinsic driving force state variable D(t) and a core emotion state vector M(t). The intrinsic driving force state variable includes energy value, comfort level and social satisfaction. The core emotion state vector is used to represent the instantaneous emotions of the electronic pet in a continuous multidimensional emotional space. Step 3: Dynamically update the internal drive state variable D(t) according to the differential equation dD / dt=f(D)+g(E), where f(D) is the natural decay function and g(E) is the influence function of the emotional event vector on the internal drive; Step 4: Based on the updated intrinsic motivation state variable D(t), the current core emotion state vector M(t), the emotional event vector E(t), and the random walk noise term N(t), the core emotion evolution equation is used. dM / dt=AM(t)+BD(t)+CE(t)+N(t) Calculate the core emotional state vector M(t+Δt) for the next moment, where A, B, C are personality parameter matrices, representing emotional inertia, drive weight, and sensitivity to external stimuli, respectively. Step 5: Through the behavior expression mapping module, the updated core situation state vector M(t+Δt) is mapped to one or more output behaviors of the electronic pet. The output behaviors include the electronic pet's body movements, synthesized sounds, and displayed facial expressions.
7. The method for simulating the emotions of electronic pets based on multi-sensor feedback according to claim 6, characterized in that: The continuous multidimensional emotional space in step two is the PAD three-dimensional emotional space, where P represents pleasure, A represents agent activation, and D represents agent dominance. The core emotional state vector M(t) is a point in the PAD three-dimensional emotional space.
8. The method for simulating the emotions of electronic pets based on multi-sensor feedback according to claim 6, characterized in that: The preprocessing of the original signal in step one includes filtering, feature extraction, and normalization, so that sensor signals from different sources are quantized into emotional event vector components with a uniform numerical range.
9. The method for simulating the emotions of electronic pets based on multi-sensor feedback according to claim 6, characterized in that: The random walk noise term N(t) in step four follows a Gaussian distribution, and its variance is controlled by the emotion stability parameter in the personality parameter configuration module.
10. The method for simulating the emotions of electronic pets based on multi-sensor feedback according to claim 6, characterized in that: The simulation method further includes step six: repeating steps one to five, so that the emotional state of the electronic pet continues to evolve under the combined effect of internal driving force and external interaction, forming a closed-loop interaction.