Bionic intelligent pet robot with multi-mode emotional interaction capability

Abstract

The invention discloses a bionic intelligent pet robot with a multi-mode emotion interaction capability. A carbon fiber composite skeleton-medical silica gel skin-bionic hair three-layer bionic structure is adopted, and a 16-degree-of-freedom bionic movement system and a terminal cloud collaborative emotion calculation framework are matched, so that the intelligent partner with real touch feeling, flexible posture and deep emotion interaction capability is realized. The system mainly comprises a carbon fiber composite skeleton having 16 degrees of freedom and a three-section flexible spine; the medical silica gel constant-temperature skin is connected through split type magnetic attraction, and the constant temperature of 36-38 DEG C is achieved; the modified acrylic fiber bionic hair is implanted by adopting gradient density; the multi-mode perception system realizes full coverage of visual sense, auditory sense and tactile sense; the emotion calculation engine is used for realizing emotion interaction based on a state machine and personalized memory; and a cat nest type automatic charging seat. The technical defects that an existing pet robot is cold in touch feeling, stiff in movement and shallow in interaction are overcome, and the intelligent pet robot is applied to family emotion accompanying, old people health care and children education auxiliary scenes.

Description

Technical Field

[0001] This invention belongs to the field of service robot technology, specifically relating to an intelligent pet robot that integrates a biomimetic mechanical structure, a multimodal perception system, and an emotion computing model. Background Technology

[0002] With an aging population, a growing number of people living alone, and smaller family structures, the demand for emotional companionship is becoming increasingly prominent. Existing pet robots or companion devices mainly suffer from the following shortcomings: 1. Insufficient tactile simulation: The outer shell is mostly made of hard plastic or simple plush fabric, lacking the temperature, softness and delicate fur texture of a real living organism, making it difficult to establish a deep emotional connection. 2. Limited mobility: The joints generally have fewer degrees of freedom (usually ≤8), resulting in stiff and mechanical movement trajectories that cannot simulate the agile posture and excellent balance of felines. 3. Limited interaction dimensions: Most only support simple preset voice command responses, lack the ability to integrate and understand multimodal information such as vision, touch, and hearing, and do not possess personalized memory and emotional evolution capabilities. 4. Fragmentation of Functional Scenarios: Product design often focuses on the single attributes of entertainment toys or tools, failing to organically unify "emotional companionship" with "practical functions". 5. Maintenance burden exists: Some products require manual cleaning of simulated excrement or frequent replacement of consumables, which increases the complexity of use and violates the original design intention of "burden-free companionship". Therefore, the market urgently needs an intelligent pet robot that can highly simulate biological features in appearance, touch, motion intelligence, and interaction depth, and integrate practical auxiliary functions, so as to be suitable for scenarios such as family emotional companionship, elderly health care, and children's education assistance. Summary of the Invention

[0003] This invention aims to overcome the shortcomings of existing technologies and provide a biomimetic intelligent pet robot. Through a three-layer biomimetic structure of "skeleton-skin-fur," a high-degree-of-freedom biomimetic motion system, and a multimodal perception and edge-cloud collaborative emotional computing architecture, it achieves an intelligent companion with realistic biological touch, agile movement, and deep personalized emotional interaction capabilities. To achieve the above objectives, the technical solution of the present invention is as follows: A biomimetic intelligent pet robot, characterized in that it comprises: A. A carbon fiber composite skeleton, consisting of carbon fiber tubes, 3D printed nylon connectors and titanium alloy joint shafts, has 16 degrees of freedom, including 2 degrees of freedom (DOF) for the head, 1 DOF for the neck, 2 DOF for the spine, 12 DOF for the limbs and 3 DOF for the tail. The spine is connected in three flexible segments and a silicone breathing bag is built into the chest cavity. B. Medical silicone skin, covering the outside of the skeleton, made of platinum vulcanized silicone material, 2-5mm thick, with an embedded nylon mesh reinforcement layer, adopting a split design and connected by magnetic buckles, and with a built-in PTC heating film to achieve a constant temperature of 36-38℃; C. A biomimetic hair system, which uses modified acrylic fiber material to be fixed to the silicone skin in a cluster implantation process, with hair length gradient distribution of 10-40mm and density of 60,000-120,000 hairs / cm², and the surface has a nano hydrophobic coating and silver ion antibacterial treatment. D. Multimodal sensing system, including an RGB-D depth camera located on the head, a microphone array located on the top of the head, and distributed tactile sensors distributed on the back / abdomen / head / ears / beard / sole of the feet; E. The Emotional Computing Engine runs on a heterogeneous computing architecture, integrates visual, auditory, and tactile multimodal data, and achieves personalized emotional interaction through a lightweight large language model and reinforcement learning algorithms; F. The cat-shaped charging dock features a biomimetic design and incorporates a magnetic positioning mechanism, an infrared guidance system, and a Qi standard wireless charging module, supporting automatic recharging of the robot. Furthermore, the 16-DOF motion system adopts a hierarchical control strategy: the core joints (hip / shoulder / knee) use high-torque servo motors with torque ≥4Nm, accuracy ≤0.1°, and full-state feedback; the secondary joints (wrist / ankle / tail root) use medium-torque servo motors with torque ≥1.5Nm; and the micro joints (ear / whisker / tail tip) use 9g-class metal gear servos. The motion control is based on the ZMP (zero torque point) dynamic balance algorithm and Bézier curve trajectory planning, with a control frequency of 1000Hz. Furthermore, the medical silicone skin comprises four separate components: a head cover, a tail cover, a torso cover, and limb covers, with Z-shaped overlapping seams. The magnetic clasp includes a neodymium iron boron magnet (Φ10mm) embedded in the skeleton and a galvanized steel sheet embedded inside the skin, with a single-point attraction force of 5N and a total holding force of 180N. PTC heating films are distributed on the back, abdomen, and inner earlobes, working in conjunction with NTC thermistors to achieve closed-loop temperature control with an accuracy of ±0.5℃, and featuring dual-path hardware overheat protection. The skin has a multi-layered composite structure, comprising, from the inside out: a nylon reinforcing fabric layer integrated with the skeleton, a waterproof film layer, a PTC heating film layer, an elastic silicone layer, and an outermost biomimetic hair layer. Furthermore, the hair length exhibits a gradient distribution, generally within the range of 10-40 mm. The head has shorter hair, with a smaller value (10-15 mm), while the tail and body hair are relatively longer, with a larger value (30-40 mm), and the limb hair is of medium length (15-30 mm). Furthermore, the emotion computing engine includes: a multimodal input fusion layer that uses a Transformer network to fuse four-dimensional data of vision, hearing, touch, and environment; an emotion state judgment layer based on a finite state machine (FSM) that defines four core states: happy, sleepy, curious, and attached; a personalized memory layer supported by a vector database that stores the user's face, voiceprint, preferences, and interaction history; and a behavior output layer that generates motion commands, voice output, facial expression display, and body temperature regulation signals. Furthermore, the cat-shaped charging base includes: a biomimetic cat-shaped structure with dimensions of 350×300×150mm and a bottom weight of 2kg; a Qi standard wireless charging transmitter module with a power of 15W and a coil diameter of 50mm; a magnetic positioning mechanism, including four magnets with alternating polarities on the bottom of the robot and corresponding magnetic steel plates and a central positioning cone inside the charging base; an infrared guidance system, including a 940nm infrared emitting tube and an infrared receiving tube on the robot's head, with an effective guidance distance of 2 meters; and an intelligent control circuit that supports FOD (Foreign Object Detection), charging status indication, and automatic disconnection when fully charged. Furthermore, the robot has educational assistance functions: it has a built-in K12 synchronous curriculum content library covering Chinese, mathematics, and English subjects; it integrates an English oral dialogue training system that supports pronunciation correction and progress tracking; it provides a visual programming interface, allowing users to control the robot's actions by dragging and dropping instruction blocks; it has a learning time management and focus guidance mechanism, and combines an affective computing engine to achieve motivational teaching. Furthermore, the robot has elderly care functions: it integrates a voice emotion analysis module to identify depression or anxiety based on acoustic features such as speech rate and tone; it has a fall detection algorithm that combines foot pressure change signals with visual posture estimation for judgment; it has an emergency call function that automatically contacts a preset emergency contact after detecting an abnormality; and it supports medication reminders and health data recording, which can be used for long-term health trend analysis. The core point of this invention is: 1. Bionic Touch System: The system features a pioneering three-layer composite structure consisting of a carbon fiber composite skeleton, medical-grade platinum vulcanized silicone skin, and modified acrylic bionic hair. Combined with active thermoregulation technology, it achieves a high degree of unity between biological-grade soft touch, warm body temperature, and realistic appearance. 2. Flexible motion architecture: It adopts a 16-DOF bionic joint configuration, combined with a three-segment flexible spine and an active tail that can act as a "fifth balancing foot", and applies a ZMP-based dynamic balance algorithm to realize complex and realistic movements such as walking, jogging, jumping, and playful rolling. 3. Multimodal emotional interaction: Deeply integrating visual, auditory, tactile, and temperature perception, an emotional model and personalized memory system based on finite state machines and reinforcement learning are constructed, enabling the robot to understand the user's emotions and make human-like responses, achieving an evolutionary companionship of "the more you interact, the more it understands you". 4. Seamless Energy and Scenario-Based Function Integration: Automatic recharging is achieved through a magnetic cat bed charging base, seamlessly integrating with the home environment; at the same time, the core of emotional companionship is deeply integrated with practical functions such as K12 education and elderly health monitoring, upgrading it from a single toy to a smart home assistant. [Beneficial Effects] Compared with the prior art, the present invention has the following significant advantages: 1. Bio-level simulation experience: The composite biomimetic structure combined with dynamic constant temperature technology highly simulates real pets in terms of touch, temperature and visual appearance, which can significantly shorten the time for users to establish an emotional connection. 2. Extremely agile motion performance: The combination of highly free bionic joints and advanced motion control algorithms enables the robot to move with a smoothness and naturalness far exceeding that of traditional products, and to perform a variety of complex behaviors. 3. Deeply Personalized Emotional Interaction: Multimodal fusion perception and emotion computing engine enable robots to understand and respond to user emotions, forming a long-term and unique bond with personalized memory, greatly enhancing user stickiness. 4. Practical Function Integration: It seamlessly integrates educational assistance and elderly care functions, expands application scenarios, and realizes the value unity of "emotional companionship" and "life assistance". 5. Truly zero-burden maintenance: No feeding or cleaning required, energy is replenished through automatic recharging, allowing users to purely enjoy the pleasure of companionship and eliminating all maintenance hassles. 6. Safe and reliable: Made entirely of medical-grade safety materials, free of sharp objects such as teeth and claws, equipped with multiple safety protection mechanisms, and has passed relevant safety certifications, making it suitable for the whole family, including children and the elderly. Attached Figure Description Figure 1 This is a schematic diagram of the overall structure of one embodiment of the present invention. Figure 2 This is a data flow diagram of the multimodal perception computing engine of the present invention. Figure 3 This is a cross-sectional schematic diagram of the biomimetic skin layered structure of the present invention. Figure 4 This is a schematic diagram of the joint structure of the carbon fiber composite skeleton of the present invention. Figure 5 This is a distribution diagram of the biomimetic animal structure and perception of the present invention. Figure 6 This is a schematic diagram of the driving mechanism of the upper and lower limbs in this invention. In the image: 001 Tail, 002 Body, 003 Head, 004 Eyes, 005 Ears, 006 Nose, 007 Mouth, 008 Neck, 009 Claws, 010 Forelimbs, 011 Lower Legs, 012 Thighs, 013 Hindlimbs; 201 Left and Right Cameras, 202 Tail / Ear / Beard Pressure Sensors, 203 Microphone Array, 204 Back Tactile Matrix, 205 Foot Pressure Sensors, 206 Dual-Core Processor, 207 Gumstix Embedded Computing Module, 208 Inertial Measurement Unit (IMU), 209 Field Programmable Gate Array (FPGA), 210 Execution Output Mechanism; 301 Bionic Hair Layer, 302 Elastic Silicone Layer, 303 Waterproof Membrane Layer, 304 PTC heating film layer, 305 nylon reinforced fabric, 306 carbon fiber skeleton, 307 device placement space; 401 tailbone, 402 body skeleton, 403 skull, 408 head and neck bones, 409 claw joint, 410 forelimb joint, 411 tibia, 412 femur, 413 hindlimb joint, 414 upper forelimb joint, 415 lower forelimb joint, 416 upper hindlimb joint, 417 lower hindlimb joint; a. animal structure, b. intelligent pet cat, 501 light sensor, 502 infrared receiver, 503 sound sensor, 504 ear touch sensor. 505 Sound generator, 506 Back tactile sensor, 507 Airbag, 508 Tail sensor, 509 External interface, 510 Battery pack, 511 Foot sensor, 512 Joint torque sensor, 513 Counterweight, 514 Controller, 515 Communication module, 516 Temperature and humidity sensor, 517 Smoke sensor, 518 Beard sensor; 601 Servo motor, 602 Servo motor rocker arm, 603 Ball joint, 604 Lower joint of link, 605 Knee joint, 606 Lower leg support, 607 Thigh support, 608 Upper joint of link, 609 Hip joint. Detailed Implementation To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only for explaining this invention and are not intended to limit the scope of protection of this invention. like Figure 1 As shown, the bionic intelligent pet robot of the present invention is designed in the shape of a feline animal and can also be called an intelligent pet cat. It mainly includes a head (003), neck (008), body (002), tail (001), forelimbs (010), and hindlimbs (013). The head integrates bionic eyes (004), ears (005), nose (006), and mouth (007). The limbs are similar in structure to real felines. The forelimbs (010) and hindlimbs (013) each include a thigh (012), a lower leg (011), and a paw (009). Figure 5The biomimetic animal structure and sensory distribution are demonstrated. The biomimetic structure (a) of the intelligent pet cat (b) integrates multiple sensors. The visual module includes: a main visual module located in the biomimetic eyes (004) on the head, using Intel RealSense D435i left and right cameras (201), supporting RGB (1920×1080@30fps, 86°×57°) and depth (1280×720@30fps, 0.3-10m) perception, and integrating a 6-axis IMU (200Hz); and an auxiliary visual module located in the center of the abdomen of the body (002), using an OV2640 wide-angle camera (120° field of view, MIPI interface). The auditory module includes: a 4-microphone ring array (203, ReSpeaker Mic Array v2.0, 40mm in diameter) located in the ear (005), supporting sound source localization (horizontal ±5° / vertical ±15°), far-field pickup distance of 5 meters, and signal-to-noise ratio >60dB; the sound generator (505) is a 2-inch full-range unit (8Ω / 10W) located on both sides of the chest cavity, driven by a Class D amplifier. The tactile sensing system includes sensors distributed in the ears (504), back (506), tail (508), whiskers (518), soles of feet (511), and joints (512). In addition, environmental sensors such as temperature and humidity sensors (516) and smoke sensors (517) are integrated. The infrared receiver (502) is used for charging guidance, the airbag (507) simulates breathing, the external interface (509) is used for debugging, the battery pack (510) provides power, and the counterweight (513) is used to maintain balance. The specifications and functions of each tactile sensor are shown in the table below: The core system of this invention is composed as follows: 1. Bionic mechanical structure system like Figure 3 , 4 As shown, the robot's mechanical structure adopts a three-layer biomimetic design. Carbon fiber composite skeleton (306, 402): As the core load-bearing and motion frame, it is constructed using T700 grade carbon fiber tubing (306) and 3D-printed nylon connectors, with TC4 titanium alloy used for the joint axes. Figure 4 and Figure 5 As shown, the skeleton mimics a biological skeleton, including the skull (403), cervical bones (408), body skeleton (402), coccyx (401), femur (412), tibia (411), etc., and is connected by various joints (409, 410, 413, 414, 415, 416, 417) to achieve 16 degrees of freedom of movement. Silicone skin: Made of medical-grade platinum-cured silicone material, with a Shore hardness of A12, tensile strength >8MPa, and tear strength >20kN / m. The main body area is 3mm thick, the joint movement area is 2mm thick, and the pressure-bearing area is 5mm thick. It features an embedded nylon mesh reinforcement layer and a split design connected by magnetic snap-fit ​​fasteners. Figure 3 As shown, the skin has a multi-layered composite structure, including, from the inside out: a nylon reinforcing fabric (305) layer combined with the skeleton (306); a PTC heating film layer (304) covering the back (150x100mm), abdomen (80x60mm) and inner ear (20x15mm×2), which, together with the NTC thermistor, achieves closed-loop constant temperature control of 36-38℃ (accuracy ±0.5℃) and has dual-path hardware overheat protection; a waterproof film layer (303); an elastic silicone layer (302) as the main body; and the outermost bionic hair layer (301). Biomimetic hair: Made of high-end modified acrylic fiber (similar to mink cashmere), with a fineness of 3.3 dtex, a crimp grade of 3, and a density of approximately 80,000 hairs / cm². The hair length is gradient-distributed: approximately 10 mm on the head, approximately 36 mm on the tail and torso, and approximately 15 mm on the limbs. The surface is treated with nano-hydrophobic and silver ion antibacterial coatings. Motion drive unit: such as Figure 6 As shown, the movement of the limbs is driven by a servo motor (601). The servo motor (601) transmits the motion to the upper joint (608) and lower joint (604) of the connecting rod through the servo motor rocker arm (602) and the ball joint (603), thereby driving the thigh support (607) and the lower leg support (606) to move around the hip joint (609) and knee joint (605) to achieve bionic walking. 2. Multimodal sensing and computing systems like Figure 2 As shown, the robot's perception and computing system is a complete data processing closed loop. Sensory input: The system acquires visual information through the left and right cameras (201); collects audio through the microphone array (203); acquires pressure and balance information through the plantar pressure sensor (205); acquires touch information through the back tactile matrix (204) and the sensor matrix (202) distributed on the tail, ears and whiskers; and senses its own posture through the inertial measurement unit (IMU) (208). Computational Core: The aforementioned multimodal data is fed into a processing unit centered around a dual-core processor (206) and a Gumstix embedded computing module (207). A field-programmable gate array (FPGA) (209) is responsible for the preprocessing and fusion of the sensor data. The affective computing engine analyzes, understands, and makes decisions based on the fused information. Behavior output: The final calculation result is converted into specific control instructions, which drive the robot's movement, vocalization, facial expression changes, etc. through the execution output mechanism (210) to complete the interactive closed loop. 3. Affective computing and software architecture The affective computing engine is the "brain" of the robot. Its software architecture is built on ROS2 and adopts a layered design. Hardware abstraction layer: directly drives servo motors and collects sensor data. Perception processing layer: Runs algorithms such as object detection, speech recognition, and tactile pattern recognition to transform raw data into semantic information. Decision planning layer: This is the core of the emotion computing engine. It integrates multimodal information, combines personalized memory (stores user preferences and interaction history), runs an emotion model based on finite state machine (FSM) and reinforcement learning, determines the current emotion state (such as happy, curious, sleepy, attached), and generates corresponding behavioral instructions (such as approach, rub, play, rest). Edge-cloud collaboration: Local processing ensures privacy and real-time performance, while functions such as complex semantic understanding and knowledge updates are assisted by the cloud, and the local model is continuously optimized through federated learning. 4. Sensorless Energy System The robot automatically charges using a dedicated cat-shaped charging dock. When the battery level drops below 20%, the infrared receiver (502) on the robot's head detects the signal emitted by the infrared guide transmitter on the charging dock and autonomously navigates to the vicinity of the charging dock. Precise alignment is achieved through the attraction between the magnet on the bottom of the robot and the magnetic positioning steel plate and positioning cone inside the charging dock. After alignment, the charging control circuit is activated, and the Qi standard wireless charging coil (15W) begins charging the battery pack (510), which takes approximately 2-4 hours to fully charge. Once fully charged, the power is automatically cut off, and the robot leaves the charging dock to resume activity. Wired charging is also possible via an external interface (509). Example 1: Family Emotional Companionship Scenario In this embodiment, the robot primarily serves as an emotional companion for family members. When a user returns home, the robot identifies its owner through vision (201) and sound source localization (203) and rises from its cat-shaped charging dock to greet them. When the user strokes its back (tactile matrix 204), the PTC heating film (304) maintains the skin temperature at 37°C, the bionic fur (301) provides a soft touch, and the airbags (507) simulate breathing. The emotion computing engine determines that the user is in a relaxed state based on the interaction history (e.g., the user likes to pat their head) and the current touch intensity, and then drives the robot to perform behaviors such as "leg rubbing" (coordinated movements through joints 409, 410, etc.) and emitting pleasant purring sounds. In daily companionship, it can actively seek out the user, lie quietly beside the user (to maintain body temperature), or play autonomously by pouncing on toys, etc. The flexible joints of its spine and the swaying of its tail (001) make all movements extremely natural. When the battery is low, it automatically returns to the charging dock to recharge, without any human intervention. Example 2: Elderly Health Care Scenario In this embodiment, the robot focuses on health monitoring, providing companionship and conversation for elderly people living alone. An emotion computing engine continuously analyzes the elderly person's speech, identifying potential depression or anxiety through abnormal speech rate and tone, and offering reassurance through methods such as playful interaction, playing soothing music, or engaging in conversation. The robot's built-in fall detection algorithm integrates instantaneous data from plantar pressure sensors (205) and visual posture estimation for judgment. Once a fall is detected, the robot immediately inquires about the situation via voice. If no effective response is received or prolonged stillness is detected, an emergency protocol is triggered, automatically contacting a preset emergency contact via the 4G network and playing a distress message through the speaker. Simultaneously, the robot can periodically remind the elderly person to take medication via voice and record their daily activity levels, key emotional interactions, and other health data for their children to view remotely. Example 3: Children's Educational Assistance Scenarios In this embodiment, the robot serves as a learning partner and programming introduction tool for children. During learning sessions, the robot can act as a "little teacher," guiding children in English speaking practice through voice dialogue. The affective computing engine scores and corrects pronunciation, providing positive reinforcement with encouraging words. It has a built-in K-12 synchronized curriculum Q&A knowledge base to answer subject-related questions. Furthermore, by connecting to a companion visual programming app, children can program and control the robot to perform actions by dragging and dropping instruction blocks (such as "forward," "turn around," and "wag tail"). The robot vividly executes instructions and associates code logic with physical actions, enhancing the fun of learning. The affective computing engine manages learning time; when it detects a decline in a child's attention, it suggests short interactive games to achieve a balance between work and rest.

Claims

1. A biomimetic intelligent pet robot, characterized in that, include: A. A carbon fiber composite skeleton, consisting of carbon fiber tubes, 3D printed nylon connectors and titanium alloy joint shafts, has 16 degrees of freedom, including 2DOF for the head, 1DOF for the neck, 2DOF for the spine, 12DOF for the limbs and 3DOF for the tail. The spine is connected in three flexible segments and a silicone breathing bag is built into the chest cavity. B. Medical silicone skin, covering the outside of the skeleton, made of platinum vulcanized silicone material, 2-5mm thick, with an embedded nylon mesh reinforcement layer, adopting a split design and connected by magnetic buckles, and with a built-in PTC heating film to achieve a constant temperature of 36-38℃; C. A biomimetic hair system, which uses modified acrylic fiber material to be fixed to the silicone skin in a cluster implantation process, with hair length gradient distribution of 10-40mm and density of 60,000-120,000 hairs / cm², and the surface has a nano hydrophobic coating and silver ion antibacterial treatment. D. Multimodal sensing system, including an RGB-D depth camera located on the head, a microphone array located on the top of the head, and distributed tactile sensors distributed on the back / abdomen / head / ears / beard / sole of the feet; E. The Emotional Computing Engine runs on a heterogeneous computing architecture, integrates visual, auditory, and tactile multimodal data, and achieves personalized emotional interaction through a lightweight large language model and reinforcement learning algorithms; F. The cat-shaped charging dock features a biomimetic design and incorporates a magnetic positioning mechanism, an infrared guidance system, and a Qi standard wireless charging module, supporting automatic recharging of the robot.

2. The bionic intelligent pet robot according to claim 1, characterized in that, The 16-DOF motion system adopts a hierarchical control strategy: the core joints use high-torque servo motors with torque ≥4Nm, accuracy ≤0.1°, and full-state feedback; the secondary joints use medium-torque servo motors with torque ≥1.5Nm; the micro joints use 9g-class metal gear servos; the motion control is based on the ZMP dynamic balance algorithm and Bézier curve trajectory planning, with a control frequency of 1000Hz.

3. The biomimetic intelligent pet robot according to claim 1, characterized in that, The medical silicone skin comprises four separate components: a head cover, a tail cover, a torso cover, and limb covers, with Z-shaped overlapping structures at the seams. The magnetic buckle includes a neodymium iron boron magnet (Φ10mm) embedded in the skeleton and a galvanized steel sheet embedded in the inner side of the skin, with a single-point attraction force of 5N and a total holding force of 180N. PTC heating films are distributed on the back, abdomen, and inner ear, and work with NTC thermistors to achieve closed-loop temperature control with an accuracy of ±0.5℃, and have dual-path hardware overheat protection.

4. The biomimetic intelligent pet robot according to claim 1, characterized in that, The emotion computing engine includes: a multimodal input fusion layer that uses a Transformer network to fuse four-dimensional data of vision, hearing, touch, and environment; an emotion state judgment layer based on a finite state machine that defines four core states: happy, sleepy, curious, and attached; a personalized memory layer supported by a vector database that stores the user's face, voiceprint, preferences, and interaction history; and a behavior output layer that generates motion commands, voice output, facial expression display, and body temperature regulation signals.

5. The biomimetic intelligent pet robot according to claim 1, characterized in that, The cat-shaped charging base includes: a biomimetic cat-shaped structure with dimensions of 350×300×150mm and a bottom weight of 2kg; a Qi standard wireless charging transmitter module with a power of 15W and a coil diameter of 50mm; a magnetic positioning mechanism, including four magnets with alternating polarities on the bottom of the robot and corresponding magnetic steel plates and a central positioning cone inside the charging base; an infrared guidance system, including a 940nm infrared transmitter and an infrared receiver on the robot's head, with an effective guidance distance of 2 meters; and an intelligent control circuit that supports FOD (Foreign Object Detection), charging status indication, and automatic disconnection when fully charged.

6. The biomimetic intelligent pet robot according to any one of claims 1 to 5, characterized in that, The robot has educational assistance functions: it has a built-in K12 synchronous curriculum content library covering Chinese, mathematics, and English subjects; it integrates an English oral dialogue training system that supports pronunciation correction and progress tracking; it provides a visual programming interface, allowing users to control the robot's actions by dragging and dropping instruction blocks; it has a learning time management and focus guidance mechanism, and combines an affective computing engine to achieve incentive-based teaching.

7. The biomimetic intelligent pet robot according to any one of claims 1 to 5, characterized in that, The robot has elderly care functions: it integrates a voice emotion analysis module to identify depression or anxiety based on acoustic features such as speech rate and tone; it has a fall detection algorithm that combines sudden changes in plantar pressure signals with visual posture estimation for judgment; and it has an emergency call function that automatically contacts a preset emergency contact person after detecting an abnormality. It supports medication reminders and health data recording for long-term health trend analysis.

Images