A biomimetic feeding system based on multimodal perception and its whole life cycle control method

Abstract

This invention discloses a biomimetic feeding system based on multimodal perception and its full lifecycle control method, belonging to the field of biomimetic robotics and intelligent sensing technology. The system includes a biomimetic chewing module, a multimodal taste perception module, an intelligent swallowing anti-backflow module, and a main control safety unit. It achieves low-power automatic sampling of complex food chyme through biomimetic microfluidic channels relying on capillary principles; uses glutaraldehyde cross-linked modified casein hydrogel to improve sensor lifespan; and incorporates a mechanically self-locking one-way valve for backflow prevention. The control method achieves accurate food identification and safe adaptive control through multimodal weighted fusion of chewing mechanical data and taste chemical data. This invention solves the problems of poor anti-interference, short lifespan, and insufficient safety in existing equipment, and can be widely applied in fields such as biomimetic robotics, medical rehabilitation, and food testing.

Description

Technical Field

[0001] This invention relates to the fields of bionic robots, intelligent sensing and flexible electronics, specifically to a digestive tract front-end system that integrates bionic chewing, multimodal taste recognition, intelligent swallowing and safety protection, and particularly to its full life cycle control method based on long-life materials and adaptive algorithms. Background Technology

[0002] Existing bionic feeding devices or taste analysis instruments are often functionally fragmented and have incomplete technical systems, and have long suffered from three major pain points: 1. Severe disconnect between perception and action: Traditional chewing mechanisms and taste sensors work independently, failing to simulate the physiological closed-loop feedback mechanism of humans "chewing, grinding, and perceiving simultaneously." This results in extremely low recognition accuracy for complex chymes containing high fat, gas, and solid particles (such as spicy red oil base, bubble tea, and compound dishes), making it impossible to reproduce the true eating perception logic.

[0003] 2. Weak resistance to environmental interference: Conventional microfluidic detection structures are easily affected by oil adsorption, air bubble resistance, and blockage by viscous media. The sensor surface is easily contaminated and passivated, resulting in poor long-term stability and making it difficult to adapt to routine detection operations in real-world food scenarios.

[0004] 3. Fatal shortcomings in safety and service life: Existing equipment lacks emergency removal of hard objects and redundant protection mechanisms to prevent backflow during power outages, resulting in a high failure rate. At the same time, the sensing layer mostly uses ordinary short-term polymer materials, and the service life of conventional equipment is generally less than 1 year, resulting in high maintenance frequency and high costs, which cannot meet the requirements for long-term stable operation in commercial applications.

[0005] Furthermore, significant technological gaps also exist in the fields of medical rehabilitation, elderly care, and food industry quality inspection: existing swallowing function assessment equipment, taste diagnosis instruments, and food sensory testing equipment are mostly static detection modes. The equipment is bulky and has a simple detection logic, which cannot simulate the real physiological processes of dynamic chewing, real-time grinding, and dynamic sampling in the human oral cavity. This results in a large deviation between the detection, assessment, and training results and the actual human eating state. There is a lack of high-precision, integrated, safe, and long-life bionic eating detection and execution equipment. Summary of the Invention

[0006] This invention aims to construct a biomimetic feeding system with a fully closed-loop capability encompassing human-like perception, decision-making, execution, protection, and self-maintenance. Through a unique biomimetic microfluidic architecture, a bio-modified long-life material system, a multimodal data fusion algorithm, and a dual-safety redundancy structure design, it completely solves the industry pain points of existing technologies, such as low recognition accuracy, poor anti-interference ability, insufficient security, short lifespan, and weak scenario adaptability.

[0007] 1. System Hardware Architecture (Four-in-One Linked Closed Loop) (1) Bionic chewing module It adopts a biomimetic tooth enamel structure of zirconia-toughened alumina (ZrO2-ATZ) and is matched with a gear transmission group with a high-precision torque limiter to accurately simulate the flexible grinding movement of human chewing muscles. It can adapt to the grinding of soft and hard food, and at the same time collect chewing resistance, hardness and deformation data in real time, providing mechanical dimension parameter support for multimodal perception.

[0008] (2) Multimodal taste perception module Testing hardware: High-stability platinum-iridium alloy electrode array, corrosion-resistant, interference-resistant, and suitable for extreme acid and alkali, high and low temperature complex working conditions; Core functional coating: glutaraldehyde crosslinked modified casein hydrogel coating, which constructs a human-like biological sensing interface and has an ultra-long service life and biocompatibility. Sampling and delivery structure: Based on the principle of rush, a biomimetic microfluidic pipeline is designed. The inner wall of the pipeline integrates a nano-level hydrophilic coating and a precision micro-guide ridge structure. Relying on the principle of natural capillary low-power delivery, it can achieve stable micro-sampling of complex chyme with high viscosity, high oil content and gas content.

[0009] (3) Intelligent swallowing and anti-reflux module It adopts a double-lobe PTFE sealed one-way valve structure, with built-in nickel-titanium alloy memory spring and spiral return spring, and is equipped with a flexible micro-pipe network conveying channel to achieve adaptive flow guidance of chyme, high-pressure backflow prevention, and mechanical self-locking in the event of power failure, thus balancing smooth conveying and sealing performance under extreme working conditions.

[0010] (4) Main control and security unit It integrates multi-channel signal acquisition and processing circuits, fault diagnosis modules, and audible and visual early warning modules to achieve intelligent control of the entire process, including foreign object jamming identification, pressure overload protection, power failure self-locking protection, abnormal signal self-checking, and reverse output execution.

[0011] 2. Core Innovation Points and Quantitative Technical Indicators (1) Imperceptible anti-interference microfluidic technology based on the principle of rush pith Abandoning the traditional power pump sampling method, this product uses a pipeline-specific nano-level hydrophilic coating and a micro-guide ridge structure design to enhance the capillary effect. It achieves stable delivery of high-oil, high-viscosity, and gaseous chyme without additional power and with low energy consumption, reducing the interference rate of air bubble resistance and oil blockage by 90%, and completely solving the long-standing problem of sampling failure in the industry.

[0012] (2) Mechanical + electronic multi-level safety redundancy mechanism Mechanical redundancy protection: In the instant of power failure, the one-way valve’s built-in spiral spring can complete the full mechanical locking within 15ms, completely eliminating the risk of food backflow and leakage, with no power dependence and high reliability; Logic-based intelligent protection: When the system detects chewing resistance >120N or abnormal fluctuations in taste sensor signals, it triggers a reverse dispensing procedure + audible and visual alarm within 0.3s and completes overload shutdown within 0.08s, preventing equipment jamming, damage, and media leakage.

[0013] (3) Ultra-long lifespan biomodified material system The casein hydrogel structure is optimized through a precise cross-linking modification process using glutaraldehyde, significantly improving the material's corrosion resistance, temperature resistance, and aging resistance. Based on 5 years of accelerated aging test data, combined with dual extrapolation using the Arrhenius model and the Coffin-Manson model, the theoretical service life of the core sensing components under normal operating conditions can reach 20 years. The equipment can stably adapt to extreme chemical environments ranging from -20℃ to 100℃ rapid temperature shocks and pH 1.0 strong acids to pH 12.0 strong alkalis, demonstrating extremely strong adaptability to operating conditions. Detailed Implementation

[0014] Example 1: System Interaction and High Anti-interference Performance Test Actual testing was conducted using red oil hot pot base: the microfluidic pipeline automatically extracted 0.5mL of standardized food porridge sample based on the capillary drainage structure. Combined with the hydrophobic and self-cleaning properties of the material, the residual oil content of the food was <0.5mg / cm², and the taste recognition accuracy rate reached 97.5% under complex working conditions.

[0015] Swallowing anti-backflow test: Under continuous impact of 30kPa high-intensity reverse pressure, the PTFE double-disc check valve maintained zero leakage throughout the process, demonstrating excellent sealing stability.

[0016] Example 2: Extreme Working Conditions and Foreign Object Safety Protection Test The system uses 3mm hard steel balls to simulate foreign objects getting stuck in the teeth. It stops quickly in 0.08 seconds and simultaneously starts a reverse push program to automatically remove the foreign object without jamming or structural damage.

[0017] The equipment underwent 10 cycles of rapid alternation between high and low temperatures from -20℃ to 100℃, and the micro-pipe network structure remained intact with no sealing failures. The detection error of the taste sensor was stably controlled within ±5%. After continuous operation for 24 hours in strong acid and strong alkali media, the modified hydrogel coating showed no dissolution or peeling, and the equipment's functional retention rate was >95%.

[0018] Example 3: Blind Testing Verification of Multimodal Algorithm in Real-World Scenarios A blind taste test was conducted on 100 real-life dishes. The system uses a multimodal weighted fusion algorithm combining chewing mechanical hardness data and taste chemical potential data to accurately identify complex flavors, achieving an overall average accuracy rate of 97.0%. It can accurately distinguish between spicy, numbing, salty, and umami complex tastes, while also identifying physical properties such as the hardness and viscosity of food particles. Beneficial effects

[0019] 1. Super strong anti-interference and high-precision perception: Overcomes the problems of grease blockage, bubble interference and sampling failure of viscous media, and has high recognition accuracy in complex scenes.

[0020] 2. Comprehensive safety protection: dual redundancy of mechanical power failure self-locking and electronic intelligent early warning to avoid leakage, backflow, and machine jamming.

[0021] 3. Ultra-long lifespan and low-cost operation and maintenance: The core components have a theoretical lifespan of 20 years, which greatly reduces maintenance costs.

[0022] 4. Humanoid bionic closed loop: Realize a fully closed-loop bionic logic of chewing-sampling-perception-judgment-execution.

[0023] 5. Cross-domain commercial value: It can be used in multiple fields such as bionic robots, medical rehabilitation, health care equipment, and food quality inspection. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the overall system structure of the present invention; Figure 2 This is a cross-sectional view of the biomimetic microfluidic channel of the present invention; Figure 3 This is a schematic diagram of the intelligent swallowing anti-backflow module of the present invention; Figure 4 This is a flowchart of the multimodal sensing and control method of the present invention.

[0025] Figure 1 In the middle: 1—Bionic chewing module, 2—Multimodal taste perception module, 3—Bionic microfluidic channel, 4—Intelligent swallowing anti-backflow module, 5—Main control and safety unit, 6—Chysate flow path. Figure 2 In the middle: 31—pipe body, 32—nano-level hydrophilic coating, 33—micro-guide ridge, 34—capillary drainage direction, 35—grease and bubble isolation and drainage path.

[0026] Figure 3 In the middle: 41—PTFE double-disc valve disc, 42—nickel-titanium alloy memory spring, 43—volume return spring, 44—equipment in energized and conductive state, 45—equipment in de-energized and mechanically locked state.

[0027] Figure 4 In the middle: 10—chewing resistance data input terminal, 20—taste potential data input terminal, 30—multimodal weighted fusion algorithm unit, 40—food characteristic recognition output terminal, 50—foreign object removal control command output terminal, 60—self-cleaning mode start command output terminal, 70—fault audible and visual alarm command output terminal.

Claims

1. A biomimetic feeding system based on multimodal perception, comprising a biomimetic chewing module, a multimodal taste perception module, an intelligent swallowing anti-reflux module, and a main control safety unit, characterized in that: The taste perception module is connected to the chewing cavity through a biomimetic microfluidic channel to achieve synchronous micro-sampling and perception during chewing. The swallowing anti-backflow module is equipped with a power-off mechanical self-locking one-way valve to achieve anti-backflow protection without power dependence.

2. The biomimetic feeding system based on multimodal perception according to claim 1, characterized in that: The sensitive functional coating of the taste perception module is a glutaraldehyde-crosslinked casein hydrogel.

3. The biomimetic feeding system based on multimodal perception according to claim 1, characterized in that: The inner wall of the microfluidic channel is coated with a nano-level hydrophilic coating and micro-guide ridges, and low-power stable sampling is achieved by relying on the capillary principle of rush pith.

4. The biomimetic feeding system based on multimodal perception according to claim 1, characterized in that: The power-off mechanical self-locking one-way valve described in claim 1 has a built-in reset spring structure, which completes mechanical locking within 15ms after power failure.

5. A biomimetic feeding system based on multimodal perception according to claim 1, characterized in that: The system is compatible with a temperature range of -20℃ to 100℃ and a pH range of 1.0 to 12.0, and can operate stably in acidic or alkaline environments.

6. A biomimetic feeding lifecycle control method based on multimodal perception, characterized in that, Includes the following steps: A. When the bionic chewing module is working, it simultaneously collects mechanical data on the hardness and resistance of the food during chewing; B. Microfluidic channels automatically collect chyme samples using capillary action, and combine them with electrode arrays to obtain taste chemistry data; C. Multimodal weighted fusion of mechanical data and taste data to achieve accurate identification of food characteristics; D. Based on the detection data and operating parameters, adaptively trigger foreign object removal, self-cleaning, backflow prevention locking, and fault warning actions.

7. The control method according to claim 6, characterized in that: By using a weighted fusion algorithm of chewing resistance curve and taste potential signal, multi-dimensional and accurate identification of compound ingredients can be achieved.

Images