Ai-driven personalized consumable preparation system with multi-modal user feature mapping

Abstract

A system for personalized consumable preparation includes a plurality of precursor hoppers, a preparation assembly with at least one actuator, and a controller. The controller is configured to acquire a multi-modal signal stream comprising biometric or environmental data and transform the stream into a multi-dimensional numerical representation of user features. The numerical representation is mapped to a preparation parameter set using a set of weighted adaptation parameters stored in memory. The controller then commands the actuator to modulate a physical state parameter of a consumable precursor material based on the preparation parameter set to execute a material transformation. Embodiments include personalized coffee blending systems using optical and physiological sensors to drive motor- actuated bean proportioning, as well as solid consumable preparation systems utilizing high- pressure extrusion and mechanical state modulation based on sensor-derived user feature data.

Description

DescriptionTitle of Invention : Al-Driven Personalized Consumable Preparation System with Multi-Modal User Feature Mapping

[0001] The present invention provides a technical platform for the production of personalized consumable products through the automated modulation of physical preparation parameters. By transforming multi-modal signal streams into machine-executable commands, the system enables a deterministic control loop that bridges sensor-derived user feature data with the mechanical preparation of food and beverages.

[0002] In one aspect, the invention provides a personalized coffee preparation system that assesses user-state indicators — such as derived numerical values for stress or energy levels — using optical or physiological sensors. These signals are converted into multi-dimensional numerical representations and mapped to specific bean ratios and brewing parameters. The system then commands motor- driven actuators, such as augers or valves, to dispense a specific ratio of coffee bean types in accordance with the mapped parameters.

[0003] In another aspect, the invention provides a hardware-agnostic consumables platform capable of preparing solid or semi-solid products. In these embodiments, the system personalizes products by modulating physical state parameters such as compression force, extrusion rates, and curing temperatures. The system may further incorporate identity management and data persistence logic to support preference continuity across preparation sessions and device types, including embodiments supporting anonymous, partially identified, or fully identified user states.TERMINOLOGY BRIDGING AND FUNCTIONAL EQUIVALENCY

[0004] For the purposes of this disclosure, and consistent with the priority application, the terms “artificial intelligence (Al),” “machine learning (ML),” and “algorithm” are technically realized through the transformation of raw signal data into numerical feature representations. This includes the use of multi-dimensional user feature vectors and the mapping of such vectors through a set of weightedadaptation parameters — such as matrices, rule sets, or lookup tables — stored in a non-transitory memory.

[0005] Any reference to “interpreting” or “analyzing” user state (e.g., mood, stress, or energy levels) is defined as a technical process of converting multi-modal sensor streams into digital data structures to drive the physical operation of actuators. The material transformation of a precursor into a personalized consumable product is achieved through the modulation of physical state parameters, including but not limited to dispensing ratios, temperature, and mechanical force, based on these numerical structures. Accordingly, all personalization described herein is implemented as a deterministic or probabilistic control process that results in reproducible machine behavior rather than subjective assessment. Technical Field

[0006] The present invention relates to the field of food and beverage personalization. More specifically, the invention relates to a system and method for personalizing consumable products using multi-modal signal analysis — including behaviorally inferred user context derived from sensor data — to drive machine-executable actuator commands for material transformation.Summary of Invention

[0007] This is a sample text. In certain Offices, however, Rule 20.4(c) is incompatible with the applicable national law. For as long as that incompatibility continues, that Rule will not apply for those Offices; all elements of an international application filed with those Offices as receiving Office must therefore comply with the language requirements of Rule 12.1 before an international filing date can be accorded (see Annex C for details).Technical Problem

[0008] Conventional automated consumable preparation systems, including coffee blending and food dispensing machines, suffer from several technical limitations that prevent consistent and adaptive preparation outcomes:a. Lack of Real-Time Physical AdaptationExisting systems are generally incapable of dynamically adjusting physical preparation parameters — such as ingredient ratios, dispensingtiming, mechanical force, or temperature — in response to real-time sensor-derived user or environmental conditions.b. Dependence on Explicit Manual InputCustomization is typically achieved through fixed presets or manual selection, which limits the system’s ability to respond to sensor-derived indicators such as physiological measurements or facial feature data captured during a preparation session.c. Inconsistent Material Transformation Across Precursors Automated systems frequently struggle to maintain consistent physical outcomes when handling heterogeneous precursor materials, such as different coffee bean varieties, powders, or compressible solids, particularly in the absence of closed-loop machine control.

[0009] These limitations reduce reproducibility, constrain personalization accuracy, and hinder the deployment of adaptive preparation systems across varied device types and usage contexts.Solution to Problem

[0010] The invention provides a technical solution by implementing a deterministic, device-anchored control architecture that converts multi-modal sensor inputs into machine-executable preparation commands.

[0011] The solution includes:a. Sensor-to-Feature TransformationAcquiring a multi-modal signal stream from one or more sensors and transforming the signal stream into a digital data structure comprising a multi-dimensional numerical representation of user features. b. Numerical Mapping to Preparation ParametersProcessing the numerical representation through a set of weighted adaptation parameters stored in memory to determine a preparation parameter set defining machine-level control values, including dispensing ratios, actuation timing, temperature settings, or mechanical force values.c. Physical State Modulation via ActuationCommanding one or more motor-driven actuators of a preparation assembly to modulate a physical state of consumable precursor materials in accordance with the preparation parameter set, thereby executing a controlled material transformation to produce a personalized consumable product.Advantageous Effects of Invention

[0012] Deterministic and Reproducible PersonalizationPersonalization is achieved through numerical feature mapping and actuator control, enabling reproducible preparation outcomes independent of subjective interpretation.

[0013] Closed-Loop Preparation ControlIn expansion embodiments, preparation-state sensors enable real-time adjustment of actuator operation to maintain consistency during material transformation, improving reliability across diverse precursor materials.

[0014] Cross-Session and Cross-Device ContinuityThe system supports association of preparation sessions with stored user profiles, enabling consistent preparation behavior across multiple devices and locations without requiring persistent user identification.

[0015] Improved System TransparencyBy presenting preparation rationale indicators that correlate sensor-derived feature values with specific machine actions, the system provides a transparent and technically explainable preparation process.Brief Description of Drawings

[0016] The accompanying drawings, which form a part of this disclosure, illustrate embodiments of the invention and, together with the description, serve to explain the principles of the invention. The drawings are not intended to limit the scope of the invention, and like reference numerals refer to like elements throughout the drawings.Fig. 1 System Architecture and Environment

[0017] Fig. 1 illustrates a high-level block diagram of the integrated network environment for personalized consumable preparation. The system includes a physical preparation machine 102 — such as an automated coffee or nutritional solid dispenser — configured to communicate with a mobile computing device 104 and a cloud-based profile and synchronization platform 106. Multi-modal sensors 108, which may be integrated into the machine 102 or external wearable devices, provide the real-time signal streams required for personalization.Fig. 2 Technical Eligibility Flow (Signal to Actuation)

[0018] Fig. 2 provides a detailed flow diagram illustrating the transformation of raw sensor signals into machine-executable actuator commands.a. Layer 202: Raw multi-modal signal acquisition (e.g., facial landmark tracking, physiological heart rate streams, and ambient environmental data).b. Layer 204: A feature extraction layer that converts raw signals into a normalized numerical user feature vector.c. Layer 206: The application of a set of weighted adaptation parameters (e.g., a weight matrix or lookup table) to map the feature vector to a preparation parameter set.d. Layer 208: The determination of specific machine-executable actuator commands, such as motor duty-cycles or heater pulse-width modulation, to effectuate material transformation.Fig. 3: Automated Proportioning Hardware (Priority Embodiment)

[0019] Fig. 3 is a structural view of the automated proportioning assembly utilized in the coffee-priority embodiment.a. 302a-n: A plurality of coffee bean hoppers, each associated with distinct flavor profiles.b. 304: At least one motor-driven actuator, such as an auger or proportioning valve, dedicated to each hopper for controlled dispensing. c. 308: A mixing and preparation chamber where the physical dispensing ratio is realized before brewing or processing.Fig. 4: Identity and Session Management

[0020] Fig. 4 illustrates the logic flow for resolving user identity and synchronizing preferences.a. 402: The system initializes a preparation session in an anonymous state to provide immediate recommendations based on current sensor cues. b. 404: Triggering of a biometric handshake or an authenticated device interaction.c. 406: The logic for associating the current session with a stored user profile.d. 408: Execution of data persistence rules for associating current session data with stored user profile information to support continuity across preparation sessions.Fig. 5: User Interface and Parameter Attribution

[0021] Fig. 5 illustrates the interactive interface configured to enhance transparency and trust.a. 502: A transparent preparation visualization area allowing the user to observe the material transformation in real time.b. 504: A display section providing preparation rationale indicators that visually correlate the detected user features, such as derived numerical user feature values (e.g., stress-related indicators inferred from sensor data), with specific adjustments in the preparation parameters.c. 506: Augmented reality (AR) overlays providing educational data regarding precursor origins and flavor profiles.Fig. 6: Non-Beverage Consumable Embodiment (Expansion)

[0022] Fig. 6 illustrates a variant hardware architecture for solid consumable products, supporting the expansion claims (13-22).a. 602: Precursor hoppers containing dry ingredients or nutritional concentrates.b. 604: A high-pressure extrusion and compression assembly.c. 608: collection point for mixtured. 610: Support rail for the solid hopperse. 612: machine control unitf. 614: wall mount for group of solid hoppers arrayg. 616: connecting tubes between hoppers and mixing pointFig. 7: Exploded view of machine main controller

[0023] Fig. 7 illusrrates an exploded view of the machine main controllera. 702: power supply AC-DC converter for solid hoppers powering b. 704: Main controller and WiFi module for communicating with cloud with specific credentialsc. 706: RJ sockets for power transmission to solid hoppers array d. 708: AC socket for AC poweringFig. 8: Cross section of solid beans dispenser

[0024] Fig. 8 illustrates a corss section of the solid beans dispensera. 802: Solid hopper controllerb. 804: DC motor for dispensing mechanismc. 806: Servo motor for dispensing solid ingredientsd. 808: Loadcell for weighinge. 810: weighing spoonf. 812: Auger for dispensing the solid ingredientsDescription of EmbodimentsSystem Architecture and Data Flow

[0025] The system is implemented as a hardware-agnostic platform for the personalized preparation of consumable products. As illustrated in FIG. 1, the architecture comprises a physical preparation machine 102, a mobile computing device 104, and a cloud-based profile synchronization platform 106.

[0026] The core technical operation involves a closed-loop control process where multi-modal signals are converted into specific machine-executable actuator commands. This process is orchestrated by a controller comprising processing circuitry and memory storing weighted adaptation parameters. By decoupling the personalization logic (feature extraction and mapping) from the specific machine hardware (actuator control), the system enables consistent personalization across heterogeneous device types.Multi-Modal Signal Acquisition and Feature Extraction

[0027] Personalization begins with the acquisition of a multi-modal signal stream via one or more sensors 108. These sensors include optical sensors for facial landmark tracking, physiological sensors for heart rate and hydration monitoring, and, in some embodiments, environmental sensors for ambient light and temperature detection.

[0028] The controller performs a technical transformation of these raw signals into a digital data structure, defined herein as a multi-dimensional numerical user feature vector. For example, facial landmarks are processed to derive numerical values representing a stress-related indicator or an energy-level indicator.Physiological streams, such as heart rate variability (HRV), are normalized and integrated into the feature vector. This conversion ensures that "user state" is handled as a deterministic numerical input rather than a subjective assessment, satisfying technical eligibility requirements.Numerical Mapping and Weighted Adaptation

[0029] Once the user feature vector is generated, the controller applies a set of weighted adaptation parameters to determine a preparation parameter set. These parameters are stored in memory as a weight matrix, a lookup table, or a deterministic rule set.

[0030] In one embodiment, the mapping logic applies specific weights to the features within the vector to calculate machine-control values. For instance, a high numerical value for a stress-related indicator may be mapped to a preparation parameter set that increases the ratio of a lower-caffeine coffee bean type or a coffee bean type associated with a mellow flavor profile while decreasing the ratio of a high-caffeine precursor. This mapping results in a preparationparameter set that defines the physical state of the final product, such as ingredient ratios, temperature, and mechanical processing force.Coffee-Priority Embodiments (Claims 1-12)

[0031] In the priority embodiment directed to coffee preparation, the machine 102 includes a plurality of coffee bean hoppers 302a-n, as shown in FIG. 3. Each hopper contains a distinct bean variety (e.g., varying by roast level or origin). The automated proportioning assembly utilizes motor-driven actuators 304, such as high-precision augers, to dispense beans into a mixing chamber 308.

[0032] The controller modulates the physical state of the assembly by sending individual duty-cycle instructions to the actuators. For example, to achieve a specific 70 / 30 blend ratio, a first auger is commanded to operate at a first speed for a first duration, while a second auger is commanded to operate at a second speed for a second duration. The combined beans are then ground and brewed. The system ensures reproducibility by anchoring the dispensing process to the numerical preparation parameters derived from the user feature vector.Consumables Expansion and Material Transformation (Claims 13-22)

[0033] Beyond coffee, the system is configured for the preparation of a wide range of consumable products, including nutritional solids and semi-solids. As illustrated in FIG. 6, the preparation assembly may include high-pressure extrusion and compression hardware 604.

[0034] Personalization in these embodiments involves modulating physical state parameters such as compression force and extrusion rates. For a solid nutritional product, the preparation parameters define a specific mechanical pressure to be applied by an actuator to a mixture of precursors (e.g., powders and binders).

[0035] In expansion embodiments corresponding to Claims (13-22), the system may utilize acoustic sensors to monitor the material transformation. The controller analyzes acoustic signals indicative of a preparation state — such as the resonance of a grinding cycle or the density of an extrusion flow — to adjust the actuator commands in real time. This closed-loop feedback ensures the physical properties of the consumable product match the personalized target attributes.Identity Orchestration and Data Persistence

[0036] The system manages user continuity through the logic illustrated in FIG.4. A session typically begins in an anonymous state, allowing for immediate personalization based on current sensor cues (e.g., a walk-up customer whose facial landmarks are analyzed).

[0037] Upon a biometric handshake (e.g., facial recognition) or an authenticated device interaction (e.g., a mobile app handshake), the controller associates the current session data with a stored user profile. In priority-safe embodiments, this involves synchronizing session-specific preferences to support continuity across locations. In expansion embodiments, the controller may selectively assign a higher priority to verified profile data over inferred session data to resolve conflicts between historical preferences and current situational context.Interactive Interface and Rationale Visualization

[0038] To provide a transparent and engaging experience, the system includes a display interface 504. This interface presents preparation rationale indicators that visually correlate the derived numerical user feature values with the adjustments made to the product. For example, the display may show: "High workload-related indicators detected: adjusting blend to 80% smooth roast. "This visualization provides the user with an understanding of the deterministic link between the sensed data and the physical machine behavior.Power delivery and dispensing order

[0039] The controller is powered by a AC cord connected to AC socket 708 to transfer the power to power supply 702 which provide the required power for coffee beans dispensing units 304. The main controller 704 connects to internet via internal WiFi communication to communicate with cloud and coffee dispensing units 304.Dispensing mechanism

[0040] The machine consists of array of dispensing units 304 each with a different coffee beans blend, coffee beans is fed to the machine via a silo 602, the feeding screw 812 then feeds the beans until it falls into the spoon 810, the spoon in attached to a load cell 808, when the load cell weighs the required amount ofbeans the screw stops the spoon rotates via a servo motor 808 to spill the beans into a transparent tube 616 then different beans meet at the tubes junction 608.

[0041] When an order is sent from a PC software 104 it first routs to the cloud server then sent to the main controller 704, then main controller 704 send data to the solid hopper controller 802, then solid hopper controller 802 runs the DC motor 804 at maximum speed till 80% of targeted weight and then the it slow down to 10% of speed till the loadcell 808 connected to spoon 810 reads the target weight then the servo motor 810 rotates to dispense the solid beans to transparent tubes 616 and delivering the beans to collection point 608^

Claims

Claims

1. A system for personalized coffee blending, the system comprising: a. a plurality of coffee bean hoppers, each associated with a distinct flavor profile;b. an automated proportioning assembly comprising at least one motor- driven actuator configured to dispense coffee beans from the plurality of coffee bean hoppers;c. one or more sensors configured to acquire a multi-modal signal stream comprising at least one of facial expression data, questionnaire data, or physiological data; andd. a controller in communication with the one or more sensors and the automated proportioning assembly, the controller comprising a processor and a memory storing instructions that, when executed, cause the processor to:i. transform the multi-modal signal stream into a multi-dimensional numerical representation of user features;ii. map the multi-dimensional numerical representation of user features to a preparation parameter set by processing the multi-dimensional numerical representation through a set of weighted adaptation parameters stored in memory; andiii. command the operation of the at least one motor-driven actuator based on the preparation parameter set to drive dispensing of a specific ratio of coffee beans from the plurality of coffee bean hoppers to form a personalized coffee blend.

2. A method for personalizing a coffee blend, the method comprising: a. acquiring, via one or more sensors, a multi-modal signal stream comprising at least one of facial expression data, questionnaire data, or physiological sensor data;b. converting, via a processor, the multi-modal signal stream into a digital data structure representing user state features;c. determining a preparation parameter set by mapping the digital data structure to machine control values using a set of weighted transformation parameters; andd. modulating the physical state of an automated proportioning assembly based on the preparation parameter set to dispense a specific ratio of coffee bean types from a plurality of coffee bean hoppers into a preparation chamber to form a personalized coffee blend.

3. A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform operations for personalized coffee preparation, the operations comprising: a. receiving a multi-modal signal stream including user-specific biometric or environmental data;b. converting the multi-modal signal stream into a numerical representation of user features;c. processing the numerical representation of user features through a set of weighted adaptation values to define preparation parameters for a coffee preparation machine; andd. causing operation of one or more actuators of the coffee preparation machine to effectuate physical dispensing of a plurality of coffee bean types from a plurality of hoppers according to the defined preparation parameters.

4. The system of claim 1, wherein the one or more sensors are selected from a group consisting of: an optical sensor for facial analysis and a physiological sensor for acquiring heart rate or hydration data.

5. The system of claim 1, wherein the multi-dimensional numerical representation of user features comprises at least one value corresponding to an inferred stress level or an inferred energy level.

6. The system of claim 1, wherein the set of weighted adaptation parameters stored in memory comprises at least one of a lookup table mapping defined user feature ranges to specific bean ratios or a rule set defining minimum and maximum thresholds for the preparation parameter set.

7. The system of claim 1, wherein the controller is configured to command the operation of the at least one motor-driven actuator by modulating at least one of a motor speed, an actuator duty-cycle, or a dispensing duration.

8. The system of claim 1, wherein the automated proportioning assembly is commanded to perform a multi-stage dispensing operation comprising a primary dispensing phase followed by a stabilization phase.

9. The system of claim 1, wherein the controller is further configured to initialize a coffee preparation session in an anonymous state, detect a specific user identity via a biometric handshake or a communication with a mobile device, and associate current session data with a stored user profile associated with the specific user identity.

10. The system of claim 1, wherein the controller is configured to execute data persistence rules for synchronizing current session data with the stored user profile to support continuity across preparation sessions.

11. The system of claim 1, further comprising a display interface configured to present preparation rationale indicators that visually correlate the multi-dimensional numerical representation of user features with specific adjustments made to the ratio of coffee beans.

12. The system of claim 1, further comprising a transparent preparation area and an educational display, wherein the educational display provides real-time data regarding a coffee origin or a flavor profile of the personalized coffee blend being dispensed.

13. A system for personalized consumable preparation, the system comprising:a. a plurality of precursor hoppers, each containing a consumable precursor material;b. a preparation assembly comprising at least one actuator configured to modulate a physical state of the consumable precursor material; c. one or more sensors configured to acquire a multi-modal signal stream comprising at least one of biometric data or environmental data; andd. a controller in communication with the one or more sensors and the preparation assembly, the controller comprising a processor and a memory storing instructions that, when executed, cause the processor to:i. transform the multi-modal signal stream into a numerical representation of user features;ii. map the numerical representation of user features to a set of preparation parameters by processing the numerical representation through a set of weighted adaptation values; and iii. execute a material transformation of the consumable precursor material by commanding the at least one actuator to modulate a physical state parameter of the consumable precursor material based on the set of preparation parameters to form a personalized consumable product.

14. A method for personalizing a consumable product, the method comprising:a. acquiring, via one or more sensors, a multi-modal signal stream including user-specific biometric or environmental data;b. converting the multi-modal signal stream into a digital data structure representing user state features;c. determining a preparation parameter set by mapping the digital data structure to machine control values using a set of weighted transformation parameters; andd. modulating a physical state of a plurality of consumable precursors via a preparation assembly to physically transform the plurality of consumable precursors into a personalized consumable product based on the preparation parameter set, wherein modulating the physical state comprises at least one of a dispensing ratio adjustment, a temperature modulation, or a compression force adaptation.

15. A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform operations for personalized consumable preparation, the operations comprising:a. receiving a multi-modal signal stream including user-specific biometric or environmental data;b. converting the multi-modal signal stream into a numerical representation of user features;c. processing the numerical representation of user features through a set of weighted adaptation values to define preparation parameters for a preparation machine; andd. causing operation of one or more actuators of the preparation machine to effectuate a material transformation of one or more consumable precursors into a personalized consumable product by modulating at least one physical state parameter according to the defined preparation parameters.

16. The system of claim 13, wherein the personalized consumable product comprises a solid or semi-solid product, and wherein the at least one actuator is configured to modulate a physical state parameter comprising at least one of a compression force or an extrusion rate.

17. The system of claim 13, wherein the material transformation further comprises a curing phase, and wherein the set of preparation parameters defines a curing temperature or a curing duration for the personalized consumable product.

18. The system of claim 13, wherein the plurality of precursor hoppers contains at least two materially different types of precursors selected from the group consisting of: a liquid concentrate, a dry powder, and a solid pellet.

19. The system of claim 13, wherein the set of weighted adaptation values is configured to adapt the material transformation of a first precursor type for use in a first consumable product domain and a second consumable product domain by modifying at least one physical state parameter.

20. The system of claim 13, wherein the one or more sensors are selected from a group consisting of: an acoustic sensor configured to detect acoustic signals indicative of a preparation state, an electrical conductivity sensor, and a thermal profile sensor.

21. The system of claim 13, wherein the controller is configured to adjust the set of preparation parameters based on a physical constraint of the preparation assembly stored in a hardware-specific lookup table.

22. The system of claim 13, wherein the controller is further configured to modulate the physical state parameter based on a historical preference retrieved from a user profile after a biometric handshake or an authenticated device interaction is performed, wherein the controller is configured to resolve data conflicts by selectively assigning a higher priority to verified profile data than to inferred session data.

Images