Life entropy minimization principle-based life-like robot or intelligent system construction method

Abstract

The invention provides a construction method of a life-like robot or intelligent system based on a life entropy minimization principle. The method aims at solving the problem that in the prior art, unified description and quantification of vital activity essence are lacked from the aspects of mathematics and information science, and a similar life intelligence system is not built accordingly. According to the method, a sensing and internal steady state system of a machine is constructed by defining ternary state coding (positive surprising, negative surprising and ground state) of a self-information unit (); constructing each function module topology of the machine by using a self-information complex (); and taking the minimum life entropy value as a core objective function of a machine system. A machine system eliminates a surprising state with high information content in real time through a PID feedback mechanism, so that the surprising state spontaneously returns to a low-entropy ordered ground state. According to the invention, the machine is endowed with the instinct of'benefiting and avoiding harm 'similar to organisms, and the method is suitable for constructing life-like robots or intelligent systems with high adaptivity.

Description

Technical Field

[0001] This invention relates to the fields of bioinformatics and artificial intelligence, and particularly to a method for constructing life-like robots or intelligent systems based on the principle of minimizing biological entropy. This invention is applicable to the architectural design of general artificial intelligence systems, the autonomous control of embodied intelligent robots, and the homeostasis management of complex network systems. Background Technology

[0002] With the development of artificial intelligence and robotics, machines have achieved performance close to or even surpassing the average human level in specific tasks (such as image recognition and path planning). However, existing intelligent machines are essentially still "open-loop" or "passive execution" systems, lacking an "adaptive homeostasis maintenance mechanism" similar to that of biological systems. Specifically: Lack of a unified internal homeostatic driving force: The behavior of traditional robots relies on preset rules or reward functions for specific tasks. When faced with unknown environmental disturbances beyond the training data, the machine cannot spontaneously perceive its own "degree of disorder" (entropy increase) and actively seek to return to the system's steady state, as complex adaptive systems do, often leading to system malfunction or collapse.

[0003] The separation of information and energy: From a thermodynamic perspective, highly ordered systems need to consume energy to resist entropy increase. However, in current robot design, energy management and information processing are separated, lacking a mathematical framework that can uniformly quantify the machine's hardware and software states into a "degree of order index" of the system.

[0004] Limitations of traditional information theory: Shannon's information theory mainly focuses on eliminating uncertainty in communication channels, making it difficult to directly guide how to construct an intelligent entity with self-sustaining and self-regulating capabilities.

[0005] Therefore, there is an urgent need for a new methodology for machine building that can transform the physical principle of "countering entropy increase" into an engineering-feasible system architecture and control algorithm, thereby creating a life-like intelligent system with high robustness and adaptability. Summary of the Invention

[0006] This invention aims to address the lack of autonomous steady-state regulation mechanisms and unified global stability indicators in existing intelligent systems, and provides a method for constructing life-like robots or intelligent systems based on the principle of minimizing life entropy. By introducing the concept of "entropy" from physics, this invention endows machines with a self-regulating characteristic centered on "maximizing system orderliness."

[0007] To achieve the above objectives, the present invention adopts the following technical solution: a method for constructing a life-like robot or intelligent system based on the principle of minimizing life entropy, comprising the following steps: Step S1, System Modularization and Unit Definition: Construct the hardware architecture and software environment of the robot or intelligent system, defining the system's external perception module, internal equilibrium state module, information processing and integration module, and output module as independent self-information units. And configure data acquisition channels to obtain real-time variable data from each unit. ; Step S2, Implanting Probability Awareness and Ternary Encoding Logic: Configure a probability statistics module for each self-information unit, based on variable data. Historical probability distribution Set steady-state threshold parameters And construct a ternary state encoding logic to determine the unit state in real time: the state falling into the high probability interval is the ground state (0), and the state falling into the low probability interval is the positive surprise state (+1) or the negative surprise state (-1). Step S3: Construct a self-information joint topology network: Based on the probability distribution of functional cooperation and signal transmission relationships between the respective information units, establish the graph theory topology structure within the system. To build a self-information union And establish a joint self-information flow chain that reflects the evolution of the system state over time. ; Step S4: Configure the entropy minimization control core: Deploy the central controller in the system and configure the life entropy value. The computational algorithm serves as the global optimization objective function of the system; The overall disorder of the system is characterized by the reciprocal of the sum of the self-information of all self-information units and the self-information of the joint self-information flow chain. Step S5: Establish a closed-loop feedback control strategy: Construct a feedback adjustment loop connecting the controller and the actuator; when the system detects an unexpected state leading to... When the temperature rises, the controller determines the minimum value. The principle generates adjustment commands, driving the actuator to adjust the variables of the relevant self-information unit. Compensation or suppression can be performed to force the system state to return to the ground state, thereby maintaining the life-like homeostatic characteristics of the system.

[0008] Furthermore, in step S1, the self-information unit Covering both living and non-living entities, specific data sources include: External sensing modules: such as data acquired from various visual sensors, auditory sensors, tactile sensors, odor sensors, etc., are used to characterize the machine's adaptability to the environment; Internal balance status module: such as battery voltage, processor temperature, motor torque, memory usage, or internal communication latency, used to characterize the health of a machine or system; Information processing and integration module: Corresponds to the system's central processing unit or neural network fusion layer, used for the integration, feature extraction and logical processing of multimodal sensing data and state data; Output module: Used to provide feedback on the response and output results of the actuator (such as a motor or robotic arm); Each module generates data within a certain time frame, resulting in a corresponding probability distribution.

[0009] Furthermore, in step S2, the self-information unit State coding is the logical foundation of machine perception and homeostasis regulation. It is determined based on ternary information coding, and the specific rules are as follows:

[0010] In mathematics, express of quantiles ( -quantile), defined as satisfying the cumulative probability The critical value; when When the value is 0, it represents a stable ground state with low information content; when... A value of +1 or -1 represents a state of positive or negative surprise with high information content.

[0011] Furthermore, in step S3, the self-information union The construction maps the various functional modules in a machine or living organism to nodes in graph theory. Mapping the information flow or energy flow between different modules as edges To form a dynamic network structure The joint self-information quantity of the self-information association Through formula Calculations are used to evaluate the orderliness and coordination of the collaborative work of the various modules of the system, among which... This represents the probability distribution of joint activities.

[0012] Furthermore, in step S4, the life entropy value The calculation formula is:

[0013] in, It represents the summation of the self-information units and joint self-information flow chains over time series of all functional modules within a machine or system; the lower the life entropy value, the higher the orderliness and complexity of the living organism.

[0014] Furthermore, in step S5, the feedback control strategy is a PID-like proportional-integral-derivative control strategy based on information entropy.

[0015] The beneficial effects of this invention are as follows: Endowing machines with "self-sustaining homeostasis": Machines are no longer merely tools for executing instructions, but possess an inherent mathematical driving force to maintain their own low-entropy ordered state, thereby significantly improving the robustness of the system in unknown environments.

[0016] A general steady-state architecture: This method provides a general mathematical paradigm that can be applied to silicon-based robots, software-level server clusters, and other complex engineering systems. The indicators are used for health assessment and automatic control.

[0017] Adaptive evolution capability: By continuously updating the probability distribution model, the machine can adapt to new environmental norms and achieve autonomous optimization and upgrading of the system control logic. Attached Figure Description

[0018] Figure 1 This is a flowchart of the present invention; Figure 2 This is a schematic diagram of the three-state encoding of the self-information unit based on probability distribution in this invention; Figure 3 This is a schematic diagram of the joint self-information chain in this invention; Figure 4 This is a block diagram of the self-information feedback control logic of the intelligent robot in this embodiment of the invention. Detailed Implementation

[0019] The present invention will be further described in detail below with reference to specific embodiments and accompanying drawings. The scope of protection of the present invention is not limited to the above embodiments. Variations and advantages that can be conceived by those skilled in the art without departing from the spirit and scope of the inventive concept are included in the present invention and are protected by the appended claims.

[0020] like Figure 1-4 As shown, this embodiment illustrates a design scheme for a biomimetic robot system using the construction method of the present invention, demonstrating how the present invention can be applied to the design, control, and intelligence level assessment of silicon-based artificial life. Figure 1 As shown, the specific steps include: Step S1: Construct the self-information mapping architecture of the life-like intelligent system and define the data channels for each functional module. During the system construction phase, the physical entities or virtual components of the intelligent system are first abstracted and mapped into four types of self-information units. And configure the corresponding data flow interface. Specifically, construct the external sensing module and define the external environment sensing unit. Used to collect environmental feature vectors, its output data is configured as follows: The internal balance state module defines the internal resource state units. Used to monitor the computing load, energy consumption level, and health indicators of the system, its output data is configured as follows: Construct an information processing integration module and define information processing integration units. It is used to integrate multimodal sensing data and state data, extract features, and perform logical operations; it constructs output modules and defines output execution units. Used to provide feedback on the dynamic response or logical operation results of the actuator, its output data is configured as follows: Through the above mapping, the standardized definition of multi-source heterogeneous data at the system's underlying layer is completed.

[0021] Step S2: Deploy a probability distribution model and a ternary state encoder in the architecture to transform the raw data into a probability distribution over any time period and further standardize the self-information state sequence. The system configuration process includes model initialization and encoding logic deployment. First, the probability density function trained based on historical runtime data is embedded. Set a baseline steady-state interval based on statistical significance for each data channel. Subsequently, a ternary comparator is deployed to calculate the input data in real time. The mapping relationship between the data and the threshold is as follows: data falling into the steady-state range is encoded as the ground state (0) representing normal conditions; data falling into the low-energy-level anomalous range is encoded as the positive surprise state (+1) representing resource scarcity; and data falling into the high-energy-level overload range is encoded as the negative surprise state (-1) representing system stress. Finally, a time alignment module is configured to synchronize the encoding results of each channel in time, and output multi-dimensional state matrix data. .like Figure 2 The figure shows a schematic diagram of the three-state encoding of the self-information unit based on probability distribution.

[0022] Step S3: Establish a self-information federation The topological network is constructed, and a joint self-information flow chain is generated based on this network. Based on the causal dependencies of the system (e.g., environmental load affects execution power consumption, thus changing internal resource consumption), the probability distribution between these dependencies is calculated, and a directed graph model is established. ,in For each node, For information coupling edges, configure a joint probability calculation module based on the topology graph. Given structural constraints, calculate the joint probability of multiple node states occurring simultaneously. Output flow chain data characterizing the overall synergy of the system. .like Figure 3 The diagram shown is a schematic of a joint self-information chain.

[0023] Step S4: Configure with life entropy The core global objective function is defined, and an entropy evaluation engine is established. The entropy evaluation engine is configured in the control layer, and this engine reads state data in real time. With Flow Chain Data According to the formula Solve the current order score of the system. Based on this, the core logic of the system controller is configured to manage the life entropy. Minimizing the value is the sole objective function for the convergence of all decision paths, meaning that all behavioral decisions of the system are ultimately guided by maximizing the overall orderliness of the system, rather than the single task completion rate.

[0024] Step S5: Construct a negative entropy feedback closed loop based on model predictive control and generate adjustment commands. The system is driven back to its ground state. The specific operational logic of the closed-loop control is as follows: Based on the current state input state transition model, the controller predicts the state evolution sequence under different candidate strategies (such as aggressive execution strategy or frequency reduction self-preservation strategy) within the future time window; for each predicted sequence, the corresponding expected lifetime entropy value is calculated. The strategy that minimizes the expected entropy is selected; then the optimal strategy is transformed into a low-level control signal. Commands such as voltage regulation, thread suspension, or motor braking are sent to the actuator. Simultaneously, the system monitors the feedback data after execution in real time. Calculate the residual between the actual entropy value and the expected entropy value, and use the residual to update the probability model online. This allows the system to adaptively evolve in response to environmental changes. For example... Figure 4 The diagram shown is a closed-loop control architecture diagram.

Claims

1. A method for constructing a life-like robot or intelligent system based on the principle of minimizing life entropy, characterized in that, Specifically, the steps include the following: Step S1, System Modularization and Unit Definition: Construct the hardware architecture and software environment of the robot or intelligent system, defining the system's external perception module, internal equilibrium state module, information processing and integration module, and output module as independent self-information units. And configure data acquisition channels to obtain real-time variable data from each unit. ; Step S2, Implanting Probability Awareness and Ternary Encoding Logic: Configure a probability statistics module for each self-information unit, based on variable data. Historical probability distribution Set steady-state threshold parameters And construct a ternary state encoding logic to determine the unit state in real time: the state falling into the high probability interval is the ground state, and the state falling into the low probability interval is the positive surprise state or the negative surprise state. Step S3: Construct a self-information joint topology network: Based on the probability distribution of functional cooperation and signal transmission relationships between the respective information units, establish the graph theory topology structure within the system. To build a self-information union And establish a joint self-information flow chain that reflects the evolution of the system state over time. ; Step S4: Configure the entropy minimization control core: Deploy the central controller in the system and configure the life entropy value. The computational algorithm is used as the global optimization objective function of the system; The overall disorder of the system is characterized by the reciprocal of the sum of the self-information of all self-information units and the self-information of the joint self-information flow chain. Step S5: Establish a closed-loop feedback control strategy: Construct a feedback adjustment loop connecting the controller and the actuator; when the system detects an unexpected state leading to... When the temperature rises, the controller determines the minimum value. The principle generates adjustment commands, driving the actuator to adjust the variables of the relevant self-information unit. Compensation or suppression can be performed to force the system state to return to the ground state, thereby maintaining the life-like homeostatic characteristics of the system.

2. The method for constructing a life-like robot or intelligent system based on the principle of minimizing life entropy according to claim 1, characterized in that, In step S1, the self-information unit Covering both living and non-living entities, specific data sources include: External sensing module: Data acquired from various visual sensors, auditory sensors, tactile sensors, and odor sensors is used to characterize the machine's adaptability to the environment; Internal balance status module: battery voltage, processor temperature, motor torque, memory usage, or internal communication latency, used to characterize the health of the machine or system; Information processing and integration module: Corresponds to the system's central processing unit or neural network fusion layer, used for the integration, feature extraction and logical processing of multimodal sensing data and state data; Output module: Used to provide feedback on the response and output results of the actuator; Each module generates data within a certain time frame, resulting in a corresponding probability distribution.

3. The method for constructing a life-like robot or intelligent system based on the principle of minimizing life entropy according to claim 1, characterized in that, In step S2, the self-information unit State coding is the logical foundation of machine perception and homeostasis regulation. It is determined based on ternary information coding, and the specific rules are as follows: In mathematics, express of quantiles ( -quantile), defined as satisfying the cumulative probability The critical value; when When the value is 0, it represents a stable ground state with low information content; when... A value of +1 or -1 indicates a state of positive or negative surprise with high information content.

4. The method for constructing a life-like robot or intelligent system based on the principle of minimizing life entropy according to claim 1, characterized in that, In step S3, the self-information union The construction maps the various functional modules in a machine or living organism to nodes in graph theory. Mapping the information flow or energy flow between different modules as edges To form a dynamic network structure The joint self-information quantity of the self-information association Through formula Calculations are used to evaluate the orderliness and coordination of the collaborative work of the various modules of the system, among which... This represents the probability distribution of joint activities.

5. The method for constructing a life-like robot or intelligent system based on the principle of minimizing life entropy according to claim 1, characterized in that, In step S4, the life entropy value The calculation formula is: in, It represents the summation of the self-information units and joint self-information flow chains over time series of all functional modules within a machine or system; the lower the life entropy value, the higher the orderliness and complexity of the living organism.

6. The method for constructing a life-like robot or intelligent system based on the principle of minimizing life entropy according to claim 1, characterized in that, In step S5, the feedback control strategy is a PID-like proportional-integral-derivative control strategy based on information entropy.

Images