A high-risk operation robot control system and method imitating human nervous system

Abstract

The application belongs to the field of robot control, and provides a high-risk operation robot control system and method imitating human nervous system. The high-risk operation robot control system imitating human nervous system comprises a sensing element, a lower computer, a middle computer and an upper computer; the sensing element is used for sensing robot state information and transmitting the information to the corresponding lower computer; the lower computer is used for transmitting the received robot state information to the upper computer through the middle computer; the upper computer is used for forming instruction information according to a high-risk operation preset target and the received robot state information, and issuing the instruction information to the corresponding lower computer through the middle computer; and the lower computer is also used for controlling the corresponding motion component to execute the corresponding instruction information, and feeding back the received robot state information to the upper computer through the middle computer, to form a sensing-decision-execution closed loop control.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of robot control, and particularly relates to a high-risk operation robot control system and method simulating human nervous system. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute prior art.

[0003] In a high-risk complex environment, it is a challenging task for a robot to perform rescue, detection and other operations. Taking the post-earthquake environment as an example, the post-earthquake environment is complex, the surface features of cities and mountainous areas are different, and there are aftershocks. When performing rescue and other tasks, the robot is required to have high reaction speed, real-time perception ability and instantaneous reaction ability. This determines that the control system cannot use distributed architecture and hybrid architecture. These two architectures are not suitable for application in this environment in terms of data real-time performance, cost and complexity. The high-risk complex environment has high requirements for the motion complexity and motion accuracy of the robot. A single controller, such as a Raspberry Pi 4B, although has good real-time performance and low compression cost, has limited computing power, poor expandability and is not suitable for high-risk complex environments. SUMMARY

[0004] In order to solve the technical problems existing in the background art, the present application provides a high-risk operation robot control system and method simulating human nervous system, which fully considers the chip computing power level, expansion capability, use convenience, high-risk complex environment and other factors, refers to the human nervous system, and can ensure the rapid reaction and motion accuracy of the robot in a high-risk complex environment.

[0005] In order to achieve the above purpose, the present application adopts the following technical solutions:

[0006] The first aspect of the present application provides a high-risk operation robot control system simulating human nervous system.

[0007] A high-risk operation robot control system simulating human nervous system comprises a sensing element, a lower computer, a middle computer and an upper computer.

[0008] The sensing element is used to perceive robot state information and transmit it to the corresponding lower computer; the lower computer is used to transmit the robot state information received by it to the upper computer through the middle computer; the upper computer is used to form instruction information according to the high-risk operation preset target and in combination with the received robot state information, and then issue it to the corresponding lower computer through the middle computer; the lower computer is also used to control the corresponding motion components to execute the corresponding instruction information, and at the same time, feedback the received robot state information to the upper computer through the middle computer, forming a perception-decision-execution closed-loop control.

[0009] The lower computer and the middle computer and the middle computer and the upper computer adopt a general instruction format communication; the general instruction format is: [frame header] [control flag] [instruction type] [data area] [check code] [frame tail].

[0010] As an embodiment, the data area is composed of a plurality of data blocks, and each data block is in the format of: [action] [target] [value]; the value is variable-length encoded according to its range and type.

[0011] As an embodiment, in the data block, if the value range is ≤255, 1 byte is used; if the value range is ≤65535, 2 bytes are used; if it is a floating point number, 4 bytes are used.

[0012] As an embodiment, in the data block, the action is 1 byte, the high 2 bits represent the action type, and the low 6 bits are the action ID.

[0013] As an embodiment, in the data block, the target is 1 byte, used to represent the target device or joint number, and its range is 0-255.

[0014] As an embodiment, the model of the upper computer is:

[0015] ;

[0016] Among them, is the upper computer instruction; S high is the perception data directly interacting with the upper computer; API LLM is the large language model API interaction result; S MLC is the feedback data integrated by the middle computer.

[0017] As an embodiment, the model of the middle computer includes: an instruction decomposition model, a data integration model and a lightweight computing model;

[0018] The instruction decomposition model is: ;

[0019] The data integration model is: ;

[0020] The lightweight computing model is: ;

[0021] Among them, is the upper computer instruction; C high,i is the decomposed upper computer sub-instruction; F is the instruction decomposition function; D mid is the middle layer data; S MLCThe backhaul data integrated for the middle machine; D low,i The data for the lower machine; The feedback data of the lower machine after light processing; G MCC The integrated function; The light processing calculation function.

[0022] As an embodiment, the lower machine model includes: a control signal generation model and a data preprocessing model;

[0023] The control signal generation model is: ;

[0024] The data preprocessing model: ;

[0025] Wherein, C low,i The lower machine instruction; S low,i The sensor data; U i The control signal; D low,i The preprocessed data; The control signal generation function; The data preprocessing function.

[0026] As an embodiment, the instruction priority of the upper machine is higher than that of the middle machine; the instruction actually executed by the lower machine Decided by the following logic:

[0027] ;

[0028] Wherein, The upper machine sub-instruction; The middle machine sub-instruction.

[0029] As an embodiment, the delay of the information transmission process between the lower machine and the middle machine and between the middle machine and the upper machine is optimized by using the information flow delay model:

[0030] The information flow delay model of the upper machine to the middle machine is: ;

[0031] The information flow delay model of the middle machine to the lower machine is: ;

[0032] The information flow delay model of the lower machine to the middle machine is: ;

[0033] The information flow delay model of the middle machine to the upper machine is: ;

[0034] Wherein, T high→mid The delay of the upper machine sending instructions to the middle machine; T mid→lowThe delay for the middle machine to send instructions to the lower machine; T low→mid The delay for the lower machine to return data to the middle machine; T mid→high The delay for the middle machine to return data to the upper machine; C low,i The lower machine instruction; The upper machine instruction; D low,i The lower machine data; The upper machine data.

[0035] The second aspect of the application provides a high-risk operation robot control method simulating the human nervous system.

[0036] A high-risk operation robot control method simulating the human nervous system, based on the high-risk operation robot control system simulating the human nervous system as described above; comprising:

[0037] The sensing element senses the robot state information and transmits it to the corresponding lower machine; the lower machine transmits the received robot state information to the upper machine through the middle machine; the upper machine forms instruction information according to the high-risk operation preset target and combines the received robot state information to form instruction information and issues it to the corresponding lower machine through the middle machine; the lower machine controls the corresponding motion component to execute the corresponding instruction information, while feeding back the received robot state information to the upper machine through the middle machine, forming a perception-decision-execution closed-loop control;

[0038] Among them, the general instruction format communication is adopted between the lower machine and the middle machine and between the middle machine and the upper machine; the general instruction format is: [frame header] [control flag] [instruction type] [data area] [check code] [frame tail].

[0039] The beneficial effects of the application are:

[0040] (1) The high-risk operation robot control system simulating the human nervous system has a hierarchical architecture to optimize system performance and scalability; among them, the modular design: the three-level hierarchical architecture of the upper machine (HLC), the middle machine (MLC) and the lower machine (LLC) realizes the function decoupling, the upper machine focuses on global task planning and intelligent decision-making (such as calling large language model API), the middle machine is responsible for instruction decomposition and lightweight calculation, and the lower machine executes specific motion control and low-precision data acquisition. This design significantly improves the parallel processing capability and resource allocation efficiency of the system; flexible expansion: through the data integration capability of the middle machine, the system can dynamically expand the number of sensors and actuators, adapt to the changing hardware needs in high-risk environments (such as adding sensors or robot joints).

[0041] (2) The application has high reliability: the lower machine directly controls the key motion components to ensure the instantaneous response of emergency instructions (such as obstacle avoidance), and the upper machine priority logic (Clow,i) ensures the control safety in high-risk scenarios.

[0042] (3) The universal instruction format of the present application is efficient and reliable:

[0043] High transmission efficiency: Dynamic data area and variable-length coding design (such as numerical field allocation of 1 / 2 / 4 bytes according to demand) reduce redundant data, and compared with Modbus RTU, the transmission efficiency of small data volume is improved to 50% (N=1), and the efficiency of large data volume reaches 89.5% (N=10).

[0044] Strong fault tolerance: Using double frame header / frame tail (0xAA55 and 0xFFEE) and CRC-16 check, the packet loss rate is reduced by 50% in a noisy environment, the error code detection rate reaches 99.998%, and the communication reliability in a complex electromagnetic environment is significantly enhanced.

[0045] Dynamic expansion support: Instruction type field (1 byte) and data block structure (action-target-value) support diversified instruction definition, compatible with multiple types of sensors and actuators, and adapt to dynamic needs in high-risk tasks.

[0046] (4) The bionic design of the present application improves environmental adaptability:

[0047] Neural system-like response: Referring to the information transmission mechanism of human nervous system (such as priority control and closed-loop feedback), an efficient closed loop of "perception-decision-execution" is realized to ensure the rapid response and motion precision of the robot in a complex environment (such as a region with frequent aftershocks).

[0048] Multi-sensor fusion: The upper computer integrates high-precision sensors such as laser radar to realize SLAM mapping, and combines with the middle computer to process low-frequency sensor data, forming multi-level environmental perception ability and improving the operation stability in complex terrain.

[0049] (5) The comprehensive performance advantage of the present application is significant:

[0050] Low delay and high real-time: Through information flow delay model optimization (such as compression of middle computer to lower computer delay Tmid→low), the overall response speed of the system meets the millisecond-level demand of high-risk scenes.

[0051] The advantages of the additional aspects of the present application will be partially given in the following description, partially will become obvious from the following description, or will be learned by the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0052] The drawings accompanying the specification of the present application serve to provide further understanding of the present application, and the illustrative embodiments of the present application and their descriptions serve to explain the present application, and do not constitute an improper limitation on the present application.

[0053] Figure 1It is a schematic diagram of a high-risk operation robot control system simulating the human nervous system according to an embodiment of the present application. DETAILED DESCRIPTION

[0054] The present application will be further described below in conjunction with the accompanying drawings and embodiments.

[0055] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present application. Unless otherwise indicated, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0056] It should be noted that the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit exemplary embodiments according to the present application. As used herein, the singular form is intended to include the plural form unless the context clearly indicates otherwise, and furthermore, it should be understood that when the terms "comprise" and / or "include" are used in the specification, there is a feature, step, operation, device, component, and / or combination thereof.

[0057] In the present application, the terms such as "upper", "lower", "left", "right", "front", "back", "vertical", "horizontal", "side", "bottom", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, which is only a relationship word determined for the convenience of describing the structural relationship of the components or elements of the present application, and cannot be understood as a limitation on the present application.

[0058] In the present application, the terms such as "fixedly connected", "connected", "connected" should be understood broadly, which means that it can be fixedly connected, integrally connected or detachably connected; it can be directly connected or indirectly connected through an intermediate medium. For related researchers or technicians in the art, the specific meaning of the above terms in the present application can be determined according to the specific circumstances, and cannot be understood as a limitation on the present application.

[0059] Term explanation:

[0060] Host computer: The top core unit of the control system, responsible for global task planning, intelligent decision-making (such as calling large language model API) and interaction with high-precision sensors (such as laser radar).

[0061] Middle machine: Intermediate layer processing unit, responsible for instruction decomposition, data integration and lightweight computing tasks.

[0062] Lower machine: Bottom execution unit, directly controls the motion components (such as motors, mechanical arms) and low-precision, low-frequency sensors.

[0063] Universal instruction format: Standardized communication protocol for instruction transmission within the system.

[0064] Variable-length encoding: encoding method that dynamically allocates storage space according to data value range.

[0065] CRC-16 check: a type of Cyclic Redundancy Check used to detect errors in data transmission.

[0066] STM32F103C8T6: microcontroller based on ARM Cortex-M3 core, used for high real-time motion control.

[0067] STM32F407: high-performance ARM Cortex-M4 microcontroller, supporting lightweight computing and multitasking.

[0068] Raspberry Pi 4B: single-board computer based on ARM architecture, used as system control core.

[0069] SLAM: Simultaneous Localization and Mapping technology, which realizes dynamic environment modeling through sensors such as laser radar.

[0070] Large language model API: intelligent interaction interface based on natural language processing (NLP), supporting semantic understanding and decision generation.

[0071] Priority control logic: mechanism to ensure that host computer instructions take precedence over mid-computer autonomous instructions.

[0072] Information delay model: quantifies the delay parameters of data transmission between different levels of the system, used to optimize real-time performance.

[0073] Dynamic scalability: system supports flexible addition of sensors or actuators.

[0074] As shown in Figure 1 The application provides a high-risk operation robot control system that simulates the human nervous system, which includes a sensing element, a lower computer, a mid-computer, and an upper computer.

[0075] The sensing element is used to perceive robot state information and transmit it to the corresponding lower computer; the lower computer is used to transmit the robot state information it receives to the upper computer through the mid-computer; the upper computer is used to form instruction information according to the high-risk operation preset target and the received robot state information, and issue it to the corresponding lower computer through the mid-computer; the lower computer is also used to control the corresponding motion components to execute the corresponding instruction information, while feeding back the received robot state information to the upper computer through the mid-computer, forming a perception-decision-execution closed-loop control.

[0076] In the embodiment of the application, the sensing element is a sensor, wherein the upper computer is connected with most of the sensors and a small part of the moving parts playing an auxiliary role, and is the control core of the whole system. The upper computer uses the API of a large language model, and the whole system is more intelligent. The control level of the upper computer to the lower computer is higher than that of the middle computer. In the designed robot model, the lower computer uses stm32f103c8t6 (hereinafter referred to as f103), the middle computer uses stm32f407 (hereinafter referred to as f407), and the upper computer uses Raspberry Pi 4B. Two f103 control a six-degree-of-freedom robot arm, are connected with f407, f407 controls four motors to control the movement of the robot, and are connected with Raspberry Pi 4B, and Raspberry Pi is connected with a laser radar to realize slam mapping.

[0077] In the embodiment, the general instruction format communication is used between the lower computer and the middle computer and between the middle computer and the upper computer; the general instruction format is: [frame header] [control flag] [instruction type] [data area] [check code] [frame tail].

[0078] Table 1 is the field details

[0079]

[0080] The data area is composed of multiple data blocks, and the format of each data block is: [action] [target] [value]. The value is variable-length encoded according to its range and type.

[0081] In the data block, if the value range is ≤255, 1 byte is used; if the value range is ≤65535, 2 bytes are used; if it is a floating-point number, 4 bytes (such as IEEE 754 standard) are used.

[0082] In the data block, the action is 1 byte, the high 2 bits represent the action type (such as moving = 00, rotating = 01), and the low 6 bits are the action ID.

[0083] The target is 1 byte, which is used to represent the target device or joint number, and its range is 0~255.

[0084] Performance index calculation:

[0085] ① Total length formula:

[0086]

[0087] Effective data:

[0088] Frame header: 2 bytes, indicating the starting position of the data frame.

[0089] Control flag: 1 byte, used to transmit control information (such as priority, retransmission flag, etc.).

[0090] Instruction type: 1 byte, indicating the type of data frame or operation instruction (such as read, write, response, etc.).

[0091] Data block overhead: 2 bytes added to each data block, usually used to store length or identification information.

[0092] L data: Length of valid data in a single data block (unit: byte).

[0093] N: Number of data blocks.

[0094] Check code: 2 bytes, used for error detection (such as CRC check).

[0095] Frame tail: 2 bytes, indicating the end position of the data frame.

[0096] Fixed overhead: total 8 bytes (frame header 2 + control flag 1 + instruction type 1 + check code 2 + frame tail 2)

[0097] ② Transmission efficiency:

[0098]

[0099] Molecule: The sum of the valid data and additional overhead of all data blocks (i.e. the "effective part" of the actual transmission).

[0100] Denominator: Total frame length (effective part + fixed overhead).

[0101] Physical meaning: Measure the proportion of effective data in total transmission data in the protocol, the value closer to 1 the higher the efficiency.

[0102] ③ Information density:

[0103]

[0104] Parameter physical meaning is consistent with transmission efficiency, here it may represent the proportion of effective data per unit length,

[0105] ④ Reliability

[0106] Use CRC-16 check: error detection rate 99.998%;

[0107] Double frame header / frame tail: reduces packet loss rate (reduces packet loss rate by 50% in noisy environment);

[0108] In the embodiment of the present application, the model of the host computer is:

[0109] ;

[0110] Among them, High-level instructions; S high Perception data for direct interaction with the host computer; API LLM Interaction results for large language model APIs; S MLC Feedback data for integration by the mid-level computer.

[0111] The model of the mid-level computer includes: instruction decomposition model, data integration model and lightweight computing model;

[0112] The instruction decomposition model is: ;

[0113] C high : Host computer instructions, representing complex instructions or complex control commands received by the system (such as "start full process monitoring").

[0114] F(): Decomposition function, responsible for decomposing high-level instructions into multiple independent executable sub-instructions.

[0115] C high,i : Sub-instructions of the host computer after decomposition (such as "sensor initialization", "data acquisition start", etc.).

[0116] n: Number of sub-instructions, depends on the complexity of high-level instructions and the division of system execution units.

[0117] The data integration model is: ;

[0118] D low,i : Lower computer data, representing raw data from different sources (such as sensors, subsystems) (such as temperature values, voltage signals).

[0119] G MCC : Integration function, responsible for fusing, filtering or structuring multiple source data (such as removing noise, time alignment).

[0120] D mid : Mid-level data, structured data set after integration (such as time series data table, feature vector).

[0121] S MCC : Effective data subset, representing key data that meets certain conditions after integration (such as anomaly detection results, key indicators).

[0122] The lightweight computing model is: ;

[0123] G ` MCC : Lightweight computing function, using simplified algorithms or optimization strategies (such as approximate calculation, dimensionality reduction processing)

[0124] D ` mid : lightweighted middle layer data, which may sacrifice part of accuracy to improve efficiency (e.g. compressed feature representation)

[0125] wherein, is a lower machine sub-instruction set; F is an instruction decomposition function;

[0126] For example:

[0127] def F(C_high):

[0128] # Predefined instruction template library

[0129] templates = {

[0130] "Monitoring category": [

[0131] "Initialize sensors (IDs: {sensor_ids})",

[0132] "Set data collection frequency to {frequency}",

[0133] "Enable real-time data streaming to {target}"

[0134] ],

[0135] "Control category": [

[0136] "Perform device calibration (mode: {mode})",

[0137] "Set threshold range: {threshold}"

[0138] ]

[0139] }

[0140] # Step 1: Parse instruction type and parameters

[0141] if "Full-process monitoring" in C_high:

[0142] instruction_type = "Monitoring category"

[0143] params = {

[0144] "sensor_ids": "S001-S005",

[0145] "frequency": "10Hz",

[0146] "target": "Cloud"

[0147] }

[0148] else:

[0149] raise ValueError("Unknown instruction type")

[0150] # Step 2: Generate sub-instructions from templates

[0151] sub_commands = []

[0152] for template in templates[instruction_type]:

[0153] sub_command = template.format(**params)

[0154] sub_commands.append(sub_command)

[0155] return sub_commands

[0156] D mid For middle layer data; S MLC For backhaul data integrated by the middle layer machine; For lower machine feedback data; For lightweight processed lower machine feedback data;G MCC For integrated function; For lightweight calculation function.

[0157] For example:

[0158] import numpy as np

[0159] def G_prime_MCC(*D_low):

[0160] # Define the lightweight processing function phi_i

[0161] def sliding_window_avg(data, window_size=100):

[0162] # Sliding window average: compress L data points into L / / window_size points

[0163] compressed = []

[0164] for i in range(0, len(data), window_size):

[0165] window = data[i:i+window_size]

[0166] compressed.append(np.mean(window))

[0167] return compressed

[0168] # Define feature extraction function (example)

[0169] def extract_features(compressed_data):

[0170] features = {

[0171] "Mean": np.mean(compressed_data),

[0172] "Variance": np.var(compressed_data),

[0173] "Peak": np.max(compressed_data)

[0174] }

[0175] return features

[0176] # Process each low-level data sequentially

[0177] D_mid_prime = {}

[0178] for idx, data in enumerate(D_low):

[0179] # Step 1: Sliding window compression

[0180] compressed = sliding_window_avg(data)

[0181] # Step 2: Extract key features

[0182] features = extract_features(compressed)

[0183] # Label data sources (e.g., temperature, humidity)

[0184] D_mid_prime[f"Sensor_{idx+1}"] = features

[0185] return D_mid_prime

[0186] The lower computer model comprises a control signal generation model and a data preprocessing model.

[0187] The control signal generation model is: ;

[0188] The data preprocessing model is: ;

[0189] Wherein, C low,i is a lower computer instruction; S low,i is a sensing data; U i is a control signal; D low,i is a lower computer preprocessing data; is a control signal generation function; is a data preprocessing function.

[0190] Taking a PID control signal as an example:

[0191] class PIDController:

[0192] def __init__(self, Kp, Ki, Kd):

[0193] self.Kp = Kp

[0194] self.Ki = Ki

[0195] self.Kd = Kd

[0196] self.integral = 0.0

[0197] self.prev_error = 0.0

[0198] def compute(self, error, dt):

[0199] self.integral += error * dt

[0200] derivative = (error - self.prev_error) / dt

[0201] output = self.Kp * error + self.Ki * self.integral + self.Kd * derivative

[0202] self.prev_error = error

[0203] return output

[0204] # Example: Control the i-th joint (e.g., i=1 for the knee joint)

[0205] f_LLC_i = PIDController(Kp=2.0, Ki=0.1, Kd=0.5)

[0206] current_angle = 30.0# Current angle (from sensors)

[0207] target_angle = 45.0# Target angle (from high-level commands)

[0208] error = target_angle - current_angle

[0209] control_signal = f_LLC_i.compute(error, dt=0.1)# dt is the sampling period

[0210] With the example of high-frequency noise removal preprocessing:

[0211] import numpy as np

[0212] from scipy.signal import butter, lfilter

[0213] def g_LLC(D_raw, fc=10.0, fs=100.0):

[0214] # Step 1: Low-pass filtering (Butterworth filter)

[0215] b, a = butter(4, fc / (0.5 * fs), 'low')

[0216] D_filtered = lfilter(b, a, D_raw)

[0217] # Step 2: Normalize to [-1, 1]

[0218] D_max = np.max(np.abs(D_filtered))

[0219] D_normalized = D_filtered / (D_max + 1e-6)# Avoid division by zero

[0220] return D_normalized

[0221] # Example: Process raw joint angular velocity data

[0222] D_raw = np.random.normal(0, 5, 1000)# Raw data with noise (mean=0, std= 5）

[0223] D_clean = g_LLC(D_raw, fc=5.0, fs=100.0)

[0224] As an embodiment, the instruction priority of the upper computer is higher than that of the middle computer;

[0225] The instruction actually executed by the lower computer Decided by the following logic:

[0226] ;

[0227] Wherein, The upper computer sub-instruction; The middle computer sub-instruction.

[0228] Specifically, the delay of the information transmission process between the lower computer and the middle computer and between the middle computer and the upper computer is optimized by using an information flow delay model:

[0229] Upper computer→middle computer: delay T high→mid : ;

[0230] Middle computer→lower computer: delay T mid→low : ;

[0231] Lower computer→middle computer: delay T low→mid : ;

[0232] Middle computer→upper computer: delay T mid→high : .

[0233] T high→mid : the delay of the upper computer sending instructions to the middle computer;

[0234] T mid→low : the delay of the middle computer sending instructions to the lower computer;

[0235] T low→mid : the delay of the lower computer returning data to the middle computer;

[0236] T mid→high : the delay of the data transmission from the lower machine to the upper machine;

[0237] C low,i : the instruction for the lower machine; : the instruction for the upper machine;D low,i : the data for the lower machine; : the data for the upper machine.

[0238] The overall information flow of the system can be represented by the following recursive relationship:

[0239] Upper machine output information flow:

[0240]

[0241] C (k) high : global target or system configuration parameters (such as total task target, operation mode), for example, "complete environmental monitoring task".

[0242] G: the state of the upper machine at time k or high-level sensor feedback (such as global positioning state, system health status).

[0243] S (k) high : the middle-level data fed back after processing by the lower machine in the last period k-1 (such as integrated environmental features, historical records).

[0244] D (k-1) mid : the interaction interface with large language models, used to generate or optimize high-level instructions (such as natural language instruction analysis).

[0245] API LLM : high-level control functions, which generate high-level instructions by integrating target, state, historical data and external models.

[0246] Lower machine processing information flow:

[0247]

[0248] C (k) high,i : high-level instructions output by the upper machine (such as "start full-process monitoring");

[0249] C (k) high : the i-th sub-instruction (such as "initialize sensor" "set collection frequency");

[0250] Lower machine execution information flow:

[0251]

[0252] U (k) i : The execution action of the lower machine k (such as motor driving signal, valve opening degree).

[0253] C (k) low,i : The low-level instruction assigned to the lower machine i by the middle machine (such as "read sensor S001 data").

[0254] S (k) low,i : The real-time state or sensor data of the lower machine i (such as current motor speed, temperature value).

[0255] Data feedback and update:

[0256]

[0257] D (k) mid : The integrated middle-layer data (such as structured feature table, abnormality detection result).

[0258] D (k) low,n : The original data of the lower machine n at time k (such as sensor raw sampling point).

[0259] In one or more embodiments, a high-risk operation robot control method simulating a human nervous system based on the high-risk operation robot control system simulating a human nervous system as described above; comprising:

[0260] The sensing element perceives the robot state information and transmits it to the corresponding lower machine; the lower machine transmits the received robot state information to the upper machine through the middle machine; the upper machine forms instruction information according to the high-risk operation preset target and combines the received robot state information, and then issues it to the corresponding lower machine through the middle machine; the lower machine controls the corresponding motion component to execute the corresponding instruction information, and at the same time, transmits the received robot state information to the upper machine through the middle machine, forming a perception-decision-execution closed-loop control;

[0261] Among them, the communication between the lower machine and the middle machine and the middle machine and the upper machine adopts a general instruction format; the general instruction format is: [frame header] [control flag] [instruction type] [data area] [check code] [frame tail].

[0262] The hardware selection of the present application takes into account the computing power and real-time performance:

[0263] Computing power balance: the host computer uses Raspberry Pi 4B to provide higher general computing power, supports interaction with large language models; the middle machine uses STM32F407 for lightweight computing to avoid waste of host computer resources; the lower machine uses STM32F103 to realize high real-time motion control. The three achieve the best balance among cost, computing power and real-time performance.

[0264] Cost controllable: the hardware selection is based on mature embedded platforms (such as STM32 series and Raspberry Pi), which greatly reduces the development and deployment cost while ensuring performance.

[0265] The above only describes the preferred embodiments of the present application and is not used to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A control system for a high-risk work robot that mimics the human nervous system, characterized in that, include: Sensing elements, lower-level machines, middle-level machines, and upper-level machines; The sensing element is used to sense the robot's status information and transmit it to the corresponding lower-level machine; The lower-level machine is used to transmit the robot status information it receives to the upper-level machine via the middle-level machine; the upper-level machine is used to form instruction information based on the preset target of high-risk operation and the received robot status information, and send it to the corresponding lower-level machine via the middle-level machine; the lower-level machine is also used to control the corresponding moving parts to execute the corresponding instruction information, and at the same time, it feeds back the robot status information it receives to the upper-level machine via the middle-level machine, forming a perception-decision-execution closed-loop control; The communication between the lower-level machine and the middle-level machine, as well as between the middle-level machine and the upper-level machine, uses a common command format. The common command format is: [frame header][control flags][command type][data area][checksum][frame tail]. The host computer model is as follows: ； in, These are instructions from the host computer. S high Sensing data that interacts directly with the host computer; API LLM This is the result of interaction with the large language model API; S MLC The data transmitted back to the mid-level computer is integrated; G represents the status of the host computer at time k or the feedback from higher-level sensors. The models of mid-level machines include: instruction decomposition model, data integration model, and lightweight computing model; The instruction decomposition model is as follows: ; The data integration model is as follows: ; The lightweight computational model is as follows: ; Among them, C high,i F represents the decomposed host computer sub-instructions; F is the instruction decomposition function. D mid This is mid-level data; S MLC Data returned by the mid-level unit; D low,i For lower-level machine data; This is the lower-level machine feedback data after lightweight processing; For integration functions; For lightweight computation functions; The lower-level machine model includes: a control signal generation model and a data preprocessing model; The control signal generation model is as follows: ; Data preprocessing model: ; Among them, C low,i For lower-level machine instructions; S low,i For sensor data; U i For control signals; D low,i Preprocess data for the lower-level machine; This is the function for generating control signals; For data preprocessing functions; The latency of information transmission between the lower-level machine and the middle-level machine, as well as between the middle-level machine and the upper-level machine, is optimized using an information flow latency model: The information flow delay model from the host computer to the intermediate computer is as follows: ; The information flow delay model from the mid-level machine to the lower-level machine is as follows: ; The information flow delay model from lower-level machine to middle-level machine is as follows: ; The information flow delay model from the intermediate computer to the host computer is as follows: ; Among them, T high→mid T represents the delay for the host computer to send instructions to the intermediate computer. mid→low The delay for the intermediate unit to send instructions to the lower unit; T low→mid T represents the delay in data transmission from the lower-level machine to the middle-level machine. mid→high The delay for the intermediate computer to transmit data back to the host computer; This is data from the host computer.

2. The high-risk operation robot control system that mimics the human nervous system as described in claim 1, characterized in that, The data area consists of multiple data blocks, each with the format: [action][target][value]; the value is encoded with variable length according to its range and type.

3. The high-risk operation robot control system that mimics the human nervous system as described in claim 2, characterized in that, Within a data block, if the numerical value is ≤255, use 1 byte; if 255 < the numerical value is ≤65535, use 2 bytes; if it is a floating-point number, use 4 bytes.

4. The high-risk operation robot control system that mimics the human nervous system as described in claim 1, characterized in that, In the data block, the action is 1 byte, with the high 2 bits indicating the action type and the low 6 bits being the action ID; Or / and in the data block, the target is 1 byte, used to represent the target device or joint number, which ranges from 0 to 255.

5. The high-risk operation robot control system that mimics the human nervous system as described in claim 1, characterized in that, Instructions from the host computer have higher priority than those from the intermediate computer; the instructions actually executed by the slave computer are... It is determined by the following logic: ； in, These are instructions from the host computer. This is a sub-instruction for the middle unit.

6. A control method for high-risk work robots that mimics the human nervous system, characterized in that, A high-risk operation robot control system based on any one of claims 1-5 that mimics the human nervous system; comprising: The sensing element perceives the robot's status information and transmits it to the corresponding lower-level machine; the lower-level machine transmits the received robot status information to the upper-level machine via the middle-level machine; the upper-level machine, based on the preset target of high-risk operation, combines the received robot status information to form instruction information and sends it to the corresponding lower-level machine via the middle-level machine; the lower-level machine controls the corresponding moving parts to execute the corresponding instruction information, and at the same time feeds back the received robot status information to the upper-level machine via the middle-level machine, forming a perception-decision-execution closed-loop control; The communication between the lower-level machine and the middle-level machine, as well as between the middle-level machine and the upper-level machine, uses a common instruction format. The common instruction format is: [frame header][control flag][instruction type][data area][checksum][frame tail].

Images