Port mobile intelligent sentinel multi-source information equipment system

Abstract

The invention provides a port mobile intelligent sentinel multi-source information equipment system, which is realized on a mobile intelligent robot and comprises a multi-source information acquisition platform, a portable experimental platform and a one-stop management platform. The multi-source information acquisition platform is used for acquiring multi-source information data of to-be-quarantine customs port imported goods; the portable experimental platform is used for identifying and sampling goods to be quarantine imported from customs ports to obtain identification and sampling results; and the one-stop management platform is used for carrying out data processing based on the multi-source information data and the identification sampling result, and obtaining a quarantine result report of the to-be-quarantine customs port imported goods. The multi-source information acquisition platform carries image sensors, chemical sensors and the like with different wave bands, and different types of information are transmitted and integrated through a multi-source information transmission technology and a data integration method; and the remote expert system realizes real-time sharing of equipment information so as to obtain professional guidance.

Description

Technical Field

[0001] The present invention belongs to the technical field of inspection of alien invasive species at customs ports, and in particular relates to a mobile intelligent outpost multi-source information equipment system at ports. Background Art

[0002] Traditional customs quarantine methods for invasive alien species primarily rely on instruments to examine physical indicators like wood density and moisture content to determine if the wood meets quality requirements. Alternatively, biological testing can be used to detect the presence of pests and pathogens to identify potential safety hazards. However, these traditional methods have limitations. The quarantine process can damage the quality and appearance of the wood, and some processes involving contamination must be conducted under conditions that ensure the personal safety of workers.

[0003] Intelligent quarantine can significantly improve the efficiency of port quarantine, save the average time ships spend in port per voyage, save operating costs, reduce the risk of "contact" cross-infection between incoming personnel and front-line customs officers, use digital technology to enhance the dynamic perception capability of port operations, realize a "one-picture" panoramic operation management of ports, and improve the digital level of port operation management.

[0004] Therefore, how to conduct intelligent quarantine of alien invasive species at customs ports and observe the disease and pest conditions of wood in real time is a technical problem that needs to be solved urgently. Summary of the Invention

[0005] To solve the problems existing in the prior art, the present invention provides a mobile intelligent outpost multi-source information equipment system for ports, which aims to observe the disease and insect pest conditions of wood in real time. By capturing high-definition image data, including using an endoscope to deeply inspect the internal conditions of the wood, it is possible to build an intelligent port and an information data integrated port.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A multi-source information equipment system for a mobile intelligent outpost at a port. The system is implemented on an intelligent outpost robot and includes: a multi-source information collection platform, a portable experimental platform, and a one-stop management platform;

[0008] The multi-source information collection platform is used to collect multi-source information data of imported goods to be quarantined at customs ports;

[0009] The portable experimental platform is used to identify and sample the imported goods to be quarantined at the customs port and obtain identification and sampling results;

[0010] The one-stop management platform is used to perform data processing based on the multi-source information data and the identification sampling results to obtain a quarantine result report for imported goods to be quarantined at the customs port.

[0011] Preferably, the multi-source information acquisition platform includes: an acquisition unit, a power supply unit and a data processing unit;

[0012] The acquisition unit is used to collect multi-source information data of the imported goods to be quarantined at the customs port based on visible light sensors, infrared sensors, near-infrared sensors, spectral sensors and chemical sensors;

[0013] The power supply unit is used to provide power to the sensor unit;

[0014] The data processing unit is used to pre-process the multi-source information data.

[0015] Preferably, the portable experimental platform comprises: an observation lens, a robotic arm, a disposal box and a wood identification device;

[0016] The observation lens is used to perform path planning and navigation obstacle avoidance for the intelligent outpost robot;

[0017] The robotic arm is used to identify and sample the imported goods to be quarantined at the customs port;

[0018] The disposal box is used to store and polish sampling tools;

[0019] The wood identification device is used to identify the wood species in the imported goods to be quarantined at the customs port.

[0020] Preferably, the one-stop management platform includes: a data transmission unit, a data integration unit, an image recognition unit and a remote assistance unit;

[0021] The data transmission unit is used to transmit the multi-source information data to a control room for real-time monitoring;

[0022] The data integration unit is used to construct a multi-dimensional feature database based on the multi-source information data;

[0023] The image recognition unit is configured to recognize real-time images in the multi-source information data based on an intelligent image recognition algorithm, and obtain a risk assessment index in combination with odor data in the multi-source information data;

[0024] The remote assistance unit is used for remote guidance.

[0025] Preferably, the data integration unit is used to construct a multidimensional feature database based on the multi-source information data, including:

[0026] Extracting color features, texture features, and shape features of the goods based on the data collected by the visible light sensor, and generating a color feature vector, a texture feature vector, and a shape feature vector;

[0027] Extracting temperature distribution features based on data collected by the infrared sensor and generating a temperature distribution feature vector;

[0028] Extracting the spectral absorption characteristics of the substance based on the data collected by the near-infrared sensor and generating a spectral absorption characteristic vector;

[0029] Based on the data collected by the spectral sensor, spectral features are extracted and spectral feature vectors are generated;

[0030] Based on the data collected by the chemical sensor, chemical composition characteristics are extracted and chemical composition feature vectors are generated;

[0031] Data fusion is performed based on the color feature vector, the texture feature vector, the shape feature vector, the temperature distribution feature vector, the spectral absorption feature vector, the spectral feature vector and the chemical composition feature vector to construct a multi-dimensional feature database.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] This invention addresses the need for multi-source information integration in mobile intelligent outposts and proposes a novel data integration solution that organically integrates different types of information to provide comprehensive support for subsequent analysis and decision-making. By developing an iteratively updateable image recognition algorithm, it enables rapid identification of invasive organisms and other risk factors, and integrates odor data for risk assessment. An insect suction device and a pan-tilt camera integrated into the robotic arm capture and identify the interior of wood. Furthermore, a remote expert assistance system is proposed, enabling real-time sharing of information collected by equipment and providing professional guidance. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0035] Figure 1 This is a schematic diagram of the composition of a multi-source information equipment system for a mobile intelligent outpost at a port according to an embodiment of the present invention;

[0036] Figure 2 This is a schematic diagram of the working principle structure of the multi-source information acquisition platform according to an embodiment of the present invention;

[0037] Figure 3 This is a schematic diagram of the working principle structure of the portable experimental platform according to an embodiment of the present invention;

[0038] Figure 4This is a schematic diagram of the working principle structure of the one-stop management platform according to an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0040] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0041] Example 1

[0042] The present invention provides a multi-source information equipment system for a mobile intelligent outpost at a port. The system is implemented on an intelligent outpost robot and includes: a multi-source information collection platform, a portable experimental platform, and a one-stop management platform.

[0043] Multi-source information collection platform, used to collect multi-source information data on imported goods to be quarantined at customs ports;

[0044] A portable experimental platform for identifying and sampling imported goods at customs ports undergoing quarantine and obtaining identification and sampling results;

[0045] A one-stop management platform for data processing based on multi-source information data and identification sampling results, and obtaining quarantine result reports for imported goods at customs ports awaiting quarantine.

[0046] Further, such as Figure 1 As shown, the specific implementation process of the present invention is as follows:

[0047] (1) Multi-source information collection platform

[0048] The platform's main function is to integrate five modules, namely visible light sensors, infrared sensors, near-infrared sensors, spectral sensors, and chemical sensors, into a multi-source information collection platform to achieve quarantine of port goods, including wood, soybeans, grains, weeds, insect viruses, etc.

[0049] 1. Integrated solution

[0050] (1) Hardware Integration

[0051] ① Physical Structure Design: Design a compact and robust housing to strategically arrange and install the five sensor modules. The housing must be well sealed and resistant to interference to withstand the complex environment of port cargo quarantine, such as dust, humidity, and electromagnetic interference. For example, a multi-layer shielding design can be used to ensure that the signals between sensor modules do not interfere with each other.

[0052] ②Power supply system: Design a unified power supply circuit based on the power consumption and voltage requirements of each sensor module. Use a voltage- and current-stabilized power supply module to provide a stable and reliable power supply for the sensors, and consider power supply redundancy to cope with emergencies.

[0053] ③Data transmission interface: Each sensor module is equipped with a suitable communication interface, such as I2C, SPI, UART, etc., to transmit the collected data to the central processing unit. At the same time, an external communication interface, such as an Ethernet interface or a wireless communication module interface, is reserved for sending data to a remote server or monitoring center.

[0054] (2) Software Integration

[0055] ① Data acquisition and preprocessing: Develop a data acquisition program to read data from each sensor module in real time. Perform unified preprocessing operations, such as filtering and normalization, on data formats and dimensions of different sensors to eliminate discrepancies and facilitate subsequent fusion processing.

[0056] ② Feature extraction and fusion: To meet the needs of port cargo quarantine, key features are extracted from the data collected by various sensors. For example, color and texture features of cargo can be extracted from visible light sensor data; temperature distribution features can be extracted from infrared sensor data; and spectral absorption features of substances can be extracted from near-infrared sensor data. These features are then fused using mid- or post-fusion methods to generate a comprehensive feature vector for subsequent quarantine analysis.

[0057] ③ Quarantine Analysis Algorithm: Establish corresponding quarantine analysis models based on the characteristics of different cargoes, such as timber, soybeans, grains, weeds, insects, and viruses. Machine learning or deep learning algorithms, such as support vector machines and convolutional neural networks, can be used to classify, identify, and detect the fused feature vectors to determine whether the cargo meets quarantine requirements.

[0058] 2. Application scenarios and specific operations

[0059] (1) Wood quarantine: Use the visible light sensor module to capture the appearance of wood, extract features such as color and texture, and identify the type and defects of wood; the infrared sensor module detects the temperature distribution of wood to determine whether there are hidden insect pests or diseased areas; the near-infrared sensor module analyzes the chemical composition of wood, such as moisture content and lignin content, to assess the quality and health of wood; the spectral sensor module obtains the spectral information of wood to further confirm the type and origin of wood; the chemical sensor module detects chemical residues on the surface or inside of wood, such as preservatives and pesticides, to ensure the safety of wood.

[0060] (2) Quarantine of soybeans and grains: The visible light sensor module is used to observe the color, shape and impurities of soybeans and grains; the infrared sensor module measures the temperature and humidity of grains to determine whether they are moldy or hot; the near-infrared sensor module detects the content of grain ingredients, such as protein, starch, fat, etc., to evaluate their nutritional value and quality; the spectral sensor module analyzes the spectral characteristics of grains to identify grains of different varieties and origins; the chemical sensor module detects pesticide residues, heavy metal content and other harmful substances in grains to ensure food safety.

[0061] (3) Weed quarantine: The visible light sensor module captures the morphological images of weeds and determines the type of weeds through image recognition technology; the infrared sensor module monitors the temperature of the weed growth environment and analyzes its growth status; the near-infrared sensor module obtains the spectral information of weeds to further assist in species identification and growth stage judgment; the spectral sensor module can be used to detect physiological indicators such as the photosynthesis efficiency of weed leaves; the chemical sensor module detects chemical substances in or around weeds, such as herbicide residues and volatile organic compounds, and evaluates their impact on the ecological environment.

[0062] (4) Insect virus quarantine: The visible light sensor module can be used to observe the appearance and behavioral characteristics of insects; the infrared sensor module monitors the changes in the insect's body temperature to determine whether it is infected with a virus or in a state of stress; the near-infrared sensor module analyzes the changes in the chemical composition of the insect body, such as changes in the content of biological macromolecules such as proteins and nucleic acids, providing clues for virus detection; the spectral sensor module obtains the spectral information of the insect to distinguish different types of insects and virus types; the chemical sensor module directly detects biomarkers such as viral nucleic acids and proteins in the insect body to achieve rapid and accurate detection of insect viruses.

[0063] 3. System testing and optimization

[0064] (1) Laboratory testing: In a laboratory environment, the integrated multi-source information acquisition platform is functionally tested and its performance evaluated. Samples of wood, soybeans, grains, weeds, insect viruses, and other materials are tested to verify the data acquisition accuracy of each sensor module, the effectiveness of the fusion algorithm, and the reliability of the quarantine analysis model. Test data and results are recorded, and existing problems are analyzed and optimized.

[0065] (2) Field testing: The optimized system was installed at the port cargo quarantine site for actual operational testing. Under different environmental conditions, a large number of actual cargoes were quarantined and tested, and compared with traditional quarantine methods to evaluate the system's accuracy, detection efficiency, and stability. Based on the field test results, the system's parameters and algorithms were further adjusted and optimized to ensure stable and efficient operation in actual applications.

[0066] (2) Portable experimental platform

[0067] 1. Shared field of view observation lens

[0068] Given the characteristics of port cruising, which typically features large open areas with excellent satellite signal reception, a relatively stable environment, and minimal route changes, the Beidou satellite navigation system was selected for autonomous driving. Beidou satellite navigation provides relatively stable and accurate positioning information, ensuring the cruising vehicle adheres to its intended route. The robot is also equipped with a time-of-flight camera and infrared or near-infrared image sensors. The time-of-flight camera lens helps the robot perceive its surroundings, enabling autonomous navigation and obstacle avoidance, while the infrared or near-infrared system enables nighttime navigation. These lenses provide more precise navigation and obstacle avoidance for the robot's operations in the port.

[0069] 2. Intelligent sampling:

[0070] The robotic arm utilizes a smart toolbox for sampling and polishing, and a smart sampling device is designed to intelligently select the most appropriate sampling strategy based on identification and assessment results. A vacuum cleaner duct is integrated into the robotic arm to form an insect extractor, which retracts and extends accordingly. For tropical wood, a camera is installed at the tip of the robotic arm, using a pan-tilt system to achieve 360° rotation, allowing observation of termite tunnels to assess termite infestation and ultimately locate ant nests. The primary scenario involves capturing and identifying termites in wood.

[0071] 3. Quick disposal box:

[0072] (1) Tools for sampling wood and wood packaging: including hardware tools such as pointed axes, quarantine chisels, sampling shovels, needle-nose pliers, etc., electric and electronic tools such as high-intensity flashlights, magnifying glasses with light sources, pin-type electronic thermometers, etc., sampling and recording tools such as brushes, oil-based markers, and signing pens, etc., and other tools such as labor protection gloves, tape measures, and white porcelain plates, etc.

[0073] (2) Grain sampling tools: equipped with customized sample splitters and bags, customized sieve sets, sampling shovels, tweezers, folding knives, scissors and other tools, as well as electric and electronic tools such as high-intensity flashlights, magnifying glasses with light sources, pin-type electronic thermometers, electronic scales, counters, etc.

[0074] (3) Chemical sensors and data acquisition equipment: Chemical sensors can collect data using data acquisition instruments. Data acquisition instruments can collect signals emitted by sensors to detect chemical substances in the external environment.

[0075] (4) Rapid drug identification equipment: Handheld Raman system that can quickly identify suspected drugs, precursor chemicals and diluents.

[0076] (5) High-throughput sequencing technology equipment: used for vector monitoring, which can quickly and accurately identify the species and related species of imported vectors, and reduce the chance of the spread of vectors themselves and the pathogens they carry.

[0077] 4. Wood identification device: captures wood cross-section texture information and identifies wood species

[0078] (1) Image acquisition:

[0079] ① Use high-precision image recognition technology to obtain characteristic information such as size, shape, and texture of wood.

[0080] ② Use the wood structure image acquisition device to capture the fine structural images of the wood cross section.

[0081] (2) Feature extraction:

[0082] ① Apply image processing technology to extract the color, grayscale, texture and other contents of wood images, including characteristic parameters such as hue, saturation, brightness, contrast, second-order angular moment, sum of variance, long-stroke weighting factor, fractal dimension, and wavelet horizontal energy ratio.

[0083] ② A wood image feature extraction method based on image segmentation extracts the wood tissue contour and then uses mathematical morphology to smooth the object contour.

[0084] (3) Deep learning model training:

[0085] ① Build a deep convolutional neural network to train and learn image big data and automatically extract wood image features.

[0086] (4) Identification of wood species:

[0087] ① Input the extracted features into the wood recognition model, which may be based on kernel principal component analysis and adaptive enhancement algorithm for wood species identification.

[0088] ② Based on the model's identification of the spectrum, the wood species is finally calculated.

[0089] (3) One-stop management platform

[0090] 1. Real-time data transmission:

[0091] This process is transmitted to the control room via the 5G network, where the checkpoint status is displayed in real time on the terminal's large screen. Real-time data transmission technology is increasingly being used in modern industry and forestry, particularly in areas such as wood sample collection and wood species identification. 5G networks, with their high speed, low latency, and high reliability, provide strong technical support for these applications.

[0092] (1) Data collection

[0093] ① Collection Unit: During the wood sample collection process, multiple camera acquisition modules are used to collect images and related data of the wood. This data includes key parameters such as wood size and odor, providing a basis for subsequent analysis and optimization.

[0094] (2) Data processing

[0095] ① Image processing and calculation: The collected data is processed by the image processing module, which has a basic model and image comparison function. The data processing module is responsible for calculating the input data.

[0096] (3) Data transmission

[0097] ① 5G network applications: The high-speed transmission capabilities of 5G networks enable real-time monitoring of wood processing equipment, production lines, and finished products, thereby acquiring large amounts of data. 5G technology offers peak rates of up to 10Gbps, meeting the needs of forestry product collection and data transmission.

[0098] ② Low latency: The low latency characteristic of 5G networks (less than 1 millisecond) ensures real-time transmission of data, which is crucial for application scenarios that require fast response.

[0099] (4) Data reception and application

[0100] ① Control room reception: Through the 5G network, the collected data can be transmitted to the control room in real time. The control room can monitor, analyze and manage this data in real time through a large screen.

[0101] 2. Data integration technology:

[0102] This technology organically integrates different types of information to provide comprehensive support for subsequent analysis and decision-making. The process constructs five databases based on data collected by five sensors on the multi-source information acquisition platform. Based on this data, the individual features are extracted to form five feature databases. These feature databases are then fused pairwise to form a multi-dimensional feature database.

[0103] (1) Data collection and database construction

[0104] ① Data collection: Use visible light sensor module, infrared sensor module, near-infrared sensor module, spectral sensor module, and chemical sensor module to collect data on port cargo.

[0105] ② Database construction: Build an independent database for each sensor module to store the collected raw data.

[0106] (2) Feature extraction

[0107] ① Visible light sensor: Extracts features such as color, texture, and shape. For example, edge detection algorithms are used to extract edge information from images, and color histograms are used to extract color distribution features.

[0108] ② Infrared sensor: Extracts temperature distribution features. For example, it uses a temperature gradient algorithm to extract temperature variation areas, and a region segmentation algorithm to extract high-temperature areas.

[0109] ③ Near-infrared sensor: Extract spectral absorption features. For example, principal component analysis (PCA) is used to extract the main spectral components, and linear discriminant analysis (LDA) is used to extract classification features.

[0110] ④ Spectral sensor: Extracts spectral features. For example, a spectral matching algorithm is used to extract the spectral features of a specific substance, and a spectral decomposition algorithm is used to extract the spectral components of a mixed substance.

[0111] ⑤ Chemical sensors: Extract chemical composition features. For example, chemometric methods are used to extract chemical concentration features, and sensor array response features are used to extract the features of mixed chemicals.

[0112] (3) Feature database construction

[0113] ① Feature extraction: Extract features from the data in each database and generate feature vectors.

[0114] ② Feature database construction: Build a feature database for each sensor module to store the extracted feature vectors.

[0115] (4) Feature Fusion

[0116] Pairwise feature library fusion: Fuse pairwise feature libraries to generate a multi-dimensional feature database. A weighted average method can be used to average the feature vectors of different sensors, with weights adjusted based on the reliability and importance of the sensors.

[0117] (5) Construction of multi-dimensional feature database

[0118] Multi-dimensional feature database: The fused feature vectors are stored in a multi-dimensional feature database.

[0119] 3. Intelligent image recognition algorithm

[0120] Mainly for early warning of disease symptoms of termites invading wood

[0121] (1) Feature extraction: Determine which image features are critical for identifying invasive organisms and other risk factors, and develop algorithms to extract these key features, such as color, shape, texture, etc.

[0122] (2) Model training and optimization: Select appropriate machine learning or deep learning models and use labeled data to train the models so that they can identify invasive organisms and other risk factors;

[0123] (3) Risk assessment algorithm: Develop an algorithm to convert the recognition results and odor data into risk assessment indicators, and classify different risk levels based on the risk assessment indicators.

[0124] 4. Remote expert assistance system

[0125] (1) Real-time image transmission technology: Develop image compression technology to reduce transmission delays and select appropriate image transmission protocols to ensure real-time and security of data;

[0126] (2) Remote communication interface: realize remote communication and guidance functions.

[0127] (3) Remote control and collaboration: The use of 5G technology breaks through geographical restrictions, realizes remote operation and maintenance of wood processing equipment, reduces maintenance costs, and supports a remote expert collaboration platform, allowing wood processing companies to communicate with industry experts anytime and anywhere.

[0128] The data analysis technology involved in the present invention includes:

[0129] 1. Multi-source information transmission technology: Different types of information are transmitted to the next stage through different paths, ensuring that a variety of information (such as images, smells, etc.) from the equipment set can be transmitted to the data processing unit in real time.

[0130] 2. Multi-source data integration: This approach integrates data from different data sources for analysis and processing within a single system. This typically involves extracting, cleaning, and transforming data to ensure consistency and usability.

[0131] 3. Intelligent image recognition algorithm: Using a convolutional neural network, by simulating the visual information processing mechanism of the human brain, it uses structures such as convolutional layers and pooling layers to automatically extract image features and abstract higher-level feature representations layer by layer to quickly identify invasive organisms and other risk factors.

[0132] Through the combination of the above modules and data analysis technology, efficient and accurate invasive species quarantine process and data analysis of the port's mobile intelligent outpost multi-source information equipment system can be realized.

[0133] The working principle of the present invention is as follows:

[0134] (1) Intelligent Outpost Robot Navigation and Positioning

[0135] 1. Environmental Perception

[0136] (1) Sensor configuration

[0137] ① Visual sensors: These sensors, such as cameras, capture information by capturing images of the surrounding environment. For example, a monocular camera can capture two-dimensional images and use image processing algorithms to identify features such as the outline and color of objects in the environment. A binocular camera, like the human eye, uses parallax to calculate the depth of objects, thereby constructing a three-dimensional environmental model.

[0138] ② LiDAR: It can emit laser beams and receive reflected lasers, and measure the distance to surrounding objects based on the laser's time of flight (TOF) or phase difference. LiDAR can quickly scan the surrounding environment and generate high-precision point cloud data that can clearly depict the contours of the environment and the location of obstacles. The TOF principle can be summarized as follows:

[0139]

[0140] Where d is the distance measured by laser ranging, c is the speed of light, and t is the round-trip time of the laser.

[0141] (2) Data fusion

[0142] Due to the limitations of a single sensor, such as the reduced perception of visual sensors in low-light or complex texture environments, and the potential for LiDAR to misjudge transparent or smooth objects, data from multiple sensors must be fused. Data fusion can be achieved using the Kalman filter algorithm, which is summarized in five equations:

[0143] State prediction equation: x k|k-1 =Ax k-1|k-1 +Bu k-1

[0144] Covariance prediction equation: P k|k-1 =AP k-1|k-1 A T +Q

[0145] Kalman gain: K k =P k|k-1 C T (CP k|k-1 C T +R) -1

[0146] Status update: x k|k =x k|k-1 +K k (z k -Cx k|k-1 )

[0147] Covariance update: P k|k =(IK k C)P k|k-1

[0148] In the state prediction equation, x is the state estimate, k is a certain moment, and similarly, k-1 represents the previous moment, x k|k-1 is the state prediction value at time k (based on the information of the previous time k-1); A is the state transition matrix; x k-1|k-1 is the state estimate at time k-1 (optimal estimate after correction); B is the control input matrix; u k-1 is the control input at time k-1.

[0149] In the covariance prediction equation, P is the covariance matrix, P k|k-1 is the covariance matrix of the predicted state, which represents the uncertainty of the predicted state; P k-1|k-1 is the covariance matrix of the state estimate at the previous moment; Q is the observation noise covariance matrix, which represents the uncertainty of the system model or external interference, and T is the transpose operator of the matrix.

[0150] In the Kalman gain equation, K k is the Kalman gain, which is used to weigh the weight of the predicted state and the observed value; C is the observation matrix, which describes how to get the observation value z from the state x; R is the observation noise covariance, which represents the uncertainty of the observation.

[0151] In the state update equation, x k|k is the optimal state estimate at time k (combining predictions and observations); z k is the actual observation value at time k; zk -Cx k|k-1 Represents the observation residual (the difference between the actual observed value and the predicted observed value).

[0152] In the covariance update equation, P k|k is the updated state covariance matrix, which represents the uncertainty of the optimal estimate; I is the identity matrix.

[0153] When a robot navigates in an indoor environment, it can more accurately determine its own position and the state of the surrounding environment by fusing data from lidar and vision sensors through Kalman filtering, thereby improving the reliability of navigation.

[0154] 2. Map construction

[0155] (1) Static map construction

[0156] In some known or relatively stable environments, static maps can be constructed in advance. In the fixed work environments of certain ports, the robot's work area is fixed. Precise measurements of the workshop layout and equipment locations can be taken manually or using specialized equipment (3D scanners). This data is then converted into map information and stored in the robot's control system. The static map includes the location and dimensions of fixed obstacles such as the ground environment, workbenches, and shelves. The robot can directly use the map for path planning when performing tasks.

[0157] (2) Dynamic Map Construction (SLAM)

[0158] In unknown or dynamically changing environments, robots need to build maps as they move. This is the key to simultaneous localization and mapping (SLAM) technology. The core of SLAM is how a robot determines its own position (localization) in an unknown environment and simultaneously creates a map of its surroundings (mapping). As the robot moves, the LiDAR continuously scans the surrounding environment, acquiring point cloud data. Feature point extraction algorithms, such as the Fast Corner Detection Algorithm (FAST), extract representative feature points from the point cloud data. These feature points can be obvious location markers such as corners and pillars. The geometric relationships between these feature points and the motion model are then used, along with algorithms such as the extended Kalman filter, to estimate the robot's position (position and posture). This scanned environmental information is then integrated into the map. As the robot continues to move, this map gradually improves, providing a basis for navigation in unknown environments.

[0159] 3. Positioning

[0160] (1) Map-based positioning

[0161] Once the robot has a map of the environment, it can determine its position by matching its own sensor data with the map information. For example, a particle filter algorithm can be used for positioning. Suppose the robot is in an indoor environment, and there are multiple feature points on its map, such as doors and windows. After the robot's sensors (such as lidar) scan the surrounding environment, they extract the currently observed feature points. The particle filter algorithm generates a series of particles, each of which represents a possible position and posture of the robot. The weight of each particle is then updated based on the degree of match between the feature points observed by the sensor and the feature points on the map. The weight of particles with a high degree of match increases, and vice versa. Through this iterative process, the position and posture represented by the particle with the largest weight is considered to be the most likely position and posture of the robot at the moment.

[0162] (2) Absolute positioning and relative positioning

[0163] ① Absolute positioning: This refers to the robot determining its position within a global coordinate system. For example, GPS (Global Positioning System) is used to locate an outdoor robot. GPS receives satellite signals and calculates the robot's latitude and longitude coordinates on the Earth's surface, thereby determining its absolute position. However, GPS signals are weak in indoor environments or when obstructed, significantly reducing positioning accuracy.

[0164] ②Relative positioning: The robot determines its displacement and posture change relative to its starting position based on its own motion information (such as wheel speed and gyroscope data). A wheeled robot can calculate its movement distance and direction relative to its starting point by measuring the rotation speed and steering angle of its left and right wheels and combining this with the robot's kinematic model. This kinematic model can be expressed as:

[0165]

[0166] Where Δx and Δy represent the displacement increments along the x-axis and y-axis, v r and v l are the rotation speeds of the left and right wheels respectively, L is the wheelbase, θ is the current posture, Δθ represents the change in posture angle within time Δt, and Δt is the time interval.

[0167] 4. Control Execution

[0168] (1) Motion control

[0169] Following the planned path, the robot needs to use a motion control algorithm to precisely control its movement. For wheeled robots, a PID (proportional-integral-differential) control algorithm can be used. This control method allows the robot to quickly and accurately reach the target speed and move smoothly along the planned path. The control signal calculation formula is:

[0170]

[0171] Among them, u(t) is the control signal, e(t) is the error signal, that is, the difference between the target value and the actual value, k p is the proportionality coefficient, k i is the integration coefficient, k d is the differential coefficient.

[0172] (2) Obstacle avoidance control

[0173] When the sensor detects an obstacle ahead, the robot needs to take timely avoidance measures. Fuzzy logic control is used for obstacle avoidance. The fuzzy logic controller can output the robot's obstacle avoidance actions (such as turning left, turning right, slowing down, etc.) based on fuzzy information (such as "close" and "fast") input from the sensor, such as the obstacle distance and relative speed. When the sensor detects that the obstacle ahead is "close" and moving toward the robot at a "fast speed," the fuzzy rules can set the robot to "turn right quickly" and "slow down," effectively avoiding the obstacle and ensuring the robot's safety.

[0174] (2) Robot pose measurement of the target: Vision-based pose measurement (PNP)

[0175] The PNP method calculates the robot's position relative to the target object by knowing the positions of several feature points on the target object in the image and combining them with the positions of these feature points in the world coordinate system. This method requires the precise location of the feature points of the target object, and feature extraction needs to be stable in a changing environment, otherwise errors are likely to occur. The basic formula for the PNP problem can be expressed as:

[0176] λ i n i =M[r|W]N i

[0177] Among them, λ i is a scaling factor; n i is the 2D projection coordinate of the i-th point on the image; M is the intrinsic parameter matrix of the camera; r is the rotation matrix; W is the translation matrix; N i is the 3D coordinate of the i-th point in the world coordinate system.

[0178] The present invention first collects data using five types of sensors on a multi-source information acquisition platform, then performs close-up operations through a portable experimental platform to identify and sample goods such as wood, and finally processes data on a one-stop management platform, completing this series of tasks under remote guidance.

[0179] Figure 2-Figure 4These are schematic diagrams of the working principles and structures of the multi-source information acquisition platform, portable experimental platform, and one-stop management platform.

[0180] The multi-source information acquisition platform uses five different sensors to collect data, which is then connected to the power supply, acquisition unit, and controller via different interfaces to generate a feature database. The portable experimental platform is navigated by a shared field of view camera, and a robotic arm performs intelligent sampling and extraction, assisted by a wood identification device and a rapid disposal box. The one-stop management platform integrates 5G network data transmission and data integration technologies, intelligent image recognition technology based on deep learning, and a remote expert assistance system.

[0181] In summary, this invention addresses the multi-source information integration needs of mobile intelligent outposts by proposing a novel data integration solution that organically integrates different types of information to provide comprehensive support for subsequent analysis and decision-making. By developing an iteratively updateable image recognition algorithm, it enables rapid identification of invasive organisms and other risk factors, and integrates odor data for risk assessment. An insect suction device and a pan-tilt camera integrated into the robotic arm capture and identify the interior of the wood. Furthermore, a remote expert assistance system is proposed, enabling real-time sharing of information collected by the equipment and obtaining professional guidance.

[0182] Example 2

[0183] The present invention further provides an application method of the port mobile intelligent outpost multi-source information equipment system according to the first embodiment, comprising the following steps:

[0184] S1. Based on the environmental factors of the port, select the deployment location of the system and deploy the intelligent outpost robot;

[0185] S2, based on the intelligent outpost robot operation system, collects multi-source information data of imported goods to be quarantined at customs ports;

[0186] S3. Based on multi-source information data, conduct analysis and sampling to generate quarantine result reports for imported goods at customs ports to be quarantined.

[0187] Furthermore, the environmental factors of the port in S1 include: geographical location factors, topographic factors, climatic conditions factors, human flow factors and vehicle flow factors.

[0188] S2 includes:

[0189] S21. The intelligent outpost robot autonomously moves to the quarantine area of ​​the imported goods at the customs port to be quarantined according to the preset route and tasks, and operates a system to collect multi-source information data of the imported goods at the customs port to be quarantined;

[0190] S22. Preprocess multi-source information data.

[0191] S3 includes:

[0192] S31. Perform image recognition analysis based on multi-source information data to obtain recognition analysis results;

[0193] S32. Using the robotic arm of the intelligent outpost robot, identify and sample imported goods at the customs port to be quarantined, and obtain sampling data;

[0194] S33. Based on the identification and analysis results and the sampling data, a quarantine result report for the imported goods to be quarantined at the customs port is generated.

[0195] The specific implementation is as follows:

[0196] 1. System Deployment and Preparation

[0197] 1. Choose the right scenario:

[0198] Based on the port's geographical location, topography, climatic conditions, pedestrian and vehicle traffic, and other environmental factors, select scenes with wide views, sufficient light, no obstructions, and convenient equipment installation and maintenance as the system deployment location, such as open areas near the border line, high places at port terminals, and near airport control towers. This ensures that the intelligent outpost robots can cover all important areas of the port. At the same time, the stability of data transmission and the operating environment of the terminal analysis software are taken into consideration to avoid electromagnetic interference, signal blockage, and other problems.

[0199] 2. Deployment of intelligent outpost robots:

[0200] First, based on the scenario characteristics and monitoring requirements, select the appropriate intelligent outpost robot model. For example, a robot with high mobility, flexibility, autonomous navigation capabilities, and multi-sensor integration is recommended to adapt to the complex port environment. Then, deployment is carried out according to the robot's deployment specifications and requirements, including positioning, securing, and connecting power and signal cables to ensure stability and safety. During deployment, attention must also be paid to the robot's orientation and angle to ensure full coverage of the monitoring area, avoiding blind spots and blind areas.

[0201] 3. Configure the data transmission module:

[0202] Based on system requirements and the on-site network environment, select the appropriate data transmission module, such as a wireless transmission module (such as a 4G / 5G module) or a wired transmission module (such as a fiber optic module). For wireless transmission modules, you need to configure their network parameters, including the APN (Access Point Network), server address, and port number, to ensure smooth network connection. For wired transmission modules, you need to perform physical connections and network configuration, including fiber splicing and fiber transceiver configuration, to ensure stable and efficient data transmission.

[0203] 4. Install terminal analysis software:

[0204] Install terminal analysis software on the server or workstation. This software should include image processing, image recognition, data analysis, and report generation capabilities. Before installation, ensure that the system hardware configuration meets the software's operating requirements, including processor performance, memory capacity, and storage space. Complete the software installation and configuration, including setting software parameters and importing relevant data models and algorithm libraries, to ensure proper operation and achieve the intended functionality.

[0205] 2. System Operation and Image Acquisition

[0206] 1. Start the Smart Outpost robot:

[0207] After system deployment and preparation are complete, the Smart Sentinel robot is activated to begin its monitoring mission. Before startup, the robot must be initialized, including calibrating sensors, setting navigation parameters, and checking power status to ensure proper operation. Once activated, the robot will autonomously move to the designated monitoring area according to the pre-set route and mission and begin collecting images.

[0208] 2. Image preprocessing:

[0209] Before being transmitted to the terminal analysis software, captured image data requires preprocessing to improve image quality and subsequent processing efficiency. Preprocessing includes operations such as denoising, enhancement, and normalization. Denoising removes noise from the image using filters such as mean and median filters. Enhancement enhances image features by adjusting parameters such as brightness, contrast, and saturation. Normalization standardizes the image data to a standard range to facilitate subsequent processing and analysis.

[0210] 3. Image data transmission:

[0211] The preprocessed image data is transmitted to the terminal analysis software via the data transmission module. Data integrity and real-time performance must be ensured during transmission. For scenarios requiring high real-time performance, the UDP (User Datagram Protocol) protocol can be used for data transmission. UDP (User Datagram Protocol) offers strong real-time performance and high transmission efficiency. For scenarios requiring high data integrity, the TCP (Transmission Control Protocol) protocol can be used for data transmission, as it provides a reliable data transmission mechanism. Data compression technology can also be used during data transmission to reduce data volume and improve transmission efficiency.

[0212] 3. Terminal Analysis and Sampling

[0213] 1. Receive image data:

[0214] After receiving the transmitted image data, the terminal analysis software will store it in memory or on the hard disk for subsequent analysis and processing. During the receiving process, the data needs to be verified to ensure its integrity and accuracy.

[0215] 2. Image recognition and analysis:

[0216] The terminal analysis software performs image recognition and analysis on the received image data. It uses a convolutional neural network to simulate the visual information processing mechanism of the human brain, and uses structures such as convolutional layers and pooling layers to automatically extract image features. It then abstracts higher-level feature representations layer by layer to quickly identify invasive organisms and other risk factors.

[0217] 3. Robotic arm sampling:

[0218] Based on real-time data collected by the gimbal, the robotic arm of the Smart Sentinel robot grinds the wood and uses an insect sucker to sample and analyze insects and wood debris. Based on image recognition and analysis, the robotic arm precisely locates the sampling location on the wood. It then grinds the wood to expose insects and wood debris. The insect sucker then sucks these insects and wood debris up for subsequent analysis and testing.

[0219] 4. Generate result report:

[0220] The terminal analysis software generates a report based on the results of image recognition and analysis, as well as data collected by the robotic arm. This report includes information on the port's operational status, security situation, abnormal events, and wood pest infestation. The report can be presented in either text or chart format, providing users with a visual understanding of the port's status. The report can also be customized to meet user needs, such as including data for a specific time period or analysis of a specific area.

[0221] 4. System maintenance and management

[0222] 1. Regular inspection and maintenance: Regularly inspect and maintain the robot's different platform hardware systems, data transmission modules, and data analysis software to ensure their normal operation.

[0223] 2. Data backup and recovery: Regularly back up received image data to prevent data loss. At the same time, establish a data recovery mechanism to ensure timely recovery in the event of data loss or damage.

[0224] 3. System upgrade and optimization: Based on the latest wood classification image data, the system will be upgraded and optimized, including improvements to image recognition algorithms and optimization of data transmission modules, to improve system performance and accuracy.

[0225] Through the above-mentioned implementation, the multi-source information equipment system for mobile intelligent outposts at ports effectively implements timber quarantine and sampling and analysis processes at customs ports. Through precise, integrated analysis between data collection and management platforms, the application of intelligent robots in customs quarantine not only improves efficiency and accuracy, reduces occupational risks, but also adapts to complex working environments, enables intelligent management, reduces costs, and promotes technological development and application.

[0226] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by persons skilled in the art should fall within the scope of protection defined by the claims of the present invention.

Claims

1. A multi-source information equipment system for mobile intelligent outposts at ports, implemented on intelligent outpost robots, characterized in that: The system includes: a multi-source information acquisition platform, a portable experimental platform and a one-stop management platform; The multi-source information collection platform is used to collect multi-source information data of imported goods to be quarantined at customs ports; The portable experimental platform is used to identify and sample the imported goods to be quarantined at the customs port and obtain identification and sampling results; The one-stop management platform is used to perform data processing based on the multi-source information data and the identification sampling results to obtain a quarantine result report for imported goods to be quarantined at the customs port.

2. The port mobile intelligent outpost multi-source information equipment system according to claim 1 is characterized in that: The multi-source information acquisition platform includes: an acquisition unit, a power supply unit and a data processing unit; The acquisition unit is used to collect multi-source information data of the imported goods to be quarantined at the customs port based on visible light sensors, infrared sensors, near-infrared sensors, spectral sensors and chemical sensors; The power supply unit is used to provide power to the sensor unit; The data processing unit is used to pre-process the multi-source information data.

3. The port mobile intelligent outpost multi-source information equipment system according to claim 2 is characterized in that: The portable experimental platform includes: an observation lens, a robotic arm, a disposal box and a wood identification device; The observation lens is used to perform path planning and navigation obstacle avoidance for the intelligent outpost robot; The robotic arm is used to identify and sample the imported goods to be quarantined at the customs port; The disposal box is used to store and polish sampling tools; The wood identification device is used to identify the wood species in the imported goods to be quarantined at the customs port.

4. The port mobile intelligent outpost multi-source information equipment system according to claim 3 is characterized in that: The one-stop management platform includes: a data transmission unit, a data integration unit, an image recognition unit and a remote assistance unit; The data transmission unit is used to transmit the multi-source information data to a control room for real-time monitoring; The data integration unit is used to construct a multi-dimensional feature database based on the multi-source information data; The image recognition unit is configured to recognize real-time images in the multi-source information data based on an intelligent image recognition algorithm, and obtain a risk assessment index in combination with odor data in the multi-source information data; The remote assistance unit is used for remote guidance.

5. The port mobile intelligent outpost multi-source information equipment system according to claim 4 is characterized in that: The data integration unit is used to construct a multidimensional feature database based on the multi-source information data, including: Extracting color features, texture features, and shape features of the goods based on the data collected by the visible light sensor, and generating a color feature vector, a texture feature vector, and a shape feature vector; Extracting temperature distribution features based on data collected by the infrared sensor and generating a temperature distribution feature vector; Extracting the spectral absorption characteristics of the substance based on the data collected by the near-infrared sensor and generating a spectral absorption characteristic vector; Based on the data collected by the spectral sensor, spectral features are extracted and spectral feature vectors are generated; Based on the data collected by the chemical sensor, chemical composition characteristics are extracted and chemical composition feature vectors are generated; Data fusion is performed based on the color feature vector, the texture feature vector, the shape feature vector, the temperature distribution feature vector, the spectral absorption feature vector, the spectral feature vector and the chemical composition feature vector to construct a multi-dimensional feature database.

Images