Elevator floor detection system based on tof sensor
Patent Information
- Application Number
- CN202521330019.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Utility models(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-26
- Publication Date
- 2026-08-18
- Estimated Expiration
- 2035-06-26
AI Technical Summary
[0002]现有的电梯系统依赖于机械开关、光电编码器或RFID标签等来进行楼层检测，这些方法容易因磨损、错位或受环境干扰等出现一些偏差；同时，电梯的平整度检查通常是使用基本的接近传感器，存在精度上的欠缺；因此，急需一种基于ToF传感器的电梯楼层检测系统来解决上述问题
[0012]本实用新型的有益效果：基于ToF传感器的电梯楼层检测系统，包括楼层块、ToF传感器以及处理单元；楼层块设置在层门门槛上，楼层块上设置有在光学上可区分的标识符；ToF传感器设置在轿厢顶部，其被配置为测量与层门门槛之间的距离以及检测楼层块上的标识符；还包括与处理单元以及电梯控制系统通信的通信模块；处理单元与ToF传感器通信，其被配置为对ToF传感器测得的距离与平整度阈值进行比较，以确定轿厢与层门门槛之间的垂直和/或水平间隙，以及，被配置为对楼层块上的标识符进行解码，以确定当前楼层；能够增强安全性，减少了机械磨损，并提高了电梯的操作精度，满足使用需求。
Smart Images
Figure CN224646432U_ABST
Abstract
Description
Technical Field
[0001] This utility model relates to the field of elevator safety systems, and in particular to an elevator floor detection system based on a ToF sensor. Background Technology
[0002] Existing elevator systems rely on mechanical switches, photoelectric encoders, or RFID tags for floor detection. These methods are prone to deviations due to wear, misalignment, or environmental interference. Meanwhile, elevator leveling checks typically use basic proximity sensors, which lack accuracy. Therefore, there is an urgent need for an elevator floor detection system based on ToF sensors to solve these problems. Utility Model Content
[0003] This invention aims to solve at least one of the technical problems existing in the prior art. To this end, this invention proposes an elevator floor detection system based on a ToF sensor.
[0004] The technical solution adopted by one embodiment of this utility model to solve its technical problem is: an elevator floor detection system based on a ToF sensor, including a floor block, a ToF sensor and a processing unit;
[0005] Floor blocks are set on the threshold of the floor doors, and optically distinguishable identifiers are set on the floor blocks;
[0006] The ToF sensor is located on the top of the car and is configured to measure the distance between itself and the landing door threshold as well as detect identifiers on the floor blocks.
[0007] It also includes a communication module that communicates with the processing unit and the elevator control system;
[0008] The processing unit communicates with the ToF sensor and is configured to compare the distance measured by the ToF sensor with a flatness threshold to determine the vertical and / or horizontal gap between the car and the landing door threshold, and is configured to decode the identifier on the floor block to determine the current floor.
[0009] As one of the preferred embodiments of this utility model, a two-dimensional identification code is provided on the floor block to form an identifier.
[0010] As one of the preferred embodiments of this utility model, the floor block is provided with a three-dimensional block of asymmetrical geometry to form an identifier.
[0011] As one of the preferred embodiments of this utility model, the flatness threshold is set to 2mm.
[0012] The beneficial effects of this utility model are as follows: The elevator floor detection system based on a ToF sensor includes a floor block, a ToF sensor, and a processing unit. The floor block is set on the landing door sill and has an optically distinguishable identifier. The ToF sensor is set on the top of the car and is configured to measure the distance between the car and the landing door sill and to detect the identifier on the floor block. It also includes a communication module that communicates with the processing unit and the elevator control system. The processing unit communicates with the ToF sensor and is configured to compare the distance measured by the ToF sensor with a flatness threshold to determine the vertical and / or horizontal gap between the car and the landing door sill, and to decode the identifier on the floor block to determine the current floor. This enhances safety, reduces mechanical wear, and improves the elevator's operational accuracy, meeting usage requirements. Attached Figure Description
[0013] The above and / or additional aspects and advantages of this utility model will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:
[0014] Figure 1 This is a schematic diagram illustrating a real-world application scenario of an elevator floor detection system based on a ToF sensor.
[0015] Figure 2 A diagram illustrating the floor detection steps of an elevator floor detection system based on a ToF sensor;
[0016] Figure 3 This is a flowchart of the floor detection process for an elevator floor detection system based on a ToF sensor.
[0017] Figure 4 A schematic diagram of one embodiment of a floor block;
[0018] Figure 5 This is a schematic diagram of the 8x8 area used by the ToF sensor. Detailed Implementation
[0019] This section will describe in detail the specific embodiments of the present utility model. The preferred embodiments of the present utility model are shown in the accompanying drawings. The purpose of the drawings is to supplement the textual description with graphics, so that people can intuitively and vividly understand each technical feature and the overall technical solution of the present utility model, but they should not be construed as limiting the scope of protection of the present utility model.
[0020] In the description of this utility model, "multiple" means two or more; "greater than," "less than," and "exceeding" are understood to exclude the stated number; "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly specifying the number of indicated technical features or their sequential relationship.
[0021] In the description of this utility model, it should be understood that the directional descriptions, such as up, down, front, back, left, right, etc., indicate the directional or positional relationship based on the directional or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this utility model and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this utility model.
[0022] In this utility model, unless otherwise explicitly defined, the terms "setting," "installing," and "connecting" should be interpreted broadly. For example, they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to a fixed connection, a detachable connection, or an integral molding; they can refer to a mechanical connection; they can refer to the internal connection of two components or the interaction between two components. Those skilled in the art can reasonably determine the specific meaning of the above terms in this utility model in conjunction with the specific content of the technical solution.
[0023] Reference Figures 1-2 An elevator floor detection system based on a ToF sensor includes a floor block 10, a ToF sensor 20, and a processing unit 30.
[0024] Floor block 10 is set on the threshold of the floor door, and floor block 10 is provided with an optically distinguishable identifier;
[0025] ToF sensor 20 is located on the top of the car and is configured to measure the distance between itself and the landing door threshold and to detect identifiers on the floor block 10.
[0026] The processing unit 30 communicates with the ToF sensor 20 and is configured to compare the distance measured by the ToF sensor 20 with a flatness threshold to determine the vertical and / or horizontal gap between the car and the landing door threshold, and is configured to decode the identifier on the floor block 10 to determine the current floor.
[0027] In this invention, 1) the floor blocks 10 are designed to be optically distinguishable, as shown in the reference. Figure 4 In some possible embodiments, floor block 10 is a three-dimensional block with an asymmetrical geometry, for example, using three-dimensional Arabic numerals to encode floor numbers; 2) Refer to Figures 1-3 and Figure 5In some possible embodiments, the ToF sensor 20 measures the distance between itself (the car) and the landing door sill, thereby checking the vertical and / or horizontal gap between the car and the landing door sill. When the deviation exceeds a set threshold, an alarm is triggered or the elevator stops moving until it is realigned. Specifically, the ToF sensor 20 detects identifiers on the floor block 10 using infrared light pulses. The ToF sensor 20 connects to the processing unit 30 using an I2C or SPI interface. In some possible embodiments, the ToF sensor 20 with 8x8 multi-area output uses a pixelated sensor array to capture spatial data, with each pixel measuring distance and intensity. The system creates a 3D depth map of the identifiers on floor block 10, and matches the detected identifiers with pre-stored templates to determine the position of the car and floor; 3) The processing unit 30 is a microcomputer system, based on an ARM CPU and NPU and an Android operating system, connected to the elevator control system via a CAN bus / serial bus. The processing unit 30 also acts as an AIoT device, using the Internet to send data to a cloud server. The processing unit 30 compares the distance data of the ToF sensor 20 with a predefined flatness threshold (e.g., ≤2mm deviation), and can also decode the identifiers on floor block 10 to determine the current floor.
[0028] The data processing procedure for the ToF sensor 20 is as follows:
[0029] ① Data collection
[0030] The ToF sensor 20 measures distance by emitting infrared light pulses and measuring their return time. Specifically, the ToF sensor 20 uses an 8x8 pixel array to capture spatial data, with each pixel measuring distance and reflectivity (intensity) to create a 3D depth map of the identifier on the floor block.
[0031] ② Determining distance and intensity
[0032] The distance and intensity of each pixel can be determined using the following steps:
[0033] Distance measurement: The ToF sensor 20 emits an infrared light pulse and measures the time from emission to return of the pulse. The distance is:
[0034]
[0035] Where c is the speed of light and t is time;
[0036] ③ Multi-frequency resolution
[0037] The ToF sensor 20 can operate at multiple modulation frequencies to improve the resolution and accuracy of distance measurement. By using multiple frequencies, the system can resolve ambiguities in distance measurement caused by the periodicity of the ToF signal.
[0038] Multi-frequency phase unwrapping: The distance d can be unwrapped at multiple frequencies f1, f2, ..., f n The phase shift Δφ of the reflected signal is measured and calculated. The phase shift at each frequency is given by the following formula:
[0039]
[0040] By combining phase measurements at multiple frequencies, the system can resolve the true distance without creating ambiguity.
[0041] ④ Filtering and calibration
[0042] Filtering and calibration are key steps to ensure the accuracy and reliability of the ToF sensor 20 measurements.
[0043] - Median filtering: Median filtering is used to remove noise and outliers in distance and intensity measurements. The filtered value for pixel i... Given by the following formula:
[0044]
[0045] Where k is the size of the filtering window.
[0046] - Calibration: Calibration involves correcting systematic errors in the measurements of the ToF sensor 20, which may include correcting offset errors, scaling errors, and environmental factors (such as temperature and humidity). Calibration parameters are typically determined through a series of controlled measurements and applied to the raw data.
[0047] ⑤ CNN Loss and Optimization
[0048] Convolutional neural networks (CNNs) are used to process 3D depth images captured by ToF sensors and identify floor markers.
[0049] - Loss Function: The loss function of a CNN is usually a combination of classification loss and regression loss. For floor marker recognition, a common choice is the cross-entropy loss for classification and the mean squared error (MSE) loss for regression.
[0050] L = L classification +λL regression
[0051] Where λ is the weighting factor.
[0052] - Optimization: CNNs are trained using stochastic gradient descent (SGD) or its variants (such as Adam). The optimization process involves minimizing the loss function with respect to the network parameters θ.
[0053]
[0054] Where η is the learning rate. This represents the gradient of the loss function with respect to the parameters.
[0055] ⑥ Confidence index
[0056] The confidence metric is used to evaluate the reliability of the ToF sensor's measurements and CNN predictions.
[0057] - Confidence score: The confidence score for distance measurement can be based on the signal-to-noise ratio (SNR) of the received signal:
[0058]
[0059] Where τ is the threshold
[0060] -CNN Confidence: For CNN predictions, the confidence score can be derived from the softmax probability of the classification output. The confidence score for predicting class c is given by the following formula:
[0061]
[0062] Among them, z c It is the logit of class c, and the summation is performed on all classes.
[0063] By combining these detailed aspects, ToF measurement systems can achieve greater accuracy, robustness, and reliability in a variety of applications, including elevator safety systems.
[0064] Reference Figures 2-3 In some embodiments, the present invention also provides a control method applied to the elevator safety system, comprising the following steps:
[0065] S1. Hardware setup and initialization;
[0066] S2. Connect the ToF sensor 20 (e.g., ST VL53L5CX 8*8 multi-area) to the processing unit 30 (ARM CPU) via I2C / SPI, and configure the interface using ST's HAL library (e.g., STM32CubeMX); calibrate the ToF sensor 20 for ambient light compensation and offset correction; integrate the floor block 10 with a predefined template in the processing unit 30;
[0067] S3, ToF sensor 20 data acquisition: captures an 8*8 depth / density matrix, reading distance (mm) and reflectivity (intensity) for each of the 64 areas, compares this data with nominal data to detect floor level and flushing conditions; filters noise, applies a medium filter to the raw data and discards outliers, checks the flatness threshold (should be below 2mm), and uses the depth matrix to check flushing degree;
[0068] S4. Floor Recognition: A lightweight convolutional neural network model supported by a ToF sensor 20 is trained using an NPU in the processing unit 30. The current floor can be obtained by comparing the detected identifier with a pre-stored template.
[0069] S5. The processed data is sent to the elevator control system or server via CAN bus or serial bus interface.
[0070] The Convolutional Neural Network (CNN) model is used to decode the identifiers on the floor tiles to determine the current floor. To supplement the specific training process of the CNN model, we can describe in detail the data preparation, model architecture, training process, and evaluation methods. The following is a complete description of the training process:
[0071] ① Data preparation
[0072] First, a training dataset needs to be prepared; this dataset consists of 8x8 depth images acquired by the ToF sensor 20 and their corresponding floor labels. Specific steps include:
[0073] - Data Acquisition: Depth images of different floors were acquired using a ToF sensor 20, and the corresponding floor for each image was labeled.
[0074] - Data augmentation: Enhance images by methods such as rotation, scaling, and translation to increase the diversity and robustness of the dataset.
[0075] - Data standardization: Standardize the pixel values of the depth image to the range of [0,1] to facilitate neural network processing.
[0076] ② Model Architecture
[0077] Design a lightweight CNN model suitable for 8x8 depth images. A simple CNN architecture can include the following layers:
[0078] -Input layer
[0079] Dimensions: (8,8,1) (Single-channel depth map)
[0080] - Convolutional layer 1
[0081] Number of convolution kernels: 8
[0082] kernel size: 3×3
[0083] Activation function: ReLU
[0084] Output: (6,6,8)
[0085] -Max pooling layer 1
[0086] Pooling size: 2×2
[0087] Step size: 2
[0088] Output: (3,3,8)
[0089] - Convolutional layer 2
[0090] Number of convolution kernels: 16
[0091] kernel size: 3×3
[0092] Activation function: ReLU
[0093] Output: (1,1,16)
[0094] - Flattening layer
[0095] Transform (1,1,16) into a 16-dimensional eigenvector
[0096] - Fully connected layer 1
[0097] Number of neurons: 32
[0098] Activation function: ReLU
[0099] - Output layer
[0100] Number of neurons: Number of categories (e.g., number of floors)
[0101] Activation function: Softmax
[0102] ③ Training process
[0103] During training, it is necessary to define the loss function, optimizer, and evaluation metrics.
[0104] - Loss function: The cross-entropy loss function is used to measure the difference between the model's prediction and the true label.
[0105] Among them, y i For real labels, p i This represents the probability predicted by the model.
[0106] - Optimizer: Use the Adam optimizer to minimize the loss function.
[0107] Adam(learning_rate=0.001)
[0108] - Training loop: Training is performed on the training set for multiple epochs. Each epoch includes forward propagation, loss calculation, backpropagation, and parameter update.
[0109]
[0110] Where, θ t Here are the current model parameters, and η is the learning rate. This represents the gradient of the loss function with respect to the parameters.
[0111] ④ Evaluation methods
[0112] During training, the model's performance needs to be evaluated periodically. A validation set can be used to assess metrics such as accuracy, precision, recall, and F1 score.
[0113] -Accuracy: The proportion of samples that the model correctly predicts on the validation set.
[0114]
[0115] - Precision and Recall: Calculate the precision and recall for each class and take the macro average or micro average.
[0116]
[0117] By following the steps above, an effective CNN model can be trained to decode the identifiers on the floor blocks and determine the current floor.
[0118] In a further embodiment, the data collected by the ToF sensor 20 is polled and filtered for noise using I2C / SPI, flatness is checked using threshold and statistical deviation, floor identification is performed using cross-correlation and CNN (convolutional neural network) inference, and communication with the elevator control system is performed using the CAN open protocol / serial bus interface.
[0119] As a first embodiment of floor block 10, floor block 10 is provided with a two-dimensional identification code to form an identifier, such as a binary code.
[0120] As a second embodiment of the floor block 10, the floor block 10 is provided with a three-dimensional block of asymmetrical geometry to form an identifier; wherein, the three-dimensional block of asymmetrical geometry can be a convex block or a concave groove.
[0121] In some embodiments, the flatness threshold is set to 2 mm.
[0122] In some embodiments, the elevator floor detection system based on ToF sensors further includes a communication module that communicates with the processing unit 30 and the elevator control system.
[0123] The advantages of this invention are: it enhances safety, reduces mechanical wear, and improves the operational precision of the elevator, thus meeting usage requirements.
[0124] Of course, this utility model is not limited to the above-described embodiments. Those skilled in the art can make equivalent modifications or substitutions without departing from the spirit of this utility model. All such equivalent modifications and substitutions are included within the scope defined by the claims of this application.
Claims
1. An elevator floor detection system based on a ToF sensor, characterized in that: comprising a floor block (10), a ToF sensor (20) and a processing unit (30); said floor block (10) is arranged on a landing door sill, said floor block (10) is provided with an optically distinguishable identifier; said ToF sensor (20) is arranged on top of the car and is configured to measure the distance to the landing door sill and to detect the identifier on said floor block (10); further comprising a communication module in communication with said processing unit (30) and an elevator control system; said processing unit (30) is in communication with said ToF sensor (20) and is configured to compare the distance measured by said ToF sensor (20) with a flatness threshold to determine a vertical and / or horizontal gap between the car and the landing door sill and to decode the identifier on said floor block (10) to determine the current floor.
2. The ToF sensor based elevator floor detection system according to claim 1, characterized in that: said floor block (10) is provided with a two-dimensional identification code to constitute said identifier.
3. The ToF sensor based elevator floor detection system according to claim 1, characterized in that: said floor block (10) is provided with a three-dimensional tile of asymmetric geometry to constitute said identifier.
4. The ToF sensor based elevator floor detection system of claim 1, wherein: said flatness threshold is set to 2 mm.