Rail train foreign matter detection training system

Abstract

The invention provides a rail train foreign matter detection training system which comprises a laser radar detection module, an attitude acquisition module, a depth camera module, a main control module, an upper industrial personal computer, a motor driving module and a crankshaft transmission module. The two-dimensional and three-dimensional point cloud map data acquisition module is used for acquiring two-dimensional and three-dimensional point cloud map data of the surrounding environment of the rail train, acquiring the angular velocity and the linear velocity of the rail train, acquiring depth image and color image information of a rail obstacle, processing data information output by the laser radar detection module, the attitude acquisition module and the depth camera module, and displaying the processed data information. And the control module is used for planning an obstacle avoidance route of the rail train according to the data information collected and output by the main control module, and outputting a corresponding control instruction to the main control module, the motor driving module and the crankshaft transmission module in sequence so as to control braking and starting of the rail train. According to the invention, rapid identification and response to track obstacles can be realized, and an obstacle avoidance route can be rapidly planned or emergency braking can be executed.

Description

Technical Field

[0001] The present invention relates to the technical field of rail train obstacle avoidance, and more specifically, to a rail train foreign body detection training system. Background Art

[0002] With the rapid development of modern urban rail transit, its safe operation issues have attracted more and more attention from the society. Foreign objects on the track will not only cause train delays and suspensions, but may also cause major accidents such as train derailment, causing huge losses to people's lives and property. Therefore, it is of great significance to study and solve and improve the accuracy and reliability of detection and improve the efficiency of foreign object detection on the track to ensure the safety of modern urban rail transit.

[0003] In the existing technology, the inspection of obstacles by manned rail trains mainly relies on the naked eye observation and personal judgment of the driver. Although some fully automatic unmanned trains are equipped with a contact obstacle detection beam at the bottom of the front of the train, although it can detect obstacles on the track, it is still necessary to stop the rail train and wait for manual handling of the obstacle. The rail train cannot avoid the obstacle by replanning the route. Summary of the invention

[0004] The present invention aims to provide a railway train foreign object detection training system that is non-contact and takes automatic emergency braking measures when necessary. The system can define and classify foreign objects appearing on the track, accurately detect the distance, speed and relative movement of obstacles in front of the train, and predict the possibility of collision. It is also tightly integrated with the train control system including the motor drive module and the crankshaft drive module to achieve rapid response, rapid judgment and execution of emergency braking.

[0005] In order to achieve the above object, the present invention provides a railway train foreign body detection training system, comprising: The laser radar detection module is used to collect two-dimensional and three-dimensional point cloud map data of the surrounding environment of the rail train, including the location information of the rail train and track obstacles; A posture acquisition module, used to acquire the angular velocity and linear velocity of the rail train; A depth camera module, used to obtain depth image and color image information of the track obstacle according to the position information of the track obstacle; A main control module, whose input end is respectively connected to the output ends of the laser radar detection module, the depth camera module, and the posture acquisition module; The main control module establishes and updates the two-dimensional and three-dimensional environment maps in real time according to the acquired two-dimensional and three-dimensional point cloud map data, and marks the location information of the rail train and the rail obstacle, and simultaneously acquires and monitors the parameters of the angular velocity and linear velocity of the rail train in real time, and quickly identifies and calibrates the outline size and name information of the rail obstacle according to the acquired depth image and color image information of the rail obstacle; A host industrial computer is used to communicate with the main control module, and plan and determine the obstacle avoidance route of the rail train according to the two-dimensional and three-dimensional environment maps, the position information of the track obstacles and the angular velocity and linear velocity of the rail train transmitted by the main control module, and output corresponding control instructions to the main control module; A motor drive module, wherein the input end of the motor drive module is connected to the output end of the main control module, and the motor drive module is used to control the braking and starting of the rail train according to the control instructions output by the main control module.

[0006] Furthermore, it also includes a crankshaft transmission module, which is arranged on the chassis of the rail train and integrates a variety of sensors and actuators. The input end of the crankshaft transmission module is connected to the output end of the motor drive module. The motor drive module controls the crankshaft transmission module to transmit power according to the control instruction, thereby controlling the rail train to brake or start.

[0007] Furthermore, the laser radar detection module uses a time-of-flight measurement method to collect two-dimensional and three-dimensional point cloud map data of the surrounding environment of the rail train.

[0008] Furthermore, the depth camera module uses structured light 3D imaging technology to obtain depth image and color image information of track obstacles.

[0009] Furthermore, the laser radar detection module remotely communicates with the main control module via an SSH remote connection tool, and the main control module uses an RViz visualization tool to display two-dimensional and three-dimensional environment maps.

[0010] Furthermore, the main control module calculates the time when the emergency braking of the rail train occurs according to the maximum values ​​of the angular velocity and linear velocity of the rail train obtained from the posture acquisition module, so as to adjust the minimum obstacle avoidance distance of the rail train.

[0011] Furthermore, the main control module uses YoLov5 algorithm to quickly identify and calibrate the outline size and name information of the track obstacle.

[0012] Furthermore, it also includes an external display and an external distribution box, the external display is used to display the image information and various parameter data acquired and generated by the system, and the external distribution box is used to supply power to the system.

[0013] Furthermore, the laser radar of the laser radar detection module and the camera of the depth camera module are arranged on the front of the rail train.

[0014] In summary, compared with the prior art, the present invention has the following beneficial effects: 1. The present invention collects two-dimensional and three-dimensional point cloud map data of the surrounding environment of the rail train through the laser radar detection module to map a digital environment model, and then processes the two-dimensional and three-dimensional point cloud map data through the main control module to obtain two-dimensional and three-dimensional digital map information, establishes and updates high-resolution two-dimensional and three-dimensional surrounding space stereoscopic structure environment maps in real time, and marks the location information of the rail train and track obstacles in the map; At the same time, the parameters of the angular velocity and linear velocity of the track train are collected through the attitude acquisition module, and the depth and color images of the track obstacles are obtained from the real-time images of the depth camera module to quickly identify and calibrate the size and name information of the obstacles; Then, through the upper industrial computer connected to the main control module for communication, the obstacle avoidance route of the rail train is planned and determined according to the two-dimensional and three-dimensional environmental maps, the location information of the track obstacles and the angular velocity and linear velocity of the rail train, so as to automatically avoid obstacles and re-identify and plan the route in the barrier-free area for the next time period, and regenerate and output the corresponding control instructions to the main control module, which then outputs the control instructions to the motor drive module. The motor drive module controls the crankshaft drive module to transmit power to the rail train chassis according to the control instructions, thereby controlling the rail train to brake or start.

[0015] Therefore, the present invention can realize rapid identification and response to foreign objects on the track, and can make rapid judgments to re-plan the route or perform emergency braking.

[0016] 2. The present invention uses a laser radar detection module to receive the signal reflected by the laser pulse in real time, sense the changes in the environment in front of the rail train, and can collect the angular velocity and linear velocity of the rail train in real time through the attitude acquisition module, and process the objects in front of the rail train and obstacles on the track into depth images and color images through the depth camera module, so as to plan the next operation path and trajectory of the rail train training system, and then select the best solution and corresponding control instructions from the obtained planning results, so as to reduce errors, improve recognition efficiency and accuracy, reduce the impact on work results, and make it easier to upgrade and maintain the rail train training system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A system structure diagram of a railway train foreign body detection training system provided by the present invention; Figure 2 This is an overall structural diagram of a rail train foreign object detection training system provided by the present invention. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0019] The disclosure below provides different implementations or examples for realizing different structures of the present application. In order to simplify the disclosure of the present application, the components and settings of specific examples are described below. Of course, they are merely examples and are not intended to limit the present application. In addition, the present application may repeat reference numbers and / or reference letters in different examples. Such repetition is for the purpose of simplicity and clarity, and does not itself indicate the relationship between the various implementations and / or settings discussed. A rail train foreign body detection training system provided by the present invention is described in detail below. It should be noted that the description order of the following embodiments is not intended to limit the preferred order of the embodiments.

[0020] Please refer to the attached Figure 1 The present invention provides a rail train foreign body detection training system, which includes a laser radar detection module, a posture acquisition module, a depth camera module, a main control module, a host industrial computer, a motor drive module, and a crankshaft drive module.

[0021] The main control module and the upper industrial computer can be connected to each other through the communication module. Specifically, the communication module can use RS-485 serial port communication for communication connection.

[0022] like Figure 2 As shown, the present invention also provides a rail train foreign object detection training system, including an external display and an external distribution box. The external display is used to display image information and various parameter data acquired and generated by the training system. The external distribution box is used to supply power to the training system and is mainly used to power the energized track.

[0023] exist Figure 2 It can be seen that the laser radar of the laser radar detection module and the camera of the depth camera module are installed at the front of the rail train. By being installed at the front of the train, the laser radar detection module and the depth camera module can detect and explore foreign objects or obstacles on the track in front of the rail train.

[0024] The laser radar detection module is used to collect two-dimensional and three-dimensional point cloud map data of the environment around the rail train, including the location information of the rail train and track obstacles.

[0025] Specifically, the LiDAR detection module uses the time-of-flight measurement method to collect two-dimensional and three-dimensional point cloud map data of the surrounding environment of the rail train. Time of Flight (ToF) is a commonly used measurement technology, and its basic principle is based on the propagation time of particles or photons in space. Usually, the detection device emits a pulse signal (such as a sound pulse, a light pulse) and receives the reflected or transmitted signal at the target. By accurately measuring the time difference between emission and reception, the distance and speed of signal propagation can be calculated. This method has the advantages of high precision, long distance, and fast response.

[0026] The posture acquisition module is used to collect the angular velocity and linear velocity of the rail train, so as to know the motion trajectory of the rail train.

[0027] Specifically, the attitude acquisition module collects the angular velocity and linear velocity of the rail train through the three-axis gyroscope and the three-axis accelerometer. The attitude acquisition module collects the angular velocity and linear velocity of the rail train and transmits them to the main control module. The main control module calculates the time when the rail train emergency brakes according to the maximum value of the angular velocity and linear velocity of the rail train, so as to calculate and adjust the minimum obstacle avoidance distance of the rail train.

[0028] The depth camera module is used to obtain the depth image and color image information of the track obstacle according to the position information of the track obstacle.

[0029] A depth camera, also known as a 3D camera, is a device that can capture the distance information between an object and the camera, and is mainly used to generate 3D scenes. The working principle of a depth camera is mainly based on 3D structured light technology. This technology uses the light emitted by the camera, and then receives and processes the time difference or diffraction of the light reflected back, thereby calculating the distance information and generating a depth image. Specifically, the light emitted by the camera is reflected back after encountering an object, and the distance from the object to the camera is determined by calculating the time difference or diffraction of these lights.

[0030] Specifically, depth cameras mainly include structured light cameras, ToF cameras, and in-situ projected mask technology depth cameras. In this embodiment, the depth camera module uses structured light 3D imaging technology to obtain depth images and color image information of track obstacles.

[0031] Structured light 3D imaging technology is a technology that projects light of a specific structure (such as gratings, stripes, etc.) onto the surface of an object to form a specific light pattern. When these lights hit the surface of an object, reflection and scattering will occur. The camera captures these reflected lights and restores the three-dimensional form of the object through image processing technology. Specifically, by analyzing the deformation or displacement of the light pattern in the image, the depth information of the object surface can be calculated, thereby constructing a three-dimensional model of the object. Structured light 3D imaging technology calculates distance information by projecting a specific light pattern onto an object and capturing the reflected light pattern. This method has high accuracy at close range and is well suited for detecting foreign objects or obstacles on train tracks.

[0032] The main control module, whose input end is connected to the output end of the laser radar detection module, the depth camera module, and the attitude acquisition module respectively; The main control module establishes and updates the 2D and 3D environment maps in real time based on the acquired 2D and 3D point cloud map data, and marks the location information of the rail train and rail obstacles. At the same time, it acquires and monitors the angular velocity and linear velocity parameters of the rail train in real time, and quickly identifies and calibrates the outline size and name information of the rail obstacles based on the acquired depth image and color image information of the rail obstacles. Specifically, the laser radar detection module can communicate remotely with the main control module through the ssh remote connection tool, and the main control module uses the RViz visualization tool to display two-dimensional and three-dimensional environmental maps. After the laser radar in the laser radar detection module detects the location information of the track obstacle, it is displayed and marked in RViz through visualization, and the barrier-free area will not be marked. The map model training system established by the laser radar can realize the function of autonomous navigation, automatically avoid obstacles when obstacles are detected, and re-plan the route of the barrier-free area, and realize the function of real-time positioning itself and autonomous navigation in a barrier-free environment.

[0033] Among them, RViz (Robot Visualization) is an open source tool for visualizing robot systems, which is used to display and debug the robot's sensor data, status information, and motion planning. RViz provides rich functions and customizable interfaces, allowing users to view sensor data and environmental maps in three dimensions. It supports multiple types of visualization objects, including point clouds, mesh models, markers, paths, laser scans, and camera images. RViz can build a visual rail train model, allowing users to intuitively understand its appearance and posture, and can receive and display data from rail train sensors (such as lidar, cameras, IMU, etc.). Users can view and analyze sensor data in real time to help understand the environment around the rail train. RViz can generate navigation maps. RViz can generate and display two-dimensional or three-dimensional maps of the environment where the rail train is located by receiving data from SLAM (Simultaneous Localization and Mapping) or other mapping algorithms. RViz can debug motion planning. RViz can display the robot's path planning results and provide an interactive interface to debug and optimize motion planning algorithms. Users can visualize information such as virtual paths, obstacles, and collision detection. RViz is customizable and provides a wealth of configuration options, allowing users to customize interface layout, visualization objects, color styles, etc. Users can personalize settings according to actual conditions to meet specific visualization needs.

[0034] Therefore, using RViz visualization tools to display two-dimensional and three-dimensional environment maps can greatly meet the needs of observers.

[0035] In an optional implementation, the main control module uses the YoLov5 algorithm to quickly identify and calibrate the outline size and name information of the track obstacle. YOLOv5 (You Only Look Once version 5) is a real-time target detection algorithm that further improves the detection speed and model lightweight while maintaining high accuracy, and has the advantages of real-time, accuracy, and lightweight.

[0036] The process includes: 1. Input image preprocessing: Take the image containing track obstacles as input and perform necessary preprocessing operations, such as normalization and data enhancement.

[0037] 2. Feature extraction and fusion: Use the Backbone part of the YOLOv5 algorithm (such as CSPDarknet53) to extract features from the input image, and use the Neck part (such as FPN and PAN) to fuse feature maps of different levels to generate a multi-scale feature pyramid.

[0038] 3. Object detection and regression: Use the Head part of the YOLOv5 algorithm to detect objects on the fused feature map. Predict the location (bounding box) and size information of obstacles through the preset anchor boxes and regression algorithm.

[0039] 4. Category classification and confidence assessment: Classify the detected obstacles and output their name information. Calculate the confidence of each detection box to evaluate the reliability of its detection.

[0040] 5. Post-processing and result output: Apply the non-maximum suppression (NMS) algorithm to eliminate overlapping detection boxes and obtain the final detection result. Output the obstacle outline size (bounding box coordinates), name information and confidence.

[0041] Specifically, the obstacles around the rail train are read by opening the depth camera node and marking the data of the identified obstacles through the labelimg labeling tool to generate data of the position coordinate information of four points, import the data into the yolo project, and then modify the file parameters in the data set file path. The obstacle model training uses the train function and the detect function of the test obstacle to train the model. The obstacle model training is to import the labeled training data into the train function described in the previous step of model training, and then run the image processing function part of the qt project to test the trained obstacle model.

[0042] Through the real-time transmission of the depth camera in the depth camera module and the high precision of the YOLOv5 algorithm, the training system can mark and display the obstacles identified in a very short time on the image.

[0043] The upper industrial computer is used to communicate with the main control module, and plans and determines the obstacle avoidance route of the rail train according to the two-dimensional and three-dimensional environmental maps, the location information of the track obstacles and the angular velocity and linear velocity of the rail train transmitted by the main control module, and outputs corresponding control instructions to the main control module.

[0044] The motor drive module has an input end connected to an output end of the main control module, and the motor drive module is used to control the braking and starting of the rail train according to the control instructions output by the main control module.

[0045] The crankshaft drive module is installed on the chassis of the rail train and is integrated with a variety of sensors and actuators. The input end of the crankshaft drive module is connected to the output end of the motor drive module. The motor drive module controls the crankshaft drive module to transmit power according to the control instructions, thereby controlling the rail train to brake or start.

[0046] By precisely controlling the movement of the motor and accurately sensing the environment of the rail train, it can provide smooth and efficient power transmission, reduce energy loss, and facilitate the planning of obstacle avoidance routes in complex environments.

[0047] In summary, the present invention collects two-dimensional and three-dimensional point cloud map data of the surrounding environment of the rail train including the position information of the rail train and the rail obstacles, the angular velocity and linear velocity of the rail train, the depth image and color image information of the rail obstacles through the laser radar detection module, the attitude acquisition module, and the depth camera module, and obtains and processes these data information respectively through the main control module to obtain two-dimensional and three-dimensional environment maps marked with the position information of the rail train and the rail obstacles, the parameters of the angular velocity and linear velocity of the rail train, the outline size and name information of the rail obstacles, and summarizes the processed information into the upper industrial control computer. The upper industrial control computer plans and determines the obstacle avoidance route of the rail train according to this information, so as to automatically avoid obstacles and re-identify and plan the route in the barrier-free area for the next time period, and regenerates and outputs the corresponding control instructions to the main control module, and the main control module outputs the control instructions to the motor drive module. The motor drive module controls the crankshaft drive module to transmit power to the rail train chassis according to the control instructions, thereby controlling the rail train to brake or start.

[0048] Therefore, the present invention can realize rapid identification and response to foreign objects on the track, and can make rapid judgments to re-plan the route or perform emergency braking.

[0049] In addition, the present invention receives the signal reflected by the laser pulse in real time through the laser radar detection module, senses the changes in the environment in front of the rail train, and can collect the angular velocity and linear velocity of the rail train in real time through the attitude acquisition module, and process the objects in front of the rail train and the obstacles on the track into depth images and color images through the depth camera module, so as to carry out the next step of planning for the next operation path and trajectory of the rail train training system, and then select the best solution and corresponding control instructions from the obtained planning results, so as to reduce errors, improve recognition efficiency and accuracy, reduce the impact on work results, and make it easier to upgrade and maintain the rail train training system.

[0050] The above is a detailed introduction to a rail train foreign object detection training system provided in an embodiment of the present application. This article uses specific examples to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the technical solution and its core idea of ​​the present application; ordinary technicians in this field should understand that: they can still modify the technical solutions recorded in the aforementioned embodiments, or replace some of the technical features therein with equivalents; and these modifications or replacements do not make the essence of the corresponding technical solution deviate from the scope of the technical solution of the embodiments of the present application.

Claims

1. A railway train foreign body detection training system, characterized in that: include: The laser radar detection module is used to collect two-dimensional and three-dimensional point cloud map data of the surrounding environment of the rail train, including the location information of the rail train and track obstacles; A posture acquisition module, used to acquire the angular velocity and linear velocity of the rail train; A depth camera module, used to obtain depth image and color image information of the track obstacle according to the position information of the track obstacle; A main control module, whose input end is respectively connected to the output ends of the laser radar detection module, the depth camera module, and the posture acquisition module; The main control module establishes and updates the two-dimensional and three-dimensional environment maps in real time according to the acquired two-dimensional and three-dimensional point cloud map data, and marks the location information of the rail train and the rail obstacle, and simultaneously acquires and monitors the parameters of the angular velocity and linear velocity of the rail train in real time, and quickly identifies and calibrates the outline size and name information of the rail obstacle according to the acquired depth image and color image information of the rail obstacle; A host industrial computer is used to communicate with the main control module, and plan and determine the obstacle avoidance route of the rail train according to the two-dimensional and three-dimensional environment maps, the position information of the track obstacles and the angular velocity and linear velocity of the rail train transmitted by the main control module, and output corresponding control instructions to the main control module; A motor drive module, wherein the input end of the motor drive module is connected to the output end of the main control module, and the motor drive module is used to control the braking and starting of the rail train according to the control instructions output by the main control module.

2. The rail train foreign body detection training system according to claim 1 is characterized in that: It also includes a crankshaft transmission module, which is arranged on the chassis of the rail train and integrates a variety of sensors and actuators. The input end of the crankshaft transmission module is connected to the output end of the motor drive module. The motor drive module controls the crankshaft transmission module to transmit power according to the control instruction, thereby controlling the rail train to brake or start.

3. The rail train foreign body detection training system according to claim 1 is characterized in that: The laser radar detection module uses a time-of-flight measurement method to collect two-dimensional and three-dimensional point cloud map data of the surrounding environment of the rail train.

4. The rail train foreign body detection training system according to claim 1 is characterized in that: The depth camera module uses structured light 3D imaging technology to obtain depth images and color image information of track obstacles.

5. The rail train foreign body detection training system according to claim 1 is characterized in that: The laser radar detection module communicates remotely with the main control module via the SSH remote connection tool, and the main control module uses the RViz visualization tool to display two-dimensional and three-dimensional environment maps.

6. The rail train foreign body detection training system according to claim 1 is characterized in that: The main control module calculates the time when the rail train emergency brake occurs according to the maximum values ​​of the angular velocity and linear velocity of the rail train obtained from the posture acquisition module, so as to adjust the minimum obstacle avoidance distance of the rail train.

7. The rail train foreign body detection training system according to claim 1 is characterized in that: The main control module uses the YoLov5 algorithm to quickly identify and calibrate the outline size and name information of the track obstacle.

8. The rail train foreign body detection training system according to claim 1 is characterized in that: It also includes an external display and an external distribution box. The external display is used to display image information and various parameter data acquired and generated by the system, and the external distribution box is used to supply power to the system.

9. The rail train foreign body detection training system according to claim 1, characterized in that: The laser radar of the laser radar detection module and the camera of the depth camera module are arranged on the front of the rail train.

Images