CN116461571A - Obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion - Google Patents
Obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion Download PDFInfo
- Publication number
- CN116461571A CN116461571A CN202211143675.2A CN202211143675A CN116461571A CN 116461571 A CN116461571 A CN 116461571A CN 202211143675 A CN202211143675 A CN 202211143675A CN 116461571 A CN116461571 A CN 116461571A
- Authority
- CN
- China
- Prior art keywords
- obstacle
- train
- information
- vision
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000004927 fusion Effects 0.000 title claims abstract description 30
- 238000001514 detection method Methods 0.000 claims abstract description 48
- 238000012545 processing Methods 0.000 claims abstract description 24
- 238000012544 monitoring process Methods 0.000 claims abstract description 22
- 230000002159 abnormal effect Effects 0.000 claims abstract description 4
- 238000010224 classification analysis Methods 0.000 claims abstract description 3
- 238000000034 method Methods 0.000 claims description 25
- 238000004891 communication Methods 0.000 claims description 10
- 230000008569 process Effects 0.000 claims description 7
- 230000033001 locomotion Effects 0.000 claims description 3
- 230000004888 barrier function Effects 0.000 claims description 2
- 230000002457 bidirectional effect Effects 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 abstract description 16
- 230000006870 function Effects 0.000 description 20
- 230000000694 effects Effects 0.000 description 10
- 238000012360 testing method Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000005286 illumination Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 238000009826 distribution Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000004807 localization Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 230000036461 convulsion Effects 0.000 description 1
- 239000008358 core component Substances 0.000 description 1
- 230000023077 detection of light stimulus Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000007499 fusion processing Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- 230000003137 locomotive effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L23/00—Control, warning or like safety means along the route or between vehicles or trains
- B61L23/04—Control, warning or like safety means along the route or between vehicles or trains for monitoring the mechanical state of the route
- B61L23/041—Obstacle detection
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/86—Combinations of lidar systems with systems other than lidar, radar or sonar, e.g. with direction finders
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
- G01S17/93—Lidar systems specially adapted for specific applications for anti-collision purposes
- G01S17/931—Lidar systems specially adapted for specific applications for anti-collision purposes of land vehicles
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
- H04N7/181—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast for receiving images from a plurality of remote sources
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Electromagnetism (AREA)
- General Physics & Mathematics (AREA)
- Mechanical Engineering (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Train Traffic Observation, Control, And Security (AREA)
Abstract
The invention discloses an obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion, which comprises a system host obstacle sensing unit and a train control monitoring system; the obstacle sensing unit collects train position information to sense and detect the obstacle and transmit the information to the system host, and the system host carries out classification analysis on the received information; identifying an infringement obstacle on the straight track; when an obstacle is detected, the abnormal condition is transmitted to a train control monitoring system, and simultaneously, a video and a picture with obstacle distance information after processing are transmitted to a ground control center and a driver. The invention adopts a multi-sensor fusion technology to effectively fuse the data acquired by the vision, laser, millimeter wave radar and the sensing unit of the beacon reader, so that the detection function of the obstacle in front of the unmanned train and the auxiliary anti-collision function are realized.
Description
Technical Field
The invention belongs to the technical field of rail transit vehicle control and detection, and particularly relates to an obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion.
Background
The subway train has the advantages of large passenger capacity, accurate arrival time, greenness, energy conservation and the like, can effectively improve urban traffic jam, and becomes one of the main greenness public transportation means for urban travel.
Along with the rapid development of subway trains, the continuous subway train collision, the collision of the subway trains on the end wall during warehouse returning, the derailment and other safety accidents caused by incapability of timely avoiding obstacles during running and the like are caused, so that economic losses are caused. The conventional technology for detecting the obstacle by adopting a single sensor cannot meet the operation requirements of high accuracy, high stability and long-distance detection under the influence of the current hardware equipment and technology.
The Chinese patent application CN110908003A, which is found by searching, discloses an active obstacle detection system based on unmanned driving. The method is characterized by comprising the following steps: the obstacle detection module and the two-door anti-pinch platform are communicated with the video processing server; the obstacle detection module comprises 1 digital high-definition camera and one infrared high-definition camera, and the two cameras shoot images synchronously in real time and send the images to the video processing server for foreign matter intrusion detection; the video processing server shoots an image pair by using two cameras, obtains three-dimensional information of the target object through solving the two-dimensional information of the target object in the image pair, and determines foreign matter intrusion risk; the two-door anti-pinch platform comprises hemispherical high-definition cameras arranged on two sides of a car door, photographs of a region between the car door and the shielding door of a train after parking are taken and sent to the video processing server, and the video processing server performs foreign matter recognition on the ROI region in the two-door space, so that the train is prevented from still driving under the condition of pinching the foreign matter.
The Chinese patent application CN 113591626A, which is found by searching, discloses an obstacle detection system based on a visual sensor. The obstacle detection system based on the vision sensor is carried on a subway train and comprises: the obstacle detection module is used for acquiring front image information and intelligently identifying obstacles for early warning and prompting; the two-door anti-pinch module is used for acquiring image information of two-door areas and intelligently identifying obstacles for early warning and prompting;
the video recording module is used for recording video information and providing original data for accident tracing; the OCC driver eye monitoring module is used for transmitting video information of all the cameras in front of the train to the OCC driver eye monitoring platform; and the communication module is used for being connected with the train control monitoring system for communication.
The two documents mainly replace a driver to continuously watch forward through a visual camera sensor, extract obstacle characteristics through an image recognition technology, determine foreign matters and judge the foreign matter intrusion risk degree, can recognize the obstacle earlier, and reduce the occurrence probability of traffic accidents.
However, in an actual running environment, in a tunnel, at a curve, a bridge is influenced by mottled light and weather, only a single visible light camera is used for shooting images in front of running of a train, the actual effective distance is short, a certain false alarm rate exists, and the detection effect and speed are poor. The subway operation safety accidents still cannot be effectively reduced. Meanwhile, haze can cause great influence on a train in an overhead traveling area. Once haze occurs in a large range, a subway company has to limit speed operation of a train in an overhead traveling area, and the operation efficiency of the whole subway is greatly influenced.
Disclosure of Invention
In order to solve the technical problems mentioned in the background art, the invention provides an obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion.
In order to achieve the technical purpose, the technical scheme of the invention is as follows:
an obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion comprises a system host installed in a cab, an obstacle sensing unit connected with the system host and a train control monitoring system; the obstacle sensing unit collects train position information, front video images and three-dimensional point cloud images, senses and detects obstacles in a rail train operation scene, transmits the collected information to the system host, and carries out classification analysis on the received information; the projection area of the obstacle at 400m or more to the straight track is not less than 0.25m 2 Identifying the size of the infringement barrier; after detecting the obstacle, the system host transmits the abnormal situation to the train control monitoring system through the defined threat level information and the train real-time data protocol interface, and the train control monitoring system takes countermeasures according to the defined threat level and transmits the video and the processed pictures with the obstacle distance information to the ground control center and the driver through train-ground communication.
Preferably, the obstacle sensing unit is provided with a binocular camera sensing module, and comprises a module consisting of two cameras with long focus and short focus, and is arranged above a roof in a cab, POE is adopted for power supply, and an electrical interface is a DB9 interface; and acquiring front high-definition stereoscopic image information, and detecting a front obstacle of the train.
Preferably, the obstacle sensing unit is provided with a multi-line laser sensing module, and the multi-line laser sensing module comprises multi-line laser radars arranged at two ends of a vehicle head, acquires point cloud images of the front obstacle and the vehicle in a bidirectional manner, and detects the front vehicle.
Preferably, the obstacle sensing unit is provided with an auxiliary anti-collision module, and the auxiliary anti-collision module comprises vehicle-mounted secondary radar equipment, ground beacon equipment, a vehicle-mounted electronic tag reader and a ground RFID beacon, wherein the vehicle-mounted secondary radar equipment is used for reading the ground beacon, acquiring the position of a train, the distance and the relative speed between two trains on the same line, whether the trains are positioned on a positive line or a storage line and the up-down state of the trains, and carrying out hierarchical alarm and adjusting the running mode of the trains according to the detected distance.
Preferably, a sensor fusion module is arranged in the system host, the time sequence of real-time data collected by the obstacle sensing unit, the image signal obtained by the camera, the obstacle position information, the millimeter wave radar information and the laser radar point cloud information obtained by processing target detection processing are unified, the alarm signal of whether the obstacle invades the track is output after processing by a fusion algorithm, and the output signal is used for alarming a cab.
Preferably, a target tracking unit is arranged in the system host, an image video signal is processed through a Deep sort algorithm, the position information of a single tracked obstacle, the movement direction information of the obstacle and the current obstacle speed information are output, whether the speed and the direction of the current obstacle invade a track range of train running or not is prejudged according to the output information, and advanced judgment and alarm are made.
Preferably, a video image enhancement module is arranged in the system host, the contrast of the image data is calculated, if the contrast is lower than a certain threshold range, a function in opencv is adopted to adjust the contrast of the image, and if the brightness is higher than or lower than a certain threshold range, the value of each pixel is adjusted to adjust the brightness of the whole image; and meanwhile, according to fusion of the image and the multi-line laser radar point cloud data, whether an obstacle invades the track range is judged in an auxiliary mode.
Preferably, an anti-vibration module is arranged in the system host, and the system host is judged by searching a reference point in a front video image and taking the reference point as a base point; selecting two base points in consideration of up-and-down vibration and left-and-right twisting; in the running process of the train, the rail is used as a reference point, and a smooth video is restored through a base point jitter removing algorithm.
The beneficial effects brought by adopting the technical scheme are that:
the invention adopts a multi-sensor fusion technology to effectively fuse the data acquired by the vision, laser, millimeter wave radar and the sensing unit of the beacon reader, so that the detection function of the obstacle in front of the unmanned train and the auxiliary anti-collision function are realized.
Drawings
FIG. 1 is a train consist diagram;
FIG. 2 is a mounting block diagram of the present invention;
fig. 3 is a block diagram of a foreign matter intrusion detection function of the present invention;
FIG. 4 is a velocity versus graph of different deep learning algorithms;
FIG. 5 is a schematic diagram of sensor fusion;
fig. 6 is a schematic diagram of target tracking.
Detailed Description
The technical scheme of the present invention will be described in detail below with reference to the accompanying drawings.
The invention discloses an obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion. Comprising the following steps:
1. foreign matter intrusion monitoring early warning alarm model:
the rail transit foreign matter intrusion monitoring early warning and alarming model comprises a high-definition camera video detection technology and a binocular stereoscopic vision technology. In the system, a camera is arranged in a cab of a running vehicle, and whether foreign matter invasion occurs is analyzed by collecting and processing video information of scenes around a track.
The rail transit foreign matter intrusion detection system based on video processing can be divided into four parts according to functions: the data acquisition module, the foreign matter intrusion detection module, the data transmission module and the remote monitoring center are designed to have the overall structure shown in figure 3.
(1) The image acquisition module comprises a group of Ethernet cameras and millimeter wave radars. The module realizes the data acquisition function. The long and short Jiao Gaoqing camera module can acquire video images of the front part of the train in operation, and the laser radar acquires point cloud images of the front part. And the collected data is transmitted to a computer for relevant processing or is transmitted back to a monitoring end for standby.
(2) The foreign matter intrusion detection module is a core component of the system, and realizes functions of image preprocessing, later analysis and the like by writing effective program codes. The main function realized by the foreign matter intrusion detection module is to judge whether the intrusion foreign matter exists through the analysis of the image data, the process is mainly realized through a software algorithm, and the detection algorithm can be divided into background image extraction and updating, image area classification, foreground image extraction and classification and suspicious foreground danger degree judgment (namely foreign matter confirmation).
(3) And a data transmission module. The module mainly realizes remote communication of image data through a vehicle-ground WLAN. The video image obtained by the camera can be directly transmitted to a remote monitoring center, and the alarm image information transmission can be extracted through a foreign matter intrusion detection module.
(4) The remote monitoring center includes a host server and a monitor. Programming a program in a foreground of a host, building a database in a background, and establishing a human-computer interaction interface, wherein the interaction system has the following realized functions: displaying the position of the warning intrusion foreign matter by using the image, performing warning processing, storing a warning record and inquiring the warning record. All information data related to the alarm foreign matters are stored in a server database, so that the later inquiry and call of alarm records are convenient. The monitor can call out video recordings of accident high incidence zones at any time, and double monitoring is carried out on the heavy point areas.
2. Image target detection technology based on deep learning:
extracting the obstacle, and classifying the obstacle by using a Yolo (target positioning and detection) based detection technology. Yolo (target positioning and detection) is one of the most commonly used algorithms of the target detection technology, compared with other target detection algorithms, yolo (target positioning and detection) ensures the detection precision, and meanwhile, the detection speed is the fastest, so that the method is most suitable for the use scene of real-time detection of the project train in the running process.
Unlike conventional recognition methods, the neural network in the Yolo (object localization and detection) model initially divides the acquired image into a number of small grids of side a×a, and then each detects which center points fall within the small grids, which in turn identifies the objects in that grid. YOLO v3 (object localization and detection) can separate the cases where an object belongs to multiple tags by improving multi-scale prediction and better underlying classification network and Logistic (regression analysis) classifier compared to previous versions, while detection of both large and small objects is improved by 32/16/8-fold field of view selection. This allows for less identification of items in pixels of the target at a far distance. As shown in fig. 4; after the target detection signal is subjected to a YOLOV 3 algorithm of deep learning target detection, outputting position information (center point x, y, target length w, target width h) and target type signals in a re-image of the detection target, outputting the signals for judging whether an obstacle invades the track range or not, and transmitting the signals into a sensor fusion node for further processing and judgment.
3. Sensor fusion technology:
the sensor fusion is an important link in the comprehensive sensor data processing, and because of the influence of various noise clutter and other non-ideal factors, the obtained data of the sensor has noise influence, and the effective measurement is likely to consider that a plurality of targets exist, so that the corresponding relation between the effective measurement and the targets needs to be established, and the target data is matched, thereby establishing the target effective relation and improving the reliability of target identification.
The target fusion process is a process of determining the corresponding relation between the measurement information received by the sensor and the target source, wherein two most critical objects are a multi-sensor and a multi-target. The main technology for developing the fusion of targets is how to unify the time sequences of real-time data of different sensors, and if multi-sensor data are used for evaluating the real characteristic parameters of the identified targets, reliable basis is provided for the subsequent target classification. As shown in fig. 5; the image signals obtained by the camera are processed by the processing target detection processing to obtain obstacle position information, millimeter wave radar information and laser radar point cloud information, and after the processing of the fusion algorithm, alarm signals of whether the obstacle invades the track or not are output, and the output signals are used for giving an alarm to the cab.
4. Target tracking technology:
when multi-target detection occurs, due to the phenomena of short frame loss or shielding of targets and the like, the continuity of target tracking can be improved by using algorithms such as deep start (multi-target tracking) and the like, so that the interference of different environmental factors such as external illumination and the like on cameras is overcome, and the detection of light and shade alternate scene targets under the illumination of lamplight in a tunnel is improved. As shown in fig. 6; after the image video signals are processed by the Deep sort algorithm, the position information, the movement direction information, the current speed information and the like of the tracked single obstacle are output, and the output signals are used for predicting whether the current speed and direction of the obstacle invade the track range of train running or not, and making advanced judgment and alarm.
5. Video image enhancement techniques:
the video enhancement function is the basis of video processing, and video processing (such as environment monitoring, evaluation and early warning, train anti-collision reminding and the like) can be performed on the basis of video enhancement, so that the influence of external factors such as haze, light and the like on video identification can be effectively eliminated.
The video image enhancement mainly enhances and displays or highlights the contrast, color, contour and other image object characteristics of the video image, so that the processed image is convenient for further processing and analysis. The current video enhancement methods can be classified into 2 types, namely, one type of method is to improve the hardware performance of the monitoring camera, such as adopting a fog-penetrating lens. Another type of method is to enhance the acquired degraded video by using an enhancement algorithm, and because the video is also a sequence image, many image enhancement algorithms can be applied to the video enhancement field, such as spatial domain methods like gray level transformation and histogram transformation, filtering methods based on frequency domain, and the like. The video enhancement process of the method will be described in detail below using a histogram equalization method as an example.
The histogram equalization method of the image is to change the histogram distribution of the image and adjust the scattered areas with relatively concentrated pixels so as to achieve the effect of homogenizing the histogram, thereby improving the contrast of the whole image. At this time, by converting the histogram of the original image into a basic direction idea of a uniform distribution form, the luminance can be better distributed on the histogram. In addition, the method can be used for enhancing local contrast without affecting the overall contrast. Therefore, the function of histogram equalization can be realized by expanding the common brightness. This approach is more applicable to images where the overall style is so bright or dark that the details of the overall picture are not visible. In addition, the histogram equalization method is a simple and feasible reversible operation, and the calculated amount is small.
The train can pass through various complex environments in operation, so that the driving of the train can be influenced to a certain extent, the video enhancement function provides a display system in the cab, environmental factors influencing the driving are eliminated through a software algorithm, the scene in front of the train is displayed on the display system in the cab as clearly as possible, and the driving of the train is assisted. The following 3-point functions are mainly implemented by using 2 methods to process and reduce the influence:
1. the contrast of the image data is calculated, if the contrast is lower than a certain threshold range, the contrast of the image is adjusted by a function in opencv, and if the brightness is higher than or lower than a certain threshold range, the value of each pixel is added or subtracted to improve the brightness of the whole image or reduce the brightness.
2. The penetrability effect of the laser radar is utilized, the influence of the image on the detection effect is reduced, and whether an obstacle invades the track range is judged in an auxiliary mode according to fusion of the image and the laser radar point cloud data.
Haze removal
Haze can cause great influence on the train in an overhead traveling area. Once haze occurs in a large range, a subway company has to limit speed operation of a train in an overhead traveling area, and the operation efficiency of the whole subway is greatly influenced.
In addition, even under meteorological conditions where haze is not heavy, some areas on the overhead may have a cluster-like haze generated, so that a certain technical means is required objectively to counter the influence of haze on train running.
The haze removing function of the system is mainly to remove haze influencing driving in an overhead driving area through a software algorithm, so that the definition of a front scene in a monitoring video is effectively improved.
Strong light resistance
Sunlight in sunny days has great influence on driving of an overhead running train, and the strong light resistant function of the system can restore a scene in front on a display screen as much as possible.
Low light intensity illumination
The image enhancement processing is a technical method for highlighting the edge (namely, the boundary line of the image tone mutation or the ground object type) with larger difference of brightness values (or tone) of adjacent pixels (or areas) of an image (or image). The image after the edge enhancement can more clearly display the boundaries of different object types or phenomena or the trails of the linear images, so that the identification of different object types and the delineation of the distribution range of the different object types are facilitated.
6. Vibration resistant technology:
during train travel, various modes of vibration may occur due to the insufficient smoothness of the track. In severe cases, the large vibration may misjudge an object that is not originally in the traveling direction as an object in the traveling direction, and thus misjudgment occurs. For this purpose, a special anti-vibration algorithm is needed:
the base point method. The method is characterized by searching a reference point in the video and taking the reference point as a base point for judgment. In view of the possibility of side-to-side twisting in addition to up-and-down vibration, two reference points are required.
All envelopes are referenced to the rail during the travel of the train. So the rail is used as a reference point to resist the vibration of the running train. And restoring the smooth video through a base point jitter removing algorithm.
As shown in fig. 1, tc vehicle is a trailer with a driver, mp vehicle is a motor vehicle with a pantograph, and M parking space is a motor vehicle without a pantograph.
The obstacle sensing system project based on the integration of vision, millimeter wave and multi-line laser has the functions of train positioning and tracking, obstacle detection in front of train operation and auxiliary anti-collision, can give an alarm and prompt in a cab and an OCC according to an application scene, and stores and uploads a video/image to the OCC for the OCC to remotely retrieve a field image.
The obstacle sensing system comprises a system host, a long-short focal camera, a millimeter wave radar, a secondary radar and a beacon reader.
As shown in fig. 2:
and the TC vehicles at the two ends are respectively provided with a system host, and the host is communicated with the train control monitoring system through a train real-time data protocol interface protocol.
The heads of the TC vehicles at the two ends are respectively provided with a long-short focal length camera and a millimeter wave radar, so that obstacle detection is realized.
The secondary radars are arranged at the head parts of the TC heads at the two ends, and the beacon readers are arranged at the bottoms of the TC heads, so that auxiliary anti-collision detection is realized.
Aiming at the front obstacle found in the real-time running of the train, the information of the front running line is acquired in real time through the intelligent environment sensing technology, and the obstacle is intelligently identified for early warning and prompting, so that the risk that the front obstacle influences the normal running of the train is reduced.
All images or videos may be stored locally or uploaded in real time to the OCC for remote retrieval of live images by the OCC.
The system host is 3U in size and height and is arranged in a Tc car electrical cabinet, and mainly used electrical interfaces comprise a power interface, an RS485 communication interface, a CAN communication interface, an Ethernet interface, a driving and driven input and output interface and an Ethernet communication interface of a real-time data protocol interface of a train. The host is powered by a DC110V power supply.
The long-short-focus binocular camera is arranged above the built-in roof of the locomotive and is used for detecting obstacles in front of the train. The long-short-focus binocular camera is powered by POE, and the electrical interface is a DB9 interface.
The millimeter wave radar is mounted on a Tc vehicle.
The secondary radar is mounted on the Tc vehicle.
The beacon reader is mounted at the Tc underbody.
The system can project the area of the obstacle at the position which is not less than 400m on the straight track to be not less than 0.25m 2 The obstacle with the size is effectively identified and prompted to give an alarm. The 400 meter visual distance is the maximum distance that can be identified at a straight line segment.
The identification of the infringement obstacle is not affected by different light conditions such as tunnels, overhead and the like, and the time from the detection of the obstacle to the alarm is less than 1 second.
After the system detects the obstacle, the abnormal situation is transmitted to the train control monitoring system through the interface of the real-time data protocol interface of the train according to the defined threat level information, and the train control monitoring system adopts preliminary countermeasures according to the defined threat level. Meanwhile, the video and the processed pictures with the obstacle distance information are transmitted to an OCC/driver through train-ground communication, after the OCC/driver further judges, whether the pictures are obstacles or not is determined, whether the pictures are braked or not is transmitted to a train control monitoring system, if the pictures are obstacles, the pictures are braked, the operators are arranged to clear the obstacles, and otherwise, the speed is recovered.
The acquisition unit of the auxiliary anti-collision unit consists of a radar antenna and an RFID tag reader. Through radar communication technology, realize crashproof function and the headwall crashproof function of ground circuit between the train. The system will give a graded alarm depending on the size and distance of the obstacle detected. The distance and relative speed between two trains, or the distance of a train from a particular (fixed) target host and the absolute speed of the train, can be measured. The system will give a hierarchical alarm based on the detected distance.
The vehicle-mounted electronic tag reader can read the RFID electronic tag on the line, so that the system obtains the current position of the train and judges the state of the train. The acquired information comprises whether the train is positioned on a positive line, a parking line, an up-down state and the like, and the headwall anti-collision unit automatically adjusts the system operation mode according to the information.
Functions implemented on the vehicle:
1. when 400m from the front vehicle (end wall), the video links to the center in FAM mode, the HMI buzzer sounds intermittently in non-FAM mode, and the video links on CCTV screen.
2. 200m from the front vehicle (end wall), video linkage to the center in FAM mode, HMI buzzer jerk in non-FAM mode, HMI on-screen alarm, CCTV on-screen video linkage.
Front obstacle detection static test effect: as shown in table 1;
TABLE 1 front obstacle detection static test effect table
The dynamic test effect of the front obstacle detection is shown in table 2:
TABLE 2 dynamic test Effect table for front obstacle detection
The auxiliary anti-collision test effect is shown in table 3:
TABLE 3 auxiliary anti-collision test effect table
The embodiments are only for illustrating the technical idea of the present invention, and the protection scope of the present invention is not limited by the embodiments, and any modification made on the basis of the technical scheme according to the technical idea of the present invention falls within the protection scope of the present invention.
Claims (8)
1. The obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion is characterized by comprising a system host installed in a cab, an obstacle sensing unit connected with the system host and a train control monitoring system; the obstacle sensing unit collects train position information, front video images and three-dimensional point cloud images, senses and detects obstacles in a rail train operation scene, transmits the collected information to the system host, and carries out classification analysis on the received information; the projection area of the obstacle at 400m or more to the straight track is not less than 0.25m 2 Identifying the size of the infringement barrier; after detecting the obstacle, the system host transmits the abnormal situation to the train control monitoring system through the defined threat level information and the train real-time data protocol interface, and the train control monitoring system takes countermeasures according to the defined threat level and transmits the video and the processed pictures with the obstacle distance information to the ground control center and the driver through train-ground communication.
2. The obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion according to claim 1, wherein the obstacle sensing unit is provided with a binocular camera sensing module, comprises a module consisting of two cameras with long focus and short focus, is arranged above a roof in a cab, is powered by POE, and is provided with an electrical interface as a DB9 interface; and acquiring front high-definition stereoscopic image information, and detecting a front obstacle of the train.
3. The obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion according to claim 1, wherein the obstacle sensing unit is provided with a multi-line laser sensing module, comprises multi-line laser radars arranged at two ends of a vehicle head, acquires point cloud images of a front obstacle and a vehicle in a bidirectional manner, and detects the vehicle in front.
4. The obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion according to claim 1, wherein the obstacle sensing unit is provided with an auxiliary anti-collision module, the auxiliary anti-collision module comprises vehicle-mounted secondary radar equipment, ground beacon equipment, a vehicle-mounted electronic tag reader and a ground RFID beacon, the vehicle-mounted secondary radar equipment obstacle sensing unit reads the ground beacon, the position of a train, the distance and the relative speed between two trains on the same line are obtained, whether the train is positioned on a positive line or a parking line and the up-down state of the train or not is obtained, and the classification alarm is carried out and the train running mode is adjusted according to the detected distance.
5. The obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion according to claim 1, wherein a sensor fusion module is arranged in a system host, the time sequence of real-time data collected by an obstacle sensing unit is unified, image signals obtained by a camera are processed, obstacle position information, millimeter wave radar information and laser radar point cloud information obtained by processing target detection are processed, an alarm signal of whether an obstacle invades a track is output after the processing of a fusion algorithm, and the output signal is used for alarming a cab.
6. The obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion according to claim 1, wherein a target tracking unit is arranged in a system host, an image video signal is processed through a Deep sort algorithm, position information of a single tracked obstacle, movement direction information of the obstacle and current obstacle speed information are output, whether the speed and direction of the current obstacle invade a track range of train running or not is predicted according to the output information, and advanced judgment and alarm are made.
7. A vision, radio frequency positioning and multi-line laser fusion-based obstacle sensing system according to claim 3, wherein a video image enhancement module is arranged in the system host, the contrast of image data is calculated, if the contrast is lower than a certain threshold range, a function in opencv is adopted to adjust the contrast of the image, and if the brightness is higher than or lower than a certain threshold range, the value of each pixel is adjusted to adjust the brightness of the whole image; and meanwhile, according to fusion of the image and the multi-line laser radar point cloud data, whether an obstacle invades the track range is judged in an auxiliary mode.
8. The obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion according to claim 1, wherein an anti-vibration module is arranged in the system host, and the obstacle sensing system is judged by searching a reference point in a front video image and taking the reference point as a base point; selecting two base points in consideration of up-and-down vibration and left-and-right twisting; in the running process of the train, the rail is used as a reference point, and a smooth video is restored through a base point jitter removing algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211143675.2A CN116461571A (en) | 2022-09-20 | 2022-09-20 | Obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211143675.2A CN116461571A (en) | 2022-09-20 | 2022-09-20 | Obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116461571A true CN116461571A (en) | 2023-07-21 |
Family
ID=87182979
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211143675.2A Pending CN116461571A (en) | 2022-09-20 | 2022-09-20 | Obstacle sensing system based on vision, radio frequency positioning and multi-line laser fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116461571A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104386092A (en) * | 2014-10-21 | 2015-03-04 | 卡斯柯信号有限公司 | Automatic train protection system and method based on image identification and multi- perception fusion |
| JP2016088183A (en) * | 2014-10-31 | 2016-05-23 | 株式会社Ihi | Obstacle detection system and railway vehicle |
| CN106741008A (en) * | 2016-12-29 | 2017-05-31 | 北京交通大学 | Rail track method for recognizing impurities and system |
| CN109298415A (en) * | 2018-11-20 | 2019-02-01 | 中车株洲电力机车有限公司 | A kind of track and road barricade object detecting method |
| CN113568002A (en) * | 2021-06-24 | 2021-10-29 | 中车南京浦镇车辆有限公司 | Rail transit active obstacle detection device based on laser and image data fusion |
| CN114379611A (en) * | 2021-12-03 | 2022-04-22 | 合肥湛达智能科技有限公司 | Freight train driving obstacle early warning method and system |
-
2022
- 2022-09-20 CN CN202211143675.2A patent/CN116461571A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104386092A (en) * | 2014-10-21 | 2015-03-04 | 卡斯柯信号有限公司 | Automatic train protection system and method based on image identification and multi- perception fusion |
| JP2016088183A (en) * | 2014-10-31 | 2016-05-23 | 株式会社Ihi | Obstacle detection system and railway vehicle |
| CN106741008A (en) * | 2016-12-29 | 2017-05-31 | 北京交通大学 | Rail track method for recognizing impurities and system |
| CN109298415A (en) * | 2018-11-20 | 2019-02-01 | 中车株洲电力机车有限公司 | A kind of track and road barricade object detecting method |
| CN113568002A (en) * | 2021-06-24 | 2021-10-29 | 中车南京浦镇车辆有限公司 | Rail transit active obstacle detection device based on laser and image data fusion |
| CN114379611A (en) * | 2021-12-03 | 2022-04-22 | 合肥湛达智能科技有限公司 | Freight train driving obstacle early warning method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101510356B (en) | Video detection system and data processing device thereof, video detection method | |
| CN105809679B (en) | Mountain railway side slope rockfall detection method based on visual analysis | |
| US7366325B2 (en) | Moving object detection using low illumination depth capable computer vision | |
| US11380105B2 (en) | Identification and classification of traffic conflicts | |
| CN111198371A (en) | Forward-looking obstacle detection system | |
| CN105744232A (en) | Method for preventing power transmission line from being externally broken through video based on behaviour analysis technology | |
| CN104778444A (en) | Method for analyzing apparent characteristic of vehicle image in road scene | |
| CN109703460A (en) | The complex scene adaptive vehicle collision warning device and method for early warning of multi-cam | |
| CN111507237A (en) | Video-based abnormal event detection method and device applied to expressway | |
| CN107274678B (en) | A kind of night vehicle flowrate and model recognizing method based on Kinect | |
| Wu et al. | A Real‐Time Embedded Blind Spot Safety Assistance System | |
| CN111461088A (en) | Rail transit obstacle avoidance system based on image processing and target recognition | |
| CN113568002A (en) | Rail transit active obstacle detection device based on laser and image data fusion | |
| CN102426785A (en) | Traffic flow information perception method based on contour and local characteristic point and system thereof | |
| Chen et al. | Real-time approaching vehicle detection in blind-spot area | |
| CN112101316B (en) | Target detection method and system | |
| CN109447090A (en) | A kind of shield door obstacle detection method and detection system | |
| CN111667655A (en) | Infrared image-based high-speed railway safety area intrusion alarm device and method | |
| CN115083088A (en) | Railway perimeter intrusion early warning method | |
| CN114415171A (en) | Automobile travelable area detection method based on 4D millimeter wave radar | |
| KR101089029B1 (en) | Crime Preventing Car Detection System using Optical Flow | |
| Kapoor et al. | Deep learning based object and railway track recognition using train mounted thermal imaging system | |
| CN115909223A (en) | Method and system for matching WIM system information with monitoring video data | |
| CN115546738A (en) | Rail foreign matter detection method | |
| CN115600124A (en) | Subway tunnel inspection system and inspection method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |