CN112319552A - Rail car operation detection early warning system - Google Patents
Rail car operation detection early warning system Download PDFInfo
- Publication number
- CN112319552A CN112319552A CN202011269063.9A CN202011269063A CN112319552A CN 112319552 A CN112319552 A CN 112319552A CN 202011269063 A CN202011269063 A CN 202011269063A CN 112319552 A CN112319552 A CN 112319552A
- Authority
- CN
- China
- Prior art keywords
- obstacle
- vehicle
- image
- warning system
- identification
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 57
- 238000000034 method Methods 0.000 claims abstract description 14
- 230000004888 barrier function Effects 0.000 claims abstract description 10
- 230000003287 optical effect Effects 0.000 claims description 22
- 238000012545 processing Methods 0.000 claims description 13
- 238000013528 artificial neural network Methods 0.000 claims description 12
- 230000001681 protective effect Effects 0.000 claims description 12
- 230000033001 locomotion Effects 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 7
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 description 7
- 238000010276 construction Methods 0.000 description 6
- 230000000875 corresponding effect Effects 0.000 description 5
- 230000004927 fusion Effects 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000000007 visual effect Effects 0.000 description 4
- 238000013135 deep learning Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 230000000903 blocking effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 230000003137 locomotive effect Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- HPTJABJPZMULFH-UHFFFAOYSA-N 12-[(Cyclohexylcarbamoyl)amino]dodecanoic acid Chemical group OC(=O)CCCCCCCCCCCNC(=O)NC1CCCCC1 HPTJABJPZMULFH-UHFFFAOYSA-N 0.000 description 1
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 230000001678 irradiating effect Effects 0.000 description 1
- 244000144972 livestock Species 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000011435 rock Substances 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 239000002689 soil Substances 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L15/00—Indicators provided on the vehicle or train for signalling purposes
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61C—LOCOMOTIVES; MOTOR RAILCARS
- B61C17/00—Arrangement or disposition of parts; Details or accessories not otherwise provided for; Use of control gear and control systems
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61G—COUPLINGS; DRAUGHT AND BUFFING APPLIANCES
- B61G3/00—Couplings comprising mating parts of similar shape or form which can be coupled without the use of any additional element or elements
- B61G3/10—Couplings comprising mating parts of similar shape or form which can be coupled without the use of any additional element or elements with coupling heads in the form of hook-like interengaging rigid jaws, e.g. "Willison" type
- B61G3/14—Control devices, e.g. for uncoupling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
- G06T2207/30261—Obstacle
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Automation & Control Theory (AREA)
- Transportation (AREA)
- Closed-Circuit Television Systems (AREA)
Abstract
A track car operation detection early warning system belongs to the technical field of driving safety. The invention aims at the problem that the prior rail car adopts a manual observation mode to carry out early warning and has poor reliability. The method comprises the steps of acquiring an image of an obstacle in front of a vehicle within 300 meters by a zoom camera; identifying a target object according to the obstacle image to obtain an identification result; the identification result comprises an obstacle-free and obstacle identification name; when the identification result is the identification name of the obstacle, calculating to obtain the space position information of the obstacle; calculating according to the space position information of the obstacle and the current state information of the vehicle to obtain a direction adjustment control signal of the zoom camera; and meanwhile, judging whether the barrier generates a safety threat in a protection limit of the vehicle driving direction, and if so, sending an alarm signal. The invention can realize the automatic identification of the obstacle in front of the vehicle.
Description
Technical Field
The invention relates to a detection and early warning system for the operation of a railway vehicle, belonging to the technical field of driving safety.
Background
With the rapid development of railway construction, the running speed of the train is continuously improved, and the requirements on the running safety of the train are higher and higher. The existing train control and ATP system adopts the means of signal blocking, approach response and the like to ensure the driving safety, but the train control system can only monitor the traveling train as a cooperative target, and can not early warn the occurrence of an unexpected and unpredictable non-cooperative target limit invasion event. In addition, when the railway construction work machine vehicle is under the construction work condition, the ground signal system is normally in a closed or failure state, and therefore, the safety monitoring function of the work machine vehicle cannot be achieved.
At present, the safety observation of the running of the train is carried out manually, and the running control is carried out through a ground signal system. The problem that effective distance is insufficient and the influence of energy concentration of operating personnel is caused when people look out manually is solved, and once the energy of the operating personnel is not concentrated, the watching is interrupted. When the ground control system fails or is closed according to a construction maintenance plan, or the vehicle-mounted equipment fails and cannot receive a control signal, serious driving safety hazards can be caused.
The occurrence of a ' 6.01 ' Shijiazhu electric locomotive shunting broken turnout accident, ' a ' 7.23 ' Harbin railway administration industrial machinery vehicle operation interval collision human casualty accident, ' a ' 4.28 ' rubber-economical railway special important traffic accident, ' a ' 10.23 ' Guangdong harmony train collision accident, a ' 7.23 ' temperature range line motor train unit rear-end collision human casualty accident and the like is fully explained, the driving safety is ensured only by means of manual observation and a ground signal control system, and the situation is far from sufficient.
Problems that may occur in train operation include: the over-labor intensity of the driver causes the lack of concentration, the off duty of the driver, the sudden illness of the driver and the passengers, the poor observation condition, the insufficient manual visual observation distance and the like; the situation of falling rocks, collapse and the like of the line caused by geological disasters; in addition, the situation that the ground signal/communication system fails to prompt the driver is also existed, which is very easy to cause driving accidents.
In short, under the condition that the current railway train is developing faster and faster, the locomotive vehicle does not have a matched autonomous active detection system. Therefore, it is impossible to autonomously detect and make a judgment about suspected obstacles in the traveling direction, such as targets of persons, cars/train vehicles, livestock, rockfall, soil heaps, falling trees, and the like, and cases of collapse and the like. The lookout is required to be carried out through the naked eyes during the whole driving operation.
Therefore, an active detection system independent of a ground information system is needed to realize autonomous observation of the train advancing direction, and the active long-distance detection and early warning capability is provided, so that the driving safety is ensured.
Disclosure of Invention
The invention provides a rail car operation detection early warning system, aiming at the problems that the prior rail car operation adopts a manual observation mode to carry out early warning and has poor reliability.
The invention relates to a detection early warning system for the operation of a railway vehicle, which comprises,
the image acquisition module is used for acquiring an image of an obstacle in front of the vehicle within 300 meters by using the zoom camera;
the vehicle-mounted host identification module is used for identifying a target object according to the obstacle image to obtain an identification result; the identification result comprises an obstacle-free and obstacle identification name; when the identification result is the identification name of the obstacle, calculating to obtain the space position information of the obstacle;
the state updating module is used for calculating according to the space position information of the obstacle and the current state information of the vehicle to obtain a direction adjusting control signal of the zoom camera;
and the alarm module is used for judging whether the barrier generates a safety threat in a protection boundary of the vehicle driving direction or not according to the space position information of the barrier and the current state information of the vehicle, and if so, sending an alarm signal.
According to the track car operation detection early warning system, the zoom camera is arranged in an optical cabin at the top end of a car head; the zoom camera carries out azimuth adjustment through the cloud platform, the cloud platform passes through steering wheel drive, the steering wheel is controlled through the azimuth adjustment control signal of zoom camera.
According to the detection and early warning system for the operation of the railway vehicle, the zoom camera can realize the acquisition of images of visible light and invisible light obstacles.
The track car operation detection early warning system is characterized in that the zoom camera performs optical compensation through a high-power infrared lamp arranged at the top end of a car head.
According to the track car operation detection early warning system, the vehicle-mounted host identification module identifies the target object through the deep neural network based on Darknet; the calculating to obtain the space position information of the obstacle comprises: the relative speed, distance and azimuth angle of the obstacle to the vehicle;
the identification result also comprises that the image of the identified obstacle is displayed through a display screen.
According to the rail car operation detection early warning system, the process of identifying the target object by the vehicle-mounted host identification module comprises the following steps:
resolving the obstacle image into an image data stream of 30 frames/s by the vehicle-mounted host through a hardware encoder, and performing frame extraction processing on the image data stream according to the data processing speed of the deep neural network based on Darknet to obtain an extracted image frame image; sending the extracted picture frame images into a deep neural network based on Darknet for obstacle identification and comparison, and identifying and marking objects needing protection on the extracted picture frame images; and the vehicle-mounted host machine then obtains the depth of field of the space environment of the image of the obstacle through affine transformation algorithm according to the pixel information of the extracted image frame image and the current position and angle information of the zoom camera by fitting, and places the marked object to be protected into the depth of the space environment to obtain the space position information of the obstacle relative to the vehicle.
According to the detection and early warning system for the operation of the railway vehicle, the alarm module calculates and obtains the relative position and the relative speed of the obstacle and the vehicle according to the space position information of the vehicle relative to the obstacle, and judges whether the obstacle threatens the operation of the vehicle or not by combining with the track identification.
According to the detection and early warning system for the operation of the railway vehicle, a GPS module is further arranged in the optical cabin and used for calculating the longitude and latitude, the speed, the altitude data and the time information of the vehicle in real time.
According to the track car operation detection early warning system, the zoom camera adopts a T10X-pro pan-tilt integrated camera, and the integrated camera adopts a 5-time optical zoom movement and a 20-time optical zoom movement of Haokawav.
According to the detection and early warning system for the operation of the railway vehicle, the warning signal comprises the following two forms:
1) the obstacle invades into the protective limit of the vehicle driving direction, but the distance between the obstacle and the vehicle exceeds the vehicle brake control safety distance, and the alarm signal is a voice danger prompt until the obstacle is out of the protective limit;
2) the obstacle invades the protective limit of the vehicle driving direction, but the distance between the obstacle and the vehicle is within the safe distance of vehicle brake control, and the alarm signal is a collision warning prompt until the obstacle is out of the protective limit or the driver manually releases the alarm signal.
The invention has the beneficial effects that: the system has long-distance and large-range detection capability, can observe and track suspected obstacles in front of the running train, can calculate according to the relative distance, the relative speed, the azimuth angle and the train speed of a tracked target, can prompt a driver to remind the driver, and adopts vehicle braking operation.
The system can still scan and track the front target under the condition of poor visual observation, and assist the operation behavior of a driver. The vehicle picking and hanging assisting device can also have a vehicle picking and hanging assisting function, and provides the relative distance and speed of picking and hanging vehicles for a driver in real time through image acquisition, processing and recognition so as to facilitate the driver to control the vehicle.
The method adopts an automatic obstacle identification mode to replace the existing mode of manually observing and judging the obstacle, so that the identification result is more reliable, the labor cost is saved, and the efficiency is high.
Drawings
Fig. 1 is a flow chart of the detection and early warning system for the operation of the rail car.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the embodiments and features of the embodiments may be combined with each other without conflict.
The invention is further described with reference to the following drawings and specific examples, which are not intended to be limiting.
In a first specific embodiment, shown in fig. 1, the invention provides a detection and early warning system for operation of a rail vehicle, which includes an image acquisition module, configured to acquire an image of an obstacle in front of the vehicle within 300 meters of a traveling vehicle through a zoom camera;
the vehicle-mounted host identification module is used for identifying a target object according to the obstacle image to obtain an identification result; the identification result comprises an obstacle-free and obstacle identification name; when the identification result is the identification name of the obstacle, calculating to obtain the space position information of the obstacle;
the state updating module is used for calculating according to the space position information of the obstacle and the current state information of the vehicle to obtain a direction adjusting control signal of the zoom camera;
and the alarm module is used for judging whether the barrier generates a safety threat in a protection boundary of the vehicle driving direction or not according to the space position information of the barrier and the current state information of the vehicle, and if so, sending an alarm signal.
In the embodiment, the detection distance is 300m, and the method can be used for shunting operation and construction operation of self-wheel running special equipment. According to the investigation result, for special equipment (such as a work machinery vehicle, a rail car, a contact network detection vehicle and the like) operated by a work, power supply and self-wheel operation according to the rule of a running rule, the shunting speed does not exceed 40km/h, and the detection distance is more suitable for 300 meters in consideration of the hardware cost.
The application scenarios of the present embodiment include:
1) and (3) normal interval operation: the method comprises the steps of main line running, station main line/side line passing, bridge tunnel passing, curve and small-radius curve traveling.
2) Shunting in a normal station: comprises the operations of propulsion/traction shunting, turnout/turnout group traveling and vehicle connection/disassembly.
3) Block/station blocking operation: including the driving/shunting operation and the construction operation without ground signals.
4) Driving under a dangerous environment: the method comprises the step of providing early warning support for the rescue train/rescue machine entering the rescue operation environment.
The system in the embodiment has long-distance and large-range detection capability, can observe and track suspected obstacles in front of the running train, can calculate according to the relative distance, the relative speed, the azimuth angle and the train speed of a tracked target, can prompt a driver to remind the driver, and can perform vehicle braking operation.
The vehicle-mounted host recognition module is a pre-trained module, and the training target objects of the vehicle-mounted host recognition module comprise learned recognizable target objects: such as rolling stock, people, cars, trees, poles, platforms and other buildings, etc. The vehicle-mounted host identification module can also identify a new target, namely, a target which is not recorded yet is added to the special data set through identification.
The image acquisition module of the system is arranged in a mode that the front target can still be scanned and tracked under the condition of poor visual observation, and the operation behavior of a driver is assisted.
The system can also have a vehicle picking and hanging assisting function, and relative distance and speed of picking and hanging vehicles are provided for a driver in real time through image acquisition, processing and recognition so as to facilitate the driver to control the vehicle; meanwhile, the image of the connection part is displayed, so that a driver can judge the connection condition of the car coupler, the air pipeline and the electric circuit, and the operation of the driver can be assisted. The personal safety of connecting operation personnel is ensured, and the connection of the car coupler, the air pipeline and the electric circuit is normal and safe.
The embodiment can also comprise log record equipment storage, including vehicle speed, pressure, longitude and latitude, altitude, personnel operation records, warning information, video record and the like, and can analyze the record dump.
When the detection early warning system works, the vehicle-mounted host computer realizes the perception of the environment by reading the image data of the obstacles of the zoom camera, transmits the read images to the vehicle-mounted host computer recognition module, realizes the autonomous recognition of the target by calling the deep neural network, calls the state updating module according to the recognition result, calculates corresponding control information, is used for adjusting the steering engine of the pan-tilt, the focal length of the camera and the like, and realizes the further detection and tracking of the target.
Further, the zoom camera is arranged in an optical cabin at the top end of the vehicle head; the zoom camera carries out azimuth adjustment through the cloud platform, the cloud platform passes through steering wheel drive, the steering wheel is controlled through the azimuth adjustment control signal of zoom camera.
Still further, the zoom camera can realize visible light and invisible light obstacle image acquisition.
And further, the zoom camera performs optical compensation through a high-power infrared lamp arranged at the top end of the vehicle head. When the vehicle is driven at night, the zoom camera is optically compensated by the high-power infrared lamp, so that the quality of pictures shot by the camera can be improved.
Still further, the vehicle-mounted host recognition module performs target object recognition through a deep neural network based on Darknet; the calculating to obtain the space position information of the obstacle comprises: the relative speed, distance and azimuth angle of the obstacle to the vehicle;
the identification result also comprises that the image of the identified obstacle is displayed through a display screen.
The vehicle-mounted host recognition module processes image data provided by the zoom camera, and judges whether a target object exists or not (known or unknown) and the position (relative speed, distance and azimuth angle) of the target object exist in the obstacle image through a deep neural network based on Darknet. And comparing and identifying the suspected targets of the images, integrating and processing the data to form a superposed image containing identification result information, and displaying the superposed image through a display screen. Meanwhile, the alarm module can judge whether a threat target exists in the protection limit of the running direction of the alarm module according to the current vehicle state information and the target identification result information so as to avoid the danger of the train during running.
Still further, the process of identifying the target object by the vehicle-mounted host identification module comprises:
resolving the obstacle image into an image data stream of 30 frames/s by the vehicle-mounted host through a hardware encoder, and performing frame extraction processing on the image data stream according to the data processing speed of the deep neural network based on Darknet to obtain an extracted image frame image; sending the extracted picture frame images into a deep neural network based on Darknet for obstacle identification and comparison, and identifying and marking objects needing protection on the extracted picture frame images; and the vehicle-mounted host machine then obtains the depth of field of the space environment of the image of the obstacle through affine transformation algorithm according to the pixel information of the extracted image frame image and the current position and angle information of the zoom camera by fitting, and places the marked object to be protected into the depth of the space environment to obtain the space position information of the obstacle relative to the vehicle.
The vehicle-mounted host recognition module obtains the continuous spatial position of the locked target in a mode of extracting key frames at the same time every time, so that the relative position and the relative speed of the target object and the vehicle are calculated, and in combination with track recognition, whether the detected target object has collision threat on the running vehicle is finally judged, and early warning or warning is given out according to the relative speed.
The objects needing protection comprise people, vehicles, machines and other objects.
And further, the alarm module calculates and obtains the relative position and the relative speed of the obstacle and the vehicle according to the space position information of the vehicle relative to the obstacle, and judges whether the obstacle threatens the safety of the vehicle or not by combining with the track identification.
Specific application examples of the present embodiment include:
firstly, carrying out equipment self-inspection; then recording the driver information; and then the working mode is selected.
The working modes include a daytime mode, a night mode and a continuous mode, and are specifically described as follows:
the diurnal modes include: preferentially acquiring an image of a barrier in a visible light mode, and ensuring the detection of a space environment with the vehicle running direction of 300 meters by identifying and tracking a target object in a detection distance; the alarm module preferentially judges and outputs according to the visible light information;
the night mode includes: preferentially acquiring an image of a barrier in a non-visible light mode, and irradiating and tracking a target object in a detection distance by an infrared active detection technology to ensure the detection of a space environment with a vehicle running direction of 300 meters; the alarm module preferentially judges and outputs according to the invisible light information;
the hitching mode includes: all module functions of the system are used for vehicle connection, including switching the host computer picture to the vehicle connection place until the connection operation is finished.
When the data processing result of any working mode prompts that a suspected threat target is found, the prompting information of the alarm module comprises two forms:
one) prompt information: when a target invades a vehicle running direction safety limit in the detection range and the distance between the target and the threatening target exceeds a vehicle braking control safety distance, voice prompt can be carried out on drivers and passengers and the target object is continuously tracked until the barrier exits the safety limit range;
second) alarm and control actions:
when a target invades a vehicle running direction safety limit in the detection range and the distance between the target and the threatening target is less than or equal to a vehicle brake control safety distance, a collision warning action is sent to drivers and passengers until the target is separated from the vehicle brake control range or the driver manually removes the target collision threat warning. And then records the information in the process.
Until the end.
Still further, the optical cabin is also internally provided with a GPS module for resolving longitude and latitude, speed, altitude data and time information of the vehicle in real time.
Still further, the alarm signal includes the following two forms:
1) the obstacle invades into the protective limit of the vehicle driving direction, but the distance between the obstacle and the vehicle exceeds the vehicle brake control safety distance, and the alarm signal is a voice danger prompt until the obstacle is out of the protective limit;
2) the obstacle invades the protective limit of the vehicle driving direction, but the distance between the obstacle and the vehicle is within the safe distance of vehicle brake control, and the alarm signal is a collision warning prompt until the obstacle is out of the protective limit or the driver manually releases the alarm signal.
The system of the present invention is further described below:
1) a detection part:
the detection sensor part is composed of a visible light/invisible light detection and image operation part. The visible/invisible optical detection system is integrated, and clear environment images at night in the daytime are acquired, so that the effectiveness of the image identification system is ensured; calculating the operation tracking angle of the camera according to the recognition result, sending an angle control instruction to the holder control module by the host, carrying out angle feedback by the holder control module according to a magnetic encoding shaft in the motor system, carrying out closed-loop control on the tracking direction angle, and carrying out long-distance tracking observation on the driving line by combining a high-speed zooming tracking camera; and the identification algorithm part tracks and identifies cooperative/non-cooperative targets in the running direction in the images returned by the camera through an AI (intelligent object identification) algorithm, calculates the spatial position of the target according to the camera installation information, and tracks and acquires speed information of the appeared suspected target by carrying out the same identification calculation on a plurality of continuous images. Finally, determining whether the currently detected suspected target threatens the driving safety through an information fusion alarm module, and taking corresponding alarm measures; meanwhile, the GPS module built in the optical cabin can be used for resolving longitude and latitude, speed, altitude data and time information of the vehicle in real time to assist the driving and analysis.
2) A software part:
the software part comprises a data acquisition module, a YOLO 1 target detection module depending on deep learning, a decision-level information fusion alarm module, a display control interaction module and other main modules. The data acquisition module continuously reads the data of the camera, resolves the data into an image sequence, and simultaneously acquires the feedback data of the millimeter wave radar in front of the vehicle in real time for comprehensive judgment and processing of the decision-level information fusion alarm module; the method comprises the steps that a resolved image is used as input by a YOLO 1 target detection module relying on deep learning, a deep neural network is composed of operations such as multi-level convolution operation, pooling operation and an activation function, high-dimensional image data are subjected to nonlinear calculation to obtain the type and the position of an interested target, and type and position information is transmitted to a decision-level information fusion alarm module and used for taking corresponding warning measures for suspected targets; the display control system can comprehensively display various information such as collected data, identification results, positions, warning information and the like, and is used for a user to observe the identification results of the identification system so as to take corresponding feedback measures; the decision-making level information fusion alarm module can accurately prompt and alarm the vehicle co-driver and the driver according to the standards such as the running rule and the skill rule by the identification result of the intelligent target identification module, the radar system feedback information and the optical characteristics of the locked target and combining the self state and the running direction environment of the vehicle.
3) An output section:
carrying out voice prompt on targets (determined by combining the vehicle speed) which are invaded in the running direction but do not form threats according to the relative distance and target information to remind a driver of paying attention; when the tracked target threatens the vehicle, the equipment gives out alarm acousto-optic information, and the driver can remove the alarm information through the alarm button.
Compared with the automobile unmanned technology, the detection early warning system and the automobile unmanned technology both need equipment capable of actively detecting and identifying the surrounding environment, tracking the locked target, judging the threat degree of the target and taking corresponding action; however, the unmanned automobile technology needs to perform omnidirectional scanning, the intelligent driving of the unmanned automobile has high requirements on mechanical control, and the unmanned automobile has no high requirements on detection distance, severe weather influence and the like. The system needs to detect and track the running direction in an ultra-long distance, provides a means for effectively coping with the environment with poor observation conditions, and has no overhigh requirement on the detection range and the full-automatic driving aspect.
The hardware of the detection and early warning system in the embodiment comprises:
the system consists of a host, a display screen, an optical cabin, an infrared lamp, a linked monitoring camera and a warning releasing button.
The host is directly connected with the optical cabin, the display screen, the infrared lamp and the linked monitoring camera at two ends, so as to realize control and function of each part.
The selection of the core hardware comprises:
a host part: in the task of target detection and identification of an optical image, a deep learning-based YoLO 1 target detection method is used, and the method needs to be operated on a host with an NVIDIA GPU computing unit, so that NVIDIA jetson TX2 which is an embedded GPU computing host pushed by Yingwei David and has small volume, appropriate power consumption and suitable severe application environment can be selected according to the actual situation of vehicle-mounted equipment, 256 parallel computing cores are arranged, and through tests, when a YoLO target detection algorithm is operated, the processing time of a single image is less than 100ms (the image size is 1280 × 720), and the system power consumption, the volume, the anti-vibration capability, the cost and other reasons are comprehensively considered, and the system host part uses TX2 as a main computing core, an RTSO-9003 bottom plate and the whole machine volume is 87 × 40 mm.
The main interfaces and performance indexes of the computing core are as follows:
a CPU: a dual-core Denver 2ARM CPU core + a quad-core ARM Cortex-A57 core;
GPU: NVIDIA Pascal Graphics Processing Unit (GPU) architecture (256 CUDA cores);
memory: 8GB 128bit LPDD 459.7Gb/s;
and (3) storing: 32GB eMMC (extensible by MicroSD card);
2 USB3.0 ports (5Gbps, 1A maximum supply current);
2 CAN;
1 USB2.0 (w/OTG);
2 3.3V UARTs;
1 gigabit Ethernet (10/100/1000 BASE-T);
4-path 3.3V bit programmable GPIO;
1 RTC battery interface;
1 HDMI 2.0 interface (max 6Gbps, 24bpp, 4096x2160@60 Hz);
1 SD card interface;
1I 2C interface;
1 fan control interface;
1 3.3V UART remote monitoring port (on/off, ambient temperature, input voltage monitoring);
the size of the board card is 87mm multiplied by 50 mm;
power supply requirements: + 7V- + 19V;
power consumption: maximum 10W;
the working temperature is-40 to +85 ℃.
An optical portion: in the system, the optical assembly is arranged at the top end of the vehicle in the working process, and factors such as vehicle motion, body vibration and the like can cause the influence of blurring, distortion and the like on images collected by the camera. Through tests, 5 times of optical zoom movement and 20 times of optical zoom movement of Haikangwei vision can be selected; under the condition that the installation position is not fixed, the situation that an imaging picture is blurred due to shaking, vibration and the like exists, so that the zooming control is complex, and the response speed is slow. In consideration of the specific application environment and the program control aspect in the using process, a T10X-pro pan-tilt integrated camera can be adopted. The T10X-pro pan-tilt integrated camera is designed aiming at the use environment of the unmanned aerial vehicle, the shutter time is short, and therefore the influence of factors such as carrier motion and vibration on picture distortion is very little.
Meanwhile, the T10X-pro pan-tilt integrated camera has a good angle self-stabilization function, can filter the change of the pointing angle of the camera caused by the shaking and vibration of the vehicle body, and ensures that the visual angle of the camera is not changed; and the vibration reduction rubber ball is flexibly connected with the machine body, so that the influence of vibration on the image is further eliminated.
The main performance indexes are as follows:
although the invention herein has been described with reference to particular embodiments, it is to be understood that these embodiments are merely illustrative of the principles and applications of the present invention. It is therefore to be understood that numerous modifications may be made to the illustrative embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that features described in different dependent claims and herein may be combined in ways different from those described in the original claims. It is also to be understood that features described in connection with individual embodiments may be used in other described embodiments.
Claims (10)
1. A detection early warning system for the operation of a railway vehicle is characterized by comprising,
the image acquisition module is used for acquiring an image of an obstacle in front of the vehicle within 300 meters by using the zoom camera;
the vehicle-mounted host identification module is used for identifying a target object according to the obstacle image to obtain an identification result; the identification result comprises an obstacle-free and obstacle identification name; when the identification result is the identification name of the obstacle, calculating to obtain the space position information of the obstacle;
the state updating module is used for calculating according to the space position information of the obstacle and the current state information of the vehicle to obtain a direction adjusting control signal of the zoom camera;
and the alarm module is used for judging whether the barrier generates a safety threat in a protection boundary of the vehicle driving direction or not according to the space position information of the barrier and the current state information of the vehicle, and if so, sending an alarm signal.
2. The railcar running detection pre-warning system according to claim 1,
the zoom camera is arranged in an optical cabin at the top end of the vehicle head; the zoom camera carries out azimuth adjustment through the cloud platform, the cloud platform passes through steering wheel drive, the steering wheel is controlled through the azimuth adjustment control signal of zoom camera.
3. The railcar running detection pre-warning system according to claim 2,
the zoom camera can realize visible light and invisible light obstacle image acquisition.
4. The rail car operation detection and early warning system as claimed in claim 3, wherein the zoom camera is optically compensated by a high power infrared lamp disposed at the top end of the car head.
5. The railcar running detection pre-warning system according to claim 4,
the vehicle-mounted host recognition module carries out target object recognition through a deep neural network based on Darknet; the calculating to obtain the space position information of the obstacle comprises: the relative speed, distance and azimuth angle of the obstacle to the vehicle;
the identification result also comprises that the image of the identified obstacle is displayed through a display screen.
6. The railcar running detection pre-warning system according to claim 5,
the process of identifying the target object by the vehicle-mounted host identification module comprises the following steps:
resolving the obstacle image into an image data stream of 30 frames/s by the vehicle-mounted host through a hardware encoder, and performing frame extraction processing on the image data stream according to the data processing speed of the deep neural network based on Darknet to obtain an extracted image frame image; sending the extracted picture frame images into a deep neural network based on Darknet for obstacle identification and comparison, and identifying and marking objects needing protection on the extracted picture frame images; and the vehicle-mounted host machine then obtains the depth of field of the space environment of the image of the obstacle through affine transformation algorithm according to the pixel information of the extracted image frame image and the current position and angle information of the zoom camera by fitting, and places the marked object to be protected into the depth of the space environment to obtain the space position information of the obstacle relative to the vehicle.
7. The railcar running detection pre-warning system according to claim 6,
the alarm module calculates and obtains the relative position and the relative speed of the obstacle and the vehicle according to the space position information of the vehicle relative to the obstacle, and judges whether the obstacle threatens the safety of the vehicle or not by combining with the track identification.
8. The railcar running detection pre-warning system according to claim 7,
and a GPS module is also arranged in the optical cabin and used for resolving longitude and latitude, speed, altitude data and time information of the vehicle in real time.
9. The railcar running detection pre-warning system according to claim 8,
the zoom video camera adopts a T10X-pro pan-tilt integrated camera, and the integrated camera adopts a 5-time optical zoom movement and a 20-time optical zoom movement of Haokangwei vision.
10. The railcar running detection pre-warning system according to claim 9,
the alarm signal comprises the following two forms:
1) the obstacle invades into the protective limit of the vehicle driving direction, but the distance between the obstacle and the vehicle exceeds the vehicle brake control safety distance, and the alarm signal is a voice danger prompt until the obstacle is out of the protective limit;
2) the obstacle invades the protective limit of the vehicle driving direction, but the distance between the obstacle and the vehicle is within the safe distance of vehicle brake control, and the alarm signal is a collision warning prompt until the obstacle is out of the protective limit or the driver manually releases the alarm signal.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011269063.9A CN112319552A (en) | 2020-11-13 | 2020-11-13 | Rail car operation detection early warning system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011269063.9A CN112319552A (en) | 2020-11-13 | 2020-11-13 | Rail car operation detection early warning system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112319552A true CN112319552A (en) | 2021-02-05 |
Family
ID=74318864
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011269063.9A Pending CN112319552A (en) | 2020-11-13 | 2020-11-13 | Rail car operation detection early warning system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112319552A (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112977541A (en) * | 2021-03-17 | 2021-06-18 | 上海电气泰雷兹交通自动化***有限公司 | Train protection early warning system based on multi-technology fusion |
CN113269151A (en) * | 2021-06-24 | 2021-08-17 | 中车大连机车车辆有限公司 | Train control method and equipment based on visual identification |
CN113435404A (en) * | 2021-07-14 | 2021-09-24 | 深圳市比一比网络科技有限公司 | Safe auxiliary driving method and system for shunting electric bus based on radar and image detection |
CN113655793A (en) * | 2021-08-12 | 2021-11-16 | 上海三一重机股份有限公司 | Obstacle avoidance control method and device based on fusion strategy and engineering machinery |
CN113793532A (en) * | 2021-08-11 | 2021-12-14 | 岚图汽车科技有限公司 | Remote obstacle lane information sharing method and system |
CN113954930A (en) * | 2021-11-08 | 2022-01-21 | 交控科技股份有限公司 | Automatic unlocking degraded vehicle path method and device based on rear vehicle perception |
CN114550295A (en) * | 2022-02-24 | 2022-05-27 | 陈�峰 | Sensing identification storage terminal system based on big data |
CN114701539A (en) * | 2022-03-09 | 2022-07-05 | 中铁工程装备集团隧道设备制造有限公司 | Horizontal transport locomotive and active safety system thereof |
TWI804113B (en) * | 2021-12-17 | 2023-06-01 | 正修學校財團法人正修科技大學 | Intelligent railway monitoring system and method thereof |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6163755A (en) * | 1996-02-27 | 2000-12-19 | Thinkware Ltd. | Obstacle detection system |
US20190023295A1 (en) * | 2016-01-31 | 2019-01-24 | Rail Vision Ltd | System and method for detection of defects in an electric conductor system of a train |
CN109552366A (en) * | 2018-12-24 | 2019-04-02 | 西安思科赛德电子科技有限公司 | Locomotive vehicle-mounted railway barrier intelligent detecting alarm system and its method for early warning |
CN110248858A (en) * | 2016-12-07 | 2019-09-17 | 西门子交通有限责任公司 | For in rail traffic the image analysis based on track, in particular for method, equipment and the rail vehicle of the image analysis based on rail in railway traffic, especially rolling stock |
CN110532889A (en) * | 2019-08-02 | 2019-12-03 | 南京理工大学 | Track foreign matter detecting method based on rotor unmanned aircraft and YOLOv3 |
CN210047483U (en) * | 2019-04-19 | 2020-02-11 | 黑龙江御林湾科技有限公司 | Train anticollision early warning device |
-
2020
- 2020-11-13 CN CN202011269063.9A patent/CN112319552A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6163755A (en) * | 1996-02-27 | 2000-12-19 | Thinkware Ltd. | Obstacle detection system |
US20190023295A1 (en) * | 2016-01-31 | 2019-01-24 | Rail Vision Ltd | System and method for detection of defects in an electric conductor system of a train |
CN110248858A (en) * | 2016-12-07 | 2019-09-17 | 西门子交通有限责任公司 | For in rail traffic the image analysis based on track, in particular for method, equipment and the rail vehicle of the image analysis based on rail in railway traffic, especially rolling stock |
CN109552366A (en) * | 2018-12-24 | 2019-04-02 | 西安思科赛德电子科技有限公司 | Locomotive vehicle-mounted railway barrier intelligent detecting alarm system and its method for early warning |
CN210047483U (en) * | 2019-04-19 | 2020-02-11 | 黑龙江御林湾科技有限公司 | Train anticollision early warning device |
CN110532889A (en) * | 2019-08-02 | 2019-12-03 | 南京理工大学 | Track foreign matter detecting method based on rotor unmanned aircraft and YOLOv3 |
Non-Patent Citations (1)
Title |
---|
林付春等: "基于深度学习的智能辅助驾驶***设计", 《贵州大学学报(自然科学版)》 * |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112977541A (en) * | 2021-03-17 | 2021-06-18 | 上海电气泰雷兹交通自动化***有限公司 | Train protection early warning system based on multi-technology fusion |
CN113269151A (en) * | 2021-06-24 | 2021-08-17 | 中车大连机车车辆有限公司 | Train control method and equipment based on visual identification |
WO2022267266A1 (en) * | 2021-06-24 | 2022-12-29 | 中车大连机车车辆有限公司 | Vehicle control method based on visual recognition, and device |
CN113435404A (en) * | 2021-07-14 | 2021-09-24 | 深圳市比一比网络科技有限公司 | Safe auxiliary driving method and system for shunting electric bus based on radar and image detection |
CN113793532A (en) * | 2021-08-11 | 2021-12-14 | 岚图汽车科技有限公司 | Remote obstacle lane information sharing method and system |
CN113655793A (en) * | 2021-08-12 | 2021-11-16 | 上海三一重机股份有限公司 | Obstacle avoidance control method and device based on fusion strategy and engineering machinery |
CN113954930A (en) * | 2021-11-08 | 2022-01-21 | 交控科技股份有限公司 | Automatic unlocking degraded vehicle path method and device based on rear vehicle perception |
CN113954930B (en) * | 2021-11-08 | 2024-04-26 | 交控科技股份有限公司 | Automatic unlocking degradation vehicle path method and device based on rear vehicle perception |
TWI804113B (en) * | 2021-12-17 | 2023-06-01 | 正修學校財團法人正修科技大學 | Intelligent railway monitoring system and method thereof |
CN114550295A (en) * | 2022-02-24 | 2022-05-27 | 陈�峰 | Sensing identification storage terminal system based on big data |
CN114701539A (en) * | 2022-03-09 | 2022-07-05 | 中铁工程装备集团隧道设备制造有限公司 | Horizontal transport locomotive and active safety system thereof |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112319552A (en) | Rail car operation detection early warning system | |
US10110795B2 (en) | Video system and method for data communication | |
CN110545380B (en) | Video system and method for data communication | |
CN104442924A (en) | All-weather high speed railway vehicle-mounted obstacle detection system and method | |
US20150268338A1 (en) | Tracking from a vehicle | |
CN112466141A (en) | Vehicle-road-collaboration-oriented intelligent network connection end equipment interaction method, system and storage medium | |
CN104875681A (en) | Dynamic vehicle-mounted camera control method based on application scenarios | |
CN112771858A (en) | Camera assessment techniques for automated vehicles | |
CN103839046A (en) | Automatic driver attention identification system and identification method thereof | |
CN108973861A (en) | A kind of intelligence A column driving safety system | |
CN110667475A (en) | Auxiliary monitoring system and method for blind area of passenger car | |
CN105336217A (en) | Driving safety prewarning system based on machine vision and Android platform | |
CN206383949U (en) | Driving safety system based on the pure image procossings of ADAS | |
CN104157160A (en) | Vehicle drive control method and device as well as vehicle | |
KR101861525B1 (en) | Pedestrian Protection System for autonomous car | |
CN107399341A (en) | On-board running Environmental safety supervision system and method | |
CN112896159A (en) | Driving safety early warning method and system | |
CN109866686A (en) | The intelligent active safety DAS (Driver Assistant System) and method analyzed in real time based on video | |
CN108482253A (en) | A kind of lorry blind zone detection device, method and system | |
CN210733976U (en) | Vehicle driving auxiliary system based on 3D panorama technique | |
CN210047483U (en) | Train anticollision early warning device | |
CN111681383A (en) | Railway locomotive running ground environment sensing technology | |
CN116279673A (en) | Train running line and driving state on-line monitoring system and device | |
CN107292818B (en) | Automatic positioning system and method for line capture device based on panoramic camera | |
CN110908003A (en) | Active barrier detection system based on unmanned driving |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20210205 |