Vehicle positioning method and device

Abstract

The invention provides a vehicle positioning method and device, and the method comprises the steps: collecting first data of a first target feature below a vehicle in a first sensing mode in a first preset range, and comparing the first data with a first reference template, so as to determine the first position information of the vehicle; a second preset range is determined based on the first position information, second data of a second target feature below the vehicle are collected in a second sensing mode in the second preset range, the second data are compared with a second reference template, and second position information of the vehicle is determined; the second preset range is smaller than the first preset range, the positioning precision of the second reference template is higher than the positioning precision of the first reference template, and the positioning efficiency, reliability and calculation efficiency are greatly improved by gradually reducing the search range and improving the precision of the reference templates; the problem that in the prior art, the positioning precision is insufficient due to train stopping position deviation and dynamic stretching and retracting of a train body connecting device is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of robotics, and more particularly to a vehicle positioning method and apparatus. Background Technology

[0002] With the rapid development of my country's rail transit, the operational safety and maintenance efficiency of trains have become crucial to ensuring the lifeline of public transportation. The undercarriage area has a complex structure and numerous components, making it a key focus and challenge for daily maintenance. To replace traditional, high-intensity, and inefficient manual inspections, the use of intelligent inspection robots for automated and intelligent condition monitoring of train undercarriages has become an important trend in the industry. These robots are typically equipped with cameras, lasers, and other sensors, moving along tracks within maintenance trenches to scan and image critical undercarriage components and identify defects, achieving all-weather, high-precision safety assurance.

[0003] In existing technologies, the following positioning methods are commonly used to guide inspection robots to designated inspection locations. One method is fixed beacon-based positioning, where RFID tags, photoelectric switches, or transponders are installed along the track. When a train or robot passes by, a signal is triggered, thus determining its approximate location. Another method is odometer or encoder-based positioning, which calculates the distance traveled by measuring the number of revolutions of the robot's wheels, thereby determining its current position. Additionally, there are attempts to directly utilize vision or laser systems to search for specific feature points over a large area for positioning.

[0004] The aforementioned existing technologies all have significant shortcomings in practical applications, making it difficult to meet the demands of high-precision automated detection. Firstly, the stopping position of a train is not absolutely fixed each time, exhibiting random errors at the centimeter or even decimeter level. Secondly, the connecting devices between carriages, such as couplers and windshields, are flexible, causing dynamic changes in the relative positions of each carriage. These factors render positioning methods based on fixed beacons or odometers severely inaccurate, and odometers also generate cumulative errors that are difficult to eliminate. Directly using a single sensor for large-scale searches is not only computationally intensive and time-consuming, but also prone to failing to find the target when the initial position deviation is large, leading to positioning failures and low reliability. Therefore, a technical solution that can overcome the uncertainty of vehicle stopping and achieve rapid and accurate positioning is urgently needed. Summary of the Invention

[0005] This invention provides a vehicle positioning method and apparatus to address the technical shortcomings of existing technologies that rely on fixed beacons or single sensors for positioning when facing random errors in vehicle parking and dynamic expansion and contraction of vehicle body connecting devices, resulting in low accuracy, poor reliability, and difficulty in stably aligning with the parts to be inspected. This invention achieves fast, high-precision, and robust positioning of the vehicle, thereby ensuring the efficiency and success rate of automated inspection.

[0006] This invention provides a vehicle positioning method, comprising: Within a first preset range, first data of a first target feature under the vehicle is collected using a first sensing method, and the first data is compared with a first reference template to determine the first position information of the vehicle; Based on the first location information, a second preset range is determined. Within the second preset range, a second sensing method is used to collect second data of the second target features under the vehicle. The second data is then compared with a second reference template to determine the second location information of the vehicle. Wherein, the second preset range is smaller than the first preset range, and the positioning accuracy of the second reference template is higher than that of the first reference template.

[0007] According to the vehicle positioning method provided by the present invention, the first sensing method is implemented by a first sensor, the second sensing method is implemented by a second sensor, and both the first sensor and the second sensor are mounted on a mobile device.

[0008] According to the vehicle positioning method provided by the present invention, the step of determining the second preset range based on the first location information specifically includes: controlling the mobile device to adjust its position according to the first location information so that the second target feature is located within the field of view of the second sensor, thereby determining the second preset range.

[0009] According to the vehicle positioning method provided by the present invention, the first sensor is a laser sensor and the second sensor is an image sensor.

[0010] According to the vehicle positioning method provided by the present invention, the first sensing method and the second sensing method are implemented by the same multimodal sensor in different working modes.

[0011] According to the vehicle positioning method provided by the present invention, the first data is compared with a first reference template to determine the first location information of the vehicle, specifically including: The first data is matched with the first reference template to generate a first offset representing the deviation between the vehicle's current position and the reference position; Determine whether the first offset is less than a preset first threshold. If so, use the first offset as the first position information. If not, control the mobile device to adjust its position according to the first offset, and repeatedly execute the step of collecting the first data of the first target feature under the vehicle using the first sensing method until the generated first offset is less than the first threshold.

[0012] According to the vehicle positioning method provided by the present invention, the second data is compared with a second reference template to determine the second location information of the vehicle, specifically including: A template matching algorithm based on deep learning is used to calculate the deviation between the center position of the second target feature in the second data and the preset reference position in the second reference template, and the deviation is used as the second position information of the vehicle.

[0013] According to the vehicle positioning method provided by the present invention, the first target feature is the vehicle axle, and the second target feature is the vehicle motor or brake disc caliper.

[0014] This invention provides a vehicle positioning device, comprising: The first position determination module is used to collect first data of a first target feature under the vehicle using a first sensing method within a first preset range, and compare the first data with a first reference template to determine the first position information of the vehicle. The second position determination module is used to determine a second preset range based on the first position information, collect second data of the second target features under the vehicle using a second sensing method within the second preset range, and compare the second data with a second reference template to determine the second position information of the vehicle. Wherein, the second preset range is smaller than the first preset range, and the positioning accuracy of the second reference template is higher than that of the first reference template.

[0015] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the vehicle positioning methods described above.

[0016] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the vehicle positioning method as described above.

[0017] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements any of the vehicle positioning methods described above.

[0018] The vehicle positioning method and apparatus provided by this invention first collect and compare the first target features of the vehicle using a first sensing method within a relatively large first preset range to determine a first vehicle position information that can accommodate large stopping errors. Then, based on this first position information, the positioning range is narrowed to a second preset range smaller than the first preset range. Within the second preset range, a second sensing method and a second reference template with higher positioning accuracy are used to compare the second target features to determine the final second vehicle position information. By gradually narrowing the search range and improving the accuracy of the reference template, the positioning efficiency, reliability, and computational efficiency are significantly improved, effectively solving the problem of insufficient positioning accuracy caused by train stopping position deviations and dynamic expansion and contraction of the vehicle body connecting device in the prior art. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is one of the flowcharts illustrating the vehicle positioning method provided by the present invention.

[0021] Figure 2 This is the second flowchart illustrating the vehicle positioning method provided by the present invention.

[0022] Figure 3 This is a structural schematic diagram of the vehicle positioning device provided by the present invention.

[0023] Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0025] The positioning method based on odometers or encoders mentioned in the background technology calculates the distance the inspection robot travels within the trench track by measuring the number of revolutions of its own wheels, thus determining its position relative to the track's starting point. The fatal flaw of this method is the accumulation of errors. Every tiny wheel slippage, idle spin, or mechanical deviation of the measuring equipment accumulates, resulting in a larger error in the position calculation as the robot travels further. This necessitates frequent manual or other calibrations, failing to meet the requirements of long-term, stable automation.

[0026] Another approach is positioning based on fixed beacons, such as installing RFID tags or photoelectric switches along the track. This method essentially provides discrete "dot" signals. For example, it can only inform a robot that it has "reached the general area of ​​the 5th carriage," but this is far from sufficient for locating a specific component within that carriage. More importantly, a train is not a perfectly rigid body; the connecting structures between its carriages, such as couplers, buffers, and windshields, have compression and stretching spaces on the order of centimeters. Over the entire length of the train, this uncertainty accumulates to tens of centimeters. This means that even if the theoretical position of a carriage is determined by beacons, the actual position of its internal components, such as bogies and motors, still involves a significant random variable—an inherent flaw that the fixed beacon approach cannot address.

[0027] To address the aforementioned problems, this invention proposes an innovative positioning method. This method combines phased coarse and fine positioning, first rapidly determining the approximate position of the train over a large area, and then performing high-precision positioning over a smaller area. This overcomes the challenges posed by train stopping deviations and dynamic changes in the vehicle body, while avoiding the high computational complexity of global search due to large initial pose errors, thus achieving efficient and accurate train positioning.

[0028] Before introducing the technical solutions of the embodiments of the present invention, the terms and concepts involved in the embodiments of the present invention will be explained illustratively.

[0029] The first preset range refers to the theoretical search or acquisition space of the sensor when performing the coarse positioning step. This range is usually set relatively large in order to effectively accommodate the initial position uncertainty caused by factors such as vehicle parking and driver operating habits, and to ensure that the first target feature can be reliably captured no matter how much the vehicle's parking position deviates.

[0030] First data: refers to the dataset acquired through the first sensing method to characterize the features of the first target. For example, it can be point cloud data characterizing the three-dimensional structural contour of the target, or a low-resolution wide-angle image.

[0031] First reference template: refers to the data that is pre-collected and stored as the benchmark for the first coarse localization comparison. The localization accuracy of this template is relatively low, but the features it contains are highly robust and suitable for fast and stable pattern matching over a wide range.

[0032] First sensing mode: refers to a sensing mode or behavior used to collect first data, characterized by its suitability for rapid perception of macroscopic structural features over a large area.

[0033] The first target feature refers to a component under the vehicle that is easily and reliably identifiable over a large area. It typically has a macroscopically stable and prominent structural profile, such as the axle of a vehicle, whose shape and presence are highly consistent and predictable throughout the entire train.

[0034] The second preset range refers to the sensor's search or acquisition space during the fine positioning step after the first coarse positioning step. This range is much smaller than the first preset range, and its specific location is dynamically determined based on the positioning results of the first step. Its purpose is to focus the high-precision sensor into a very small area.

[0035] Secondary data refers to the dataset acquired through a secondary sensing method, used to characterize the features of the secondary target. This data typically has high resolution and rich detail; for example, it could be a high-resolution image containing information such as the target's surface texture, color, and edges.

[0036] Second reference template: This refers to the pre-collected and stored data used as the benchmark for the second step of fine-tuning and comparison. The positioning accuracy of this template contains fine feature information sufficient for accurate matching.

[0037] Second sensing mode: refers to a sensing mode or behavior used to collect second data, characterized by its suitability for high-resolution and high-precision perception of fine features within a small area.

[0038] The second target feature refers to a component located relatively fixed beneath the vehicle and possessing rich and detailed features. It is suitable as the final reference for high-precision positioning, such as a motor on a bogie, a brake disc caliper, or a specific nameplate or bolt group.

[0039] Laser sensor: A device that measures distance by emitting a laser beam and receiving the reflected signal. It can quickly acquire the three-dimensional spatial coordinates of an object's surface and generate point cloud data. It is particularly suitable for capturing the macroscopic three-dimensional contours of components such as axles.

[0040] Image sensor: refers to a device that can capture light signals and convert them into digital images, such as industrial CCD / CMOS cameras. It can acquire rich two-dimensional visual information such as the color, texture, and brightness of an object's surface, and is particularly suitable for recognizing fine features of targets such as motor housings and nameplates.

[0041] To implement the positioning method of this invention, the hardware environment of this invention mainly includes the following core modules: Inspection robot body: It has a flexible multi-wheeled chassis, equipped with anti-slip devices and a compact design to adapt to the narrowness and unevenness of ditches; it is powered by a high-capacity lithium battery to ensure continuous working capability.

[0042] Sensor system: Laser sensors are used for coarse positioning because they are insensitive to ambient light and can capture geometric features such as axles from a distance; image sensors are used for fine positioning, capturing details of components such as motors, brake discs, and calipers at high resolution. The combination of the two achieves a positioning process from coarse to fine.

[0043] Computing module: Equipped with an embedded GPU or DSP module, featuring industrial-grade protection design, adaptable to high-temperature, humid, and dusty environments in trenches, efficiently running deep learning models and processing sensor data.

[0044] Communication module: Adopts industrial-grade wireless Wi-Fi or 4G module to ensure stable data transmission between the robot and the back-end system, enabling remote monitoring and command reception.

[0045] Power supply module: It is compatible with the robot chassis, sensors, computing modules, etc., to ensure the stable operation of each module.

[0046] The entire hardware environment design fully considers the special needs of trench inspection, with close collaboration between modules to ensure that the robot can perform inspection tasks stably and efficiently.

[0047] The following is combined Figures 1-2 This invention describes a vehicle positioning method according to an embodiment of the present invention.

[0048] Figure 1 This is one of the flowcharts illustrating the vehicle positioning method provided by the present invention, which includes: Step 101: Within a first preset range, first data of the first target feature under the vehicle is collected using a first sensing method, and the first data is compared with a first reference template to determine the first position information of the vehicle.

[0049] Specifically, step 101 is the coarse positioning stage in the secondary positioning strategy of this invention, and its core objective is to eliminate the large and uncertain initial position error generated when the vehicle stops.

[0050] In a typical application scenario, the positioning method of this invention is triggered when the detection device carrying the sensor is instructed to move to the vicinity of the theoretical coordinate position of a component to be inspected. Due to various factors such as the driver's driving habits, the stop signs on the track, and the braking response of the train itself, the actual stopping position of the vehicle usually has a non-negligible, random offset compared to the theoretical coordinates, which may reach tens of centimeters.

[0051] To reliably handle this uncertainty, the first preset range is set to a sufficiently large spatial area, for example, ±50 cm in the vehicle's direction of travel. This range ensures that regardless of the actual deviation of the vehicle's parking position, the first target feature to be searched will inevitably fall within this range, thus laying the foundation for successful positioning.

[0052] Subsequently, the system scans the area using a first sensing method to collect initial data. This first sensing method refers to a sensing pattern suitable for large-scale, rapid structural perception. Accordingly, the selected first target feature is typically a simple, stable structural component under the vehicle that is repeated throughout the train, such as an axle. The geometric contour of this component is highly identifiable over a large area and is not easily confused with other complex pipelines or small components under the vehicle. The initial data collected using this sensing method focuses on the three-dimensional spatial contour or structural silhouette representing the target feature, rather than its fine surface texture.

[0053] After acquiring the initial data containing real-time environmental information, the system compares it with a pre-stored first reference template. This first reference template can be understood as a digital archive of the first target feature in its ideal position. The comparison process is essentially a pattern matching algorithm, which accurately searches for and identifies feature regions that match the first reference template in the real-time acquired initial data, which may contain a lot of background noise.

[0054] Through this comparison process, the system can calculate the deviation between the center position of the first target feature currently acquired in real time and the ideal center position represented by the first reference template, which is the final first position information to be determined. It quantifies the offset of the vehicle's actual parking position from its theoretical position with a specific numerical value (e.g., +15.3 cm). This first position information itself is not the final positioning result, but it plays a crucial role in connecting the preceding and following steps: it converges a large-scale, fuzzy uncertainty problem into a small-scale positioning correction problem with precise numerical values, providing accurate guidance and input for the high-precision positioning in the subsequent step 102.

[0055] Step 102: Determine a second preset range based on the first location information, collect second data of the second target features under the vehicle using a second sensing method within the second preset range, and compare the second data with a second reference template to determine the second location information of the vehicle.

[0056] The second preset range is smaller than the first preset range, and the positioning accuracy of the second reference template is higher than that of the first reference template.

[0057] Specifically, in step 102, the system uses the first location information to dynamically correct the sensor's target point for the next step. In other words, if the theoretical coordinates of the second target feature are X, the system will now automatically adjust the acquisition center point to "X + 15.3 cm". Around this corrected center point, the system will set a second preset range much smaller than the first preset range, for example, only ±5 cm, creating a prerequisite for subsequent high-precision recognition.

[0058] Within this second preset range, the system will switch to a second sensing mode to collect second data. The second sensing mode is a sensing mode focused on capturing fine details, requiring higher resolution and stronger detail resolution. Accordingly, the selected second target feature is a component beneath the vehicle with rich, stable, and easily matched local features, such as a nameplate on a component's housing, a specific bolt, or a unique casting mark. This feature is unique within a small area.

[0059] The system then compares the second data with a pre-stored second reference template. This second reference template is a digital file corresponding to the second target feature, and its positioning accuracy is higher than the first reference template. This comparison process typically involves a more complex and computationally intensive high-precision matching algorithm. It no longer seeks a rough outline, but rather aligns each feature in the real-time acquired data with its corresponding feature in the template.

[0060] Through this high-precision comparison, the system can calculate a more refined offset within a second preset range, such as -0.2 cm, as the vehicle's second position information. This position information is more accurate than the first position information, providing the vehicle's precise location in space, thus offering high-precision positioning support for subsequent detection or operational tasks.

[0061] Step 102 achieves the transition from coarse positioning to fine positioning by narrowing the search range and adopting a higher precision sensing method and reference template, ensuring the efficiency and accuracy of the positioning process.

[0062] Furthermore, in the vehicle positioning method of the present invention, the first sensing method and the second sensing method are implemented by specially designed first and second sensors, respectively, both of which are integrated and installed on the mobile device. This design enables the mobile device to play a key role in the positioning process, not only being able to move autonomously but also acquiring vehicle location information through its onboard sensors.

[0063] Once the mobile device receives the first data collected by the first sensor, namely the laser sensor, it determines the approximate location of the vehicle, i.e., the first location information, based on this data. Next, to further improve positioning accuracy, the system performs calculations and judgments based on this first location information, controlling the mobile device to adjust its own position. The purpose of this adjustment process is to ensure that a second target feature below the vehicle enters the field of view of the second sensor. This second target feature, distinct from the first target feature, can be the vehicle's motor or brake disc caliper.

[0064] It's important to note that the first and second target features need to be different to meet the different requirements of coarse and fine positioning. In the coarse positioning stage, the axle serves as the first target feature because its high saliency and ease of capture allow for rapid determination of the vehicle's approximate position. In the fine positioning stage, the motor or brake disc caliper serves as the second target feature because its fixed position, complex structure, and precision make it suitable for high-precision positioning. Using only a single feature may result in low search efficiency in coarse positioning due to its lack of saliency, while in fine positioning, its large feature size may lead to insufficient accuracy. The combination of both achieves a balance between efficiency and accuracy in the positioning process.

[0065] The second sensor is an image sensor that offers higher positioning accuracy compared to a laser sensor. After capturing image information of the second target's features, this information is converted into second data. The system compares this second data with a pre-stored second reference template, which has higher positioning accuracy than the first reference template. Through this high-precision comparison, the system can determine the vehicle's precise location information, i.e., the second position information.

[0066] This method of positioning in stages using two different types of sensors fully leverages the advantages of laser sensors for rapid positioning over a large area and the high-precision positioning capabilities of image sensors within a small area, thus achieving efficient and accurate vehicle positioning. Furthermore, since both sensors are mounted on the mobile device, the flexibility and adaptability of the entire positioning system are greatly enhanced.

[0067] In the vehicle positioning method of the present invention, the first sensing method and the second sensing method can also be implemented by the same multimodal sensor in different operating modes. This multimodal sensor has multiple operating modes and can switch its operating state according to actual needs to achieve different sensing functions.

[0068] Specifically, during coarse positioning, the multimodal sensor switches to its first operating mode, or first sensing method, which is typically similar to the operating mode of a laser sensor. In this mode, the sensor can quickly scan a large area beneath the vehicle, collecting initial data on the first target features. This data is primarily used to determine the approximate position of the vehicle, i.e., the initial position information.

[0069] When precise positioning is required, the multimodal sensor switches to a second operating mode, also known as a second sensing mode, which is typically similar to the operating mode of an image sensor. In this mode, the sensor can acquire second data on the features of a second target beneath the vehicle with higher resolution and accuracy. This data is then compared with a second reference template to determine the vehicle's precise position information, i.e., the second position information.

[0070] By using a single multimodal sensor to achieve two different sensing methods, this invention not only reduces the number and cost of hardware devices but also improves system integration and flexibility. This design allows the positioning system to seamlessly switch between different operating modes, thereby improving positioning efficiency and reliability while maintaining positioning accuracy.

[0071] Furthermore, in this embodiment, the process of comparing the first data with the first reference template in step 101 to determine the first location information of the vehicle is as follows: First, the first data acquired through a first sensing method is matched with a pre-stored first reference template. This matching process aims to find the part of the first data that is most similar to the first reference template, thereby determining the deviation between the vehicle's current position and the reference position, i.e., generating a first offset. The first offset visually reflects the difference between the vehicle's actual position and the ideal position in numerical form.

[0072] Next, the generated first offset is evaluated to see if it is less than a preset first threshold. This first threshold is a standard set according to the accuracy requirements of the positioning task, used to measure whether the current positioning result meets the accuracy requirements of the coarse positioning stage.

[0073] If the first offset is less than the first threshold, it means that the current positioning result has reached the expected accuracy of the coarse positioning stage. At this time, the first offset can be directly used as the vehicle's first position information for subsequent positioning operations or decision-making processes.

[0074] However, if the first offset is greater than or equal to the first threshold, it indicates that the accuracy of the current positioning result has not yet met the requirements. In this case, the system will control the mobile device to make corresponding position adjustments based on the direction and magnitude of the first offset. The purpose of the adjustment is to enable the sensors carried by the mobile device to better align with the vehicle's first target feature, so that the first data can be re-acquired at the new location.

[0075] After adjusting the position, the system will repeat the data acquisition steps of the first sensing method to obtain new first data again, and re-match and recalculate the offset with the first reference template. This process will continue to loop until the calculated first offset is less than a first threshold, thus obtaining the vehicle's first position information that meets the accuracy requirements. Through this iterative mechanism, the system can continuously optimize the positioning results, ensuring that the final vehicle position information has sufficient accuracy and reliability.

[0076] Furthermore, the process of comparing the second data with the second reference template in step 102 to determine the second location information of the vehicle is as follows: First, the second set of data is processed using a template matching algorithm based on deep learning. This algorithm, through a pre-trained deep learning model, is able to identify and locate the specific position of the second target feature beneath the vehicle. Deep learning models are typically trained on a large amount of sample data to ensure their accuracy and robustness in practical applications.

[0077] Next, the algorithm calculates the center position of the second target feature. This position is obtained through geometric analysis of the identified feature region and is usually represented in the acquired image as pixel coordinates.

[0078] Then, the center position is compared with the preset reference position in the second reference template. The second reference template contains the ideal position information of the second target feature, which is generated based on the vehicle's design drawings or standard model and is used as a reference standard for positioning.

[0079] By calculating the deviation between the actual center position and the reference position, the difference between the vehicle's current position and the ideal position can be accurately determined. This deviation value is the vehicle's second position information, which intuitively reflects the vehicle's position deviation during the precise positioning phase in numerical form.

[0080] In practice, this deviation information can be used to further adjust the position of the mobile device, enabling it to more accurately align with specific components of the vehicle, thereby achieving high-precision positioning and subsequent detection or operational tasks. This deep learning-based template matching method combines the powerful capabilities of machine learning with the accuracy of traditional geometric analysis, providing an efficient and reliable solution for vehicle positioning.

[0081] To further understand the technical solution of the embodiments of the present invention, a specific example is provided below to illustrate the vehicle positioning method of the present invention.

[0082] See Figure 2 The vehicle positioning method of this invention includes: 201. Reference template collection.

[0083] First reference template acquisition: Point cloud data of the axle was acquired using a 2D laser camera. The axle has a distinct geometric shape, making it easy for the laser sensor to identify and suitable as the first target feature.

[0084] Second reference template acquisition: Image data of the motor (powered carriage) and brake disc calipers (non-powered carriage) were acquired using a 2D image camera. Their complex and unique structures are suitable as secondary target features.

[0085] The collected template data of axles, motors, and brake disc calipers are stored in the system, and the template collection position is set as the target position.

[0086] 202. Robot Task Start-up and Initial Movement.

[0087] After receiving the vehicle positioning task instruction, the robot starts and prepares to begin the positioning process.

[0088] The robot determines its starting position and prepares to move towards the preset target location. The robot advances along the trench track towards the target location while continuously monitoring its distance from the target location.

[0089] 203. Coarse positioning threshold judgment.

[0090] The robot calculates the distance to the target location in real time and determines whether the distance reaches the coarse positioning threshold (e.g., 1 meter). If it does, it enters the coarse positioning process; otherwise, it continues to move towards the target location.

[0091] 204. Coarse positioning algorithm module flow.

[0092] The 2D laser camera is activated to acquire a point cloud image of the axle area beneath the vehicle. An algorithm is then used to match the acquired point cloud image with a pre-stored axle template, calculating the deviation between the current position and the template position.

[0093] Determine if the calculated deviation reaches a preset first threshold (e.g., ±30 cm). If it does, initiate the fine positioning process; otherwise, the robot continues to adjust its position and repeat the data acquisition and matching process.

[0094] 205. Fine positioning algorithm module process.

[0095] The 2D laser image acquisition device is activated to capture images of the motor or brake disc caliper beneath the vehicle. A deep learning-based algorithm is then used to match the acquired images with pre-stored templates of the motor or brake disc caliper. The deep learning model, trained on a large number of samples, can accurately identify target features and calculate the deviation between its center position and the template's reference position.

[0096] The calculated deviation is used as the precise displacement value of the vehicle to determine its final parking position in the pit.

[0097] 206. Task completion and result output.

[0098] After completing the precise positioning, the system terminates the vehicle positioning task. It outputs the vehicle's second location information to provide accurate location data for subsequent maintenance work.

[0099] Through the detailed steps 201-206 above, the vehicle positioning method of this embodiment of the invention can quickly and accurately determine the position of the vehicle in the maintenance trench, effectively improving the efficiency and safety of rail transit vehicle maintenance.

[0100] The vehicle positioning device provided in the embodiments of the present invention is described below. The vehicle positioning device described below and the vehicle positioning method described above can be referred to each other.

[0101] This invention provides a vehicle positioning device, such as... Figure 3 As shown, it includes: The first position determination module 310 is used to collect first data of the first target feature under the vehicle using a first sensing method within a first preset range, and compare the first data with a first reference template to determine the first position information of the vehicle. The second position determination module 320 is used to determine a second preset range based on the first position information, collect second data of the second target features under the vehicle using a second sensing method within the second preset range, and compare the second data with a second reference template to determine the second position information of the vehicle. Wherein, the second preset range is smaller than the first preset range, and the positioning accuracy of the second reference template is higher than that of the first reference template.

[0102] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4As shown, the electronic device may include a processor 410, a communications interface 420, a memory 430, and a communication bus 440, wherein the processor 410, communications interface 420, and memory 430 communicate with each other via the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a vehicle positioning method. This method includes: within a first preset range, acquiring first data of a first target feature below the vehicle using a first sensing method, and comparing the first data with a first reference template to determine the vehicle's first position information; determining a second preset range based on the first position information, acquiring second data of a second target feature below the vehicle within the second preset range using a second sensing method, and comparing the second data with a second reference template to determine the vehicle's second position information; wherein the second preset range is smaller than the first preset range, and the positioning accuracy of the second reference template is higher than that of the first reference template.

[0103] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0104] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the vehicle positioning method provided by the above methods. The method includes: within a first preset range, collecting first data of a first target feature under the vehicle using a first sensing method, and comparing the first data with a first reference template to determine the first position information of the vehicle; determining a second preset range based on the first position information, collecting second data of a second target feature under the vehicle using a second sensing method within the second preset range, and comparing the second data with a second reference template to determine the second position information of the vehicle; wherein the second preset range is smaller than the first preset range, and the positioning accuracy of the second reference template is higher than the positioning accuracy of the first reference template.

[0105] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the vehicle positioning method provided by the above methods. The method includes: within a first preset range, acquiring first data of a first target feature under the vehicle using a first sensing method, and comparing the first data with a first reference template to determine first position information of the vehicle; determining a second preset range based on the first position information, acquiring second data of a second target feature under the vehicle within the second preset range using a second sensing method, and comparing the second data with a second reference template to determine second position information of the vehicle; wherein the second preset range is smaller than the first preset range, and the positioning accuracy of the second reference template is higher than the positioning accuracy of the first reference template.

[0106] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0107] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A vehicle positioning method, characterized in that, include: Within a first preset range, first data of a first target feature under the vehicle is collected using a first sensing method, and the first data is compared with a first reference template to determine the first position information of the vehicle; Based on the first location information, a second preset range is determined. Within the second preset range, a second sensing method is used to collect second data of the second target features under the vehicle. The second data is then compared with a second reference template to determine the second location information of the vehicle. Wherein, the second preset range is smaller than the first preset range, and the positioning accuracy of the second reference template is higher than that of the first reference template.

2. The vehicle positioning method according to claim 1, characterized in that, The first sensing method is implemented by a first sensor, the second sensing method is implemented by a second sensor, and both the first sensor and the second sensor are mounted on a mobile device.

3. The vehicle positioning method according to claim 2, characterized in that, The step of determining the second preset range based on the first location information specifically includes: Based on the first location information, the mobile device is controlled to adjust its position so that the second target feature is within the field of view of the second sensor, thereby determining the second preset range.

4. The vehicle positioning method according to claim 2, characterized in that, The first sensor is a laser sensor, and the second sensor is an image sensor.

5. The vehicle positioning method according to claim 1, characterized in that, The first sensing method and the second sensing method are implemented by the same multimodal sensor in different operating modes.

6. The vehicle positioning method according to claim 1, characterized in that, The first data is compared with the first reference template to determine the first location information of the vehicle, specifically including: The first data is matched with the first reference template to generate a first offset representing the deviation between the vehicle's current position and the reference position; Determine whether the first offset is less than a preset first threshold. If so, use the first offset as the first position information. If not, control the mobile device to adjust its position according to the first offset, and repeatedly execute the step of collecting the first data of the first target feature under the vehicle using the first sensing method until the generated first offset is less than the first threshold.

7. The vehicle positioning method according to claim 1, characterized in that, The second data is compared with the second reference template to determine the second location information of the vehicle, specifically including: A template matching algorithm based on deep learning is used to calculate the deviation between the center position of the second target feature in the second data and the preset reference position in the second reference template, and the deviation is used as the second position information of the vehicle.

8. The vehicle positioning method according to claim 1, characterized in that, The first target feature is the vehicle axle, and the second target feature is the vehicle motor or brake disc caliper.

9. A vehicle positioning device, characterized in that, include: The first position determination module is used to collect first data of a first target feature under the vehicle using a first sensing method within a first preset range, and compare the first data with a first reference template to determine the first position information of the vehicle. The second position determination module is used to determine a second preset range based on the first position information, collect second data of the second target features under the vehicle using a second sensing method within the second preset range, and compare the second data with a second reference template to determine the second position information of the vehicle. Wherein, the second preset range is smaller than the first preset range, and the positioning accuracy of the second reference template is higher than that of the first reference template.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the vehicle positioning method as described in any one of claims 1 to 8.

11. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the vehicle positioning method as described in any one of claims 1 to 8.

12. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the vehicle positioning method as described in any one of claims 1 to 8.

Images