Road berth robot fault sensing system based on edge collaboration

Abstract

The invention discloses a road berth robot fault sensing system based on edge collaboration, and relates to the technical field of intelligent parking systems, and the system comprises an interaction module which displays the state of a parking space, and a user achieves the interaction with a berth robot through the interaction module; the cooperative fault sensing module is used for realizing close-range interaction between berth robot sensing layers based on the communication module, establishing a fault sensing model, comparing berth robots through the fault sensing model, calibrating fluctuation parameters and determining fault information of the berth robots; and when the user interacts with the faulty berth robot, the dynamic scheduling module controls the faulty berth robot by scheduling and responding to the control instruction of the berth robot control center. According to the invention, edge cooperation is realized through a plurality of adjacent berth robots, fault detection of the berth robots is completed, and the adjacent berth robots can be utilized to complete control of the faulty berth robot.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent parking systems, and in particular to a road parking robot fault perception system based on edge collaboration. Background Art

[0002] The parking robot is an automated device used to manage dedicated parking spaces. It uses intelligent sensing to determine when a space is occupied and released. During the 30-minute free parking period, the robot remains flush with the road surface. After the free period, the robot automatically rises, locks the vehicle, and begins charging. Before leaving, drivers simply scan a QR code next to the parking space, pay the parking fee, and the robot automatically descends.

[0003] Road parking robots may experience various types of failures:

[0004] Faulty blocking device: unable to raise / lower, the parking lock does not rise after the vehicle leaves to prevent others from parking; the parking lock does not fall after the vehicle enters to allow parking; the raising / lowering action is stuck, slow or incomplete;

[0005] System linkage failure: unable to communicate with the backend management platform or unstable communication; unable to receive instructions issued by the backend (such as rate update, remote reset); unable to report status and transaction data to the backend;

[0006] System freeze / stuttering: The touch screen is unresponsive or responds very slowly; the device remains in a certain state for a long time (such as displaying "Processing"); and requires frequent restarts to work.

[0007] Announcement No. CN118609411A discloses an intelligent charging and optimizing road parking robot system, which includes a holographic parking space mapping unit, a dynamic billing neural network, a contactless payment trigger, a virtual parking space reservation system, an interactive sound and light guidance array, an intelligent security patrol robot, a cloud data synchronization node, an intelligent parking space recommendation system and a management and processing center. By integrating the technologies of the holographic parking space mapping unit, the dynamic billing neural network, the contactless payment trigger, the intelligent security patrol robot and the cloud data synchronization node, comprehensive intelligent management of the parking process is achieved. By simulating manual charging, the system can automatically calculate the receivable fees and deduct them from the car owner's account before the car owner leaves the parking lot, thereby improving charging efficiency and accuracy. Through the real-time monitoring and three-dimensional modeling technology of the holographic parking space mapping unit, the system can dynamically update the parking space status and provide the car owner with accurate vacant parking space information.

[0008] At present, the fault detection of berth robots is carried out by remote communication once in a period of time. Due to the large number of berth robots, each berth robot conducts remote communication at intervals to realize fault detection. First, the interval time is long. When a fault occurs, it is necessary to wait for the predetermined fault detection time point to complete the fault detection of the berth robot. Second, the berth robots all conduct remote communication, which increases the system's processing capacity, and most of the berth robots are in a non-fault state, resulting in low efficiency in handling faulty berth robots. Summary of the Invention

[0009] One of the purposes of the present invention is to provide a road berth robot fault perception system based on edge collaboration, which enables close-range communication of berth robots through edge collaboration, shares berth information, determines fault information through self-inspection and close-range interconnection, and realizes close-range fault information sharing, and collaborative processing to achieve automatic unlocking.

[0010] To achieve the above objectives, the present invention is implemented through the following technical solutions: A road berth robot fault perception system based on edge collaboration, comprising:

[0011] Parking robot, set in the corresponding parking space;

[0012] The perception layer identifies whether there is a vehicle in the parking space and matches the vehicle type. The perception layer obtains the vehicle parking information;

[0013] The control center controls the parking robot based on the vehicle parking information obtained by the perception layer and executes the lifting program;

[0014] The communication module enables remote communication and close interaction between berth robots, collaborative subscription between adjacent berth robots, and real-time receipt of fault notifications;

[0015] Interaction module, which displays the parking space status and allows users to interact with the parking robot through the interaction module;

[0016] The collaborative fault perception module uses short-range communication to retrieve the self-test results of robots in adjacent berths, and combines data from multiple machines to dynamically update fault model parameters, enabling cross-checking of robots in adjacent berths and locating the fault point.

[0017] The self-test module integrates software and hardware self-test submodules. After receiving the fault broadcast, it starts the berth robot self-test. If the fault is stuck or communication is interrupted, the local unlocking procedure is executed;

[0018] Dynamic scheduling module, when the user interacts with the faulty berth robot, the dynamic scheduling module controls the faulty berth robot by scheduling the response to the control instructions of the berth robot control center, and feeds back the collaborative progress to the maintenance platform.

[0019] In one or more embodiments of the present invention, the fault sensing system further includes:

[0020] A data management platform for storing vehicle information and billing data. The data management platform stores historical fault data, and the maintenance platform determines the repair guide based on the fault information of the faulty berth robot;

[0021] The control terminal realizes remote control of the parking robot. The communication module communicates remotely with the control terminal, and the control terminal visualizes the position, status and occupancy rate of all parking robots in real time.

[0022] In one or more embodiments of the present invention, the steps for the perception layer to identify a parking space vehicle are as follows:

[0023] Multi-sensor fusion for vehicle presence detection;

[0024] Obtain vehicle chassis, model, and metal distribution characteristics to output the model ID;

[0025] Establish a model to be labeled, and input the vehicle chassis, model, and metal distribution characteristics into the model to be labeled for cloud matching;

[0026] Generate berth information based on vehicle model ID or model to be annotated.

[0027] In one or more embodiments of the present invention, the perception layer uploads the model to be annotated to the control end through the interaction module for cloud matching. The control end inputs the model to be annotated into the vehicle model library for similarity search and returns the matching results to the perception layer:

[0028] If there is a vehicle model ID, the control end sends the vehicle model ID to the perception layer;

[0029] If there is no matching result, the control end generates a labeling task and pushes the labeling task. After the vehicle ID of the unlabeled model is manually labeled, the control end sends the vehicle ID to the perception layer.

[0030] In one or more embodiments of the present invention, the control center determines the vehicle interference range:

[0031] A coordinate system is established based on the parking robot and the size of the parking space, in which the origin is established based on the position of the parking robot;

[0032] Obtain vehicle wheel position data, determine the wheelbase based on the chassis features corresponding to the vehicle model ID, and generate the coordinate points of the corresponding wheels;

[0033] Calculate the vehicle tilt angle based on the wheel coordinate points;

[0034] Generate the vehicle projection range in the coordinate system and determine the vehicle interference range;

[0035] When the vehicle interference range exceeds the parking space, the user will be reminded through the interactive module, and the parking robot will not be able to rise;

[0036] When the vehicle interference range does not exceed the parking space, the control center can control the parking robot to rise.

[0037] In one or more embodiments of the present invention, the collaborative fault perception module uses edge short-range communication to locate the fault point in the following steps:

[0038] The berth robot continuously collects key sensor data of the local robot. When any key sensor data exceeds the local threshold or the residual suddenly increases, the fault perception submodule determines the local warning and generates a warning event, recording the local warning timestamp, sensor type and current data;

[0039] The fault time is broadcast to robots in the same group or adjacent berths in a multicast or subscription manner through short-range communication. The adjacent berth robots match the fault time in the broadcast list, enter the collaborative sensing mode, and perform the self-test procedure;

[0040] The corresponding group or subscribed robot establishes a self-test sub-process to detect the components corresponding to the triggering warning parameters and generate a self-test report. The adjacent berth robot sends the self-test report back to the faulty berth robot that triggered the warning through the edge link;

[0041] The robot at the faulty berth aggregates its own historical data with the self-test reports submitted by all adjacent robots, aligns the timestamps of each node, and removes redundant data caused by communication packet loss or time drift. It then performs denoising, outlier removal, and interpolation to form a parallel comparison dataset.

[0042] By performing parameter correction on the parallel comparison data set, dynamically updating the parameter warning threshold, and calculating the residual error of the faulty berth robot and the adjacent berth robots on the same physical quantity, the fault area is preliminarily located.

[0043] The initial fault area is subdivided into subcomponent data, and the dynamic trend of the deviation between the actual and expected signals is determined in combination with the state of the Kalman filter. The sensor or submodule with the most significant deviation is identified and declared as the fault point.

[0044] In one or more embodiments of the present invention, a recovery strategy is determined after multi-machine verification of the fault point:

[0045] Verify the fault point results with the secondary self-test report of the neighboring machine to determine the fault type and severity level. If a verification conflict occurs, enter the review process, broadcast a confirmation request, and invite a third-party robot to participate in the verification.

[0046] The final confirmed fault point, fault type, and severity level are reported to the data management platform. The data management platform sends the fault data to the maintenance platform, which triggers the corresponding recovery strategy based on the fault data.

[0047] The key data, algorithm parameters, and calibration factors in the entire process of collaborative fault location are integrated into training samples and regularly uploaded in batches to the data management platform to participate in offline training and online updates of the fault model.

[0048] In one or more embodiments of the present invention, the parallel comparison of data sets is performed by calculating the calibration factor through least squares fitting, modifying the sensor bias of each collection node, and dynamically updating the warning threshold and trust weight based on the ambient temperature, load level and historical fluctuations. After each iteration, the latest threshold configuration is sent to all participating edge collection nodes;

[0049] After benchmarking the calibrated data, the residuals of the faulty berth robot and the adjacent berth robots on the same physical quantity are calculated, the residual matrix is ​​constructed, and the nodes or sensor channels with prominent abnormalities are identified. If more than one adjacent berth robot gives consistent abnormalities, the fault area is preliminarily located.

[0050] In one or more embodiments of the present invention, the dynamic scheduling module controls the faulty berth robot by scheduling the response to the control command of the berth robot control center:

[0051] The user makes an interaction request to the faulty berth robot through the interaction module;

[0052] The faulty berth robot determines whether the interaction request interferes with the fault information. If no interference occurs, the faulty berth robot completes the corresponding interaction action according to the interaction request. If interference occurs, the dynamic scheduling model obtains the interaction request and determines the responding berth robot based on the fault perception model.

[0053] The response berth robot receives the interaction request and uploads the interaction request to the control end for response verification. After completing the response verification, the response berth robot generates a control instruction based on the interaction request and returns it to the dynamic scheduling model;

[0054] The dynamic scheduling model converts the control instructions into response instructions corresponding to the faulty berth robot and controls the actions of the faulty berth robot.

[0055] In one or more embodiments of the present invention, based on the vehicle chassis height corresponding to the faulty berthing robot, the control instruction of the response berthing robot control center is converted into a response instruction of the faulty berthing robot, and the faulty berthing robot acts based on the response instruction.

[0056] Through the above technical solution, the present invention has the following beneficial effects:

[0057] 1. This application realizes edge collaboration through multiple adjacent berth robots to complete fault detection of the berth robot, and can use adjacent berth robots to complete the control of the faulty berth robot to ensure that the faulty berth robot can complete the corresponding interactive actions when the user interacts, further avoiding the impact of the fault on the user's use of the berth robot.

[0058] 2. By analyzing the position of the vehicle, it can be determined whether the vehicle corresponding to the parking space will cause interference outside the parking space. When interference occurs, the interactive module will warn the user to avoid scratches caused by the tilt of the vehicle position.

[0059] 3. The positioning results are verified through a secondary self-test report to ensure that the located fault is real, accurate, and consistent before executing recovery or alarm. Through independent self-tests of multiple machines, the sensor readings and fault characteristics are mutually verified to eliminate deviations caused by single-point anomalies or sensor false triggering. When a robot reports an anomaly that most neighboring machines do not detect, "false positives" can be identified in a timely manner; conversely, "false negatives" can be prevented to ensure system reliability.

[0060] Other features and advantages of the present invention will be described in the following description, and part of them will become obvious from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 Schematic diagram of the fault sensing system of the present invention. DETAILED DESCRIPTION

[0062] The following drawings illustrate various embodiments of the present invention. For clarity, many practical details are included in the following description. However, it should be understood that these practical details are not intended to limit the present invention. In other words, in some embodiments of the present invention, these practical details are optional. Furthermore, features from different embodiments may be interchangeably applicable, where practically possible.

[0063] Unless otherwise defined, all words used herein (including technical and scientific terms) have their ordinary meanings as understood by those skilled in the art. Furthermore, the definitions of the above-mentioned words in commonly used dictionaries should be interpreted in the context of this specification as having the same meanings as those in the relevant field of the present invention. Unless otherwise explicitly defined, these words should not be interpreted as having idealized or overly formal meanings.

[0064] See Figure 1As shown, the present invention provides a road berth robot fault perception system based on edge collaboration, which realizes the fault perception of the berth robot, converts long-interval remote communication into short-range edge collaboration, and completes the fault perception of the berth robot based on the communication between the berth robots.

[0065] The fault sensing system includes:

[0066] Parking robot, set in the corresponding parking space;

[0067] The perception layer identifies whether there is a vehicle in the parking space and matches the vehicle type. The perception layer obtains the vehicle parking information;

[0068] The control center controls the parking robot based on the vehicle parking information obtained by the perception layer and executes the lifting program;

[0069] The communication module enables remote communication and close interaction between berth robots, collaborative subscription between adjacent berth robots, and real-time receipt of fault notifications;

[0070] Interaction module, which displays the parking space status and allows users to interact with the parking robot through the interaction module;

[0071] The collaborative fault perception module uses short-range communication to retrieve the self-test results of robots in adjacent berths, and combines data from multiple machines to dynamically update fault model parameters, enabling cross-checking of robots in adjacent berths and locating the fault point.

[0072] The self-test module integrates software and hardware self-test submodules. After receiving the fault broadcast, it starts the berth robot self-test. If the fault is stuck or communication is interrupted, the local unlocking procedure is executed;

[0073] Dynamic scheduling module, when the user interacts with the faulty berth robot, the dynamic scheduling module controls the faulty berth robot by scheduling the response to the control instructions of the berth robot control center, and feeds back the collaborative progress to the maintenance platform.

[0074] In one feasible method, the remote communication fault detection of the berth robot at fixed intervals is adjusted to communication between short-range berth robots to realize fault perception of the berth robot. Compared with the remote communication of the berth robot, the fault perception processing speed between short-range berth robots is faster, and because the edge collaboration of the berth robot processes less data, frequent fault perception can be achieved to ensure that the fault of the berth robot can be determined in time.

[0075] Among them, when the berth robot determines the fault state, it interacts with the faulty berth robot interaction module by responding to the berth robot control center, and replicates the control instructions using the collaborative fault perception module. The berth robot is controlled without calling the faulty berth robot control center, avoiding the problem that the user cannot control the berth robot blocking device through the interactive module when the berth robot fails.

[0076] In one embodiment, the fault sensing system further includes:

[0077] A data management platform for storing vehicle information and billing data. The data management platform stores historical fault data, and the maintenance platform determines the repair guide based on the fault information of the faulty berth robot;

[0078] The control terminal realizes remote control of the parking robot. The communication module communicates remotely with the control terminal, and the control terminal visualizes the position, status and occupancy rate of all parking robots in real time.

[0079] In one feasible approach, a complete fault perception system needs to build a three-level "end-edge-cloud" architecture, adjusting remote communication fault detection at long time intervals to close-range mutual inspection of berth robots, reducing the burden on the system and enabling faster detection of faulty berth robots. The normal operation of the faulty berth robots is ensured by using edge collaboration.

[0080] By storing historical fault data, a maintenance guide can be determined more quickly based on the fault information, so that the operator can repair the berth robot according to the maintenance method in the maintenance guide.

[0081] In one embodiment, the perception layer identifies parking spaces and vehicles in the following steps:

[0082] Multi-sensor fusion for vehicle presence detection;

[0083] Obtain vehicle chassis, model, and metal distribution characteristics to output the model ID;

[0084] Establish a model to be labeled, and input the vehicle chassis, model, and metal distribution characteristics into the model to be labeled for cloud matching;

[0085] Generate berth information based on vehicle model ID or model to be annotated.

[0086] In one feasible method, the perception layer is used to identify parking space vehicle information, determine whether the vehicle is parked, and the status of the vehicle, so as to avoid vehicles frequently leaving and parking during free parking time. For vehicles with unrecognized vehicle model IDs, the parking space information of the vehicle is determined by establishing a model to be labeled.

[0087] Among them, multi-sensor fusion includes initial screening by geomagnetic sensors, confirmation of object attributes by millimeter-wave radar, and verification by the visual system to determine the detailed data of the vehicle.

[0088] Vehicle chassis characteristics include wheelbase and chassis height;

[0089] Model characteristics include headlight spacing, grille shape;

[0090] The metal distribution feature determines the engine location through the geomagnetic signal waveform and is used to distinguish vehicles.

[0091] In one embodiment, the perception layer uploads the model to be annotated to the control end through the interaction module for cloud matching. The control end inputs the model to be annotated into the vehicle model library for similarity search and returns the matching results to the perception layer:

[0092] If there is a vehicle model ID, the control end sends the vehicle model ID to the perception layer;

[0093] If there is no matching result, the control end generates a labeling task and pushes the labeling task. After the vehicle ID of the unlabeled model is manually labeled, the control end sends the vehicle ID to the perception layer.

[0094] In one feasible method, the unlabeled model is uploaded to the control end for cloud matching, and the unlabeled model is further searched based on the vehicle chassis, model and metal distribution characteristics to determine the vehicle ID of the parking space.

[0095] Among them, in order to facilitate the corresponding perception layer of the berth robot to identify the vehicle, the berth robot is divided into different areas. When an unlabeled model appears in the area and the vehicle model ID is determined, the control end sends the vehicle model ID to the corresponding perception layer of all berth robots in the area to facilitate the identification of the vehicle next time.

[0096] In one embodiment, the control center determines the vehicle interference range:

[0097] A coordinate system is established based on the parking robot and the size of the parking space, in which the origin is established based on the position of the parking robot;

[0098] Obtain vehicle wheel position data, determine the wheelbase based on the chassis features corresponding to the vehicle model ID, and generate the coordinate points of the corresponding wheels;

[0099] Calculate the vehicle tilt angle based on the wheel coordinate points;

[0100] Generate the vehicle projection range in the coordinate system and determine the vehicle interference range;

[0101] When the vehicle interference range exceeds the parking space, the user will be reminded through the interactive module, and the parking robot will not be able to rise;

[0102] When the vehicle interference range does not exceed the parking space, the control center can control the parking robot to rise.

[0103] In one feasible method, the vehicle interference range is determined by the control center, and the vehicle position can be analyzed to determine whether the vehicle has abnormal parking conditions such as slightly crossing the line, encroaching on the lane, or parking diagonally. The user is prompted through the interactive module to improve the effective utilization rate of parking spaces and reduce scratches caused by parking beyond the boundary.

[0104] Since the front wheels of the vehicle are steering wheels, the vehicle tilt angle is determined based on the rear wheels of the vehicle, and the front wheels of the vehicle are corrected to determine the tilt angle of the vehicle body.

[0105] In one embodiment, the collaborative fault perception module locates the fault point using edge short-range communication in the following steps:

[0106] The berth robot continuously collects key sensor data of the local robot. When any key sensor data exceeds the local threshold or the residual suddenly increases, the fault perception submodule determines the local warning and generates a warning event, recording the local warning timestamp, sensor type and current data;

[0107] The fault time is broadcast to robots in the same group or adjacent berths in a multicast or subscription manner through short-range communication. The adjacent berth robots match the fault time in the broadcast list, enter the collaborative sensing mode, and perform the self-test procedure;

[0108] The corresponding group or subscribed robot establishes a self-test sub-process to detect the components corresponding to the triggering warning parameters and generate a self-test report. The adjacent berth robot sends the self-test report back to the faulty berth robot that triggered the warning through the edge link;

[0109] The robot at the faulty berth aggregates its own historical data with the self-test reports submitted by all adjacent robots, aligns the timestamps of each node, and removes redundant data caused by communication packet loss or time drift. It then performs denoising, outlier removal, and interpolation to form a parallel comparison dataset.

[0110] By performing parameter correction on the parallel comparison data set, dynamically updating the parameter warning threshold, and calculating the residual error of the faulty berth robot and the adjacent berth robots on the same physical quantity, the fault area is preliminarily located.

[0111] The initial fault area is subdivided into subcomponent data, and the dynamic trend of the deviation between the actual and expected signals is determined in combination with the state of the Kalman filter. The sensor or submodule with the most significant deviation is identified and declared as the fault point.

[0112] In one feasible approach, the collaborative fault perception module utilizes edge short-range communication to close the entire process from local early warning, neighboring machine self-checking to data fusion, residual analysis, and precise fault point location. It utilizes edge short-range communication and efficient data fusion to quickly and accurately locate the fault point.

[0113] Relying on edge collaboration in the event of a single point of failure, fault diagnosis, sharing, and automatic response can be quickly completed, greatly improving reliability and user experience.

[0114] In one embodiment, a recovery strategy is determined after multi-machine verification of the failure point:

[0115] Verify the fault point results with the secondary self-test report of the neighboring machine to determine the fault type and severity level. If a verification conflict occurs, enter the review process, broadcast a confirmation request, and invite a third-party robot to participate in the verification.

[0116] The final confirmed fault point, fault type, and severity level are reported to the data management platform. The data management platform sends the fault data to the maintenance platform, which triggers the corresponding recovery strategy based on the fault data.

[0117] The key data, algorithm parameters, and calibration factors in the entire process of collaborative fault location are integrated into training samples and regularly uploaded in batches to the data management platform to participate in offline training and online updates of the fault model.

[0118] In one feasible approach, secondary confirmation of the fault point results can further improve the accuracy of fault point location. Multiple machines can conduct independent self-inspections and mutually verify sensor data and fault characteristics, eliminating deviations caused by single-point anomalies or sensor mis-triggering.

[0119] The multiple verification mechanism is more tolerant to single-point failures or network jitter, avoiding erroneous operations in global scheduling or automatic recovery caused by individual anomalies.

[0120] In one embodiment, the parallel comparison data sets are used to calculate the calibration factor through least squares fitting, modify the sensor bias of each collection node, and dynamically update the warning threshold and trust weight based on the ambient temperature, load level and historical fluctuations. After each iteration, the latest threshold configuration is sent to all participating edge collection nodes;

[0121] After benchmarking the calibrated data, the residuals of the faulty berth robot and the adjacent berth robots on the same physical quantity are calculated, the residual matrix is ​​constructed, and the nodes or sensor channels with prominent abnormalities are identified. If more than one adjacent berth robot gives consistent abnormalities, the fault area is preliminarily located.

[0122] In one feasible approach, sensor deviations are corrected to eliminate range drift and environmental influences, and to eliminate measurement differences between different nodes. Dynamic thresholds are automatically adjusted based on the real-time environment and load level. Fixed threshold settings can easily lead to false alarms. Dynamic thresholds are updated in real time with historical data and multi-machine calibration results, improving fault detection sensitivity while suppressing false alarms caused by harmless jitter.

[0123] By comparing the residuals of the same type of parameters of the faulty berth robot with those of the adjacent berth robots after calibration, we can quickly identify which link or component has the most abnormal reading. Cross-verification allows the independent measurements of multiple machines to verify each other. Once multiple nodes show consistent deviations from the same physical quantity, false signals caused by individual sensor failures or communication jitter can be eliminated.

[0124] In one embodiment, the dynamic scheduling module controls the faulty berth robot by scheduling the response to the berth robot control center control command:

[0125] The user makes an interaction request to the faulty berth robot through the interaction module;

[0126] The faulty berth robot determines whether the interaction request interferes with the fault information. If no interference occurs, the faulty berth robot completes the corresponding interaction action according to the interaction request. If interference occurs, the dynamic scheduling model obtains the interaction request and determines the responding berth robot based on the fault perception model.

[0127] The response berth robot receives the interaction request and uploads the interaction request to the control end for response verification. After completing the response verification, the response berth robot generates a control instruction based on the interaction request and returns it to the dynamic scheduling model;

[0128] The dynamic scheduling model converts the control instructions into response instructions corresponding to the faulty berth robot and controls the actions of the faulty berth robot.

[0129] In one feasible manner, during the malfunction of the faulty berthing robot, the user interacts with the faulty berthing robot and can control the faulty berthing robot by responding to the robot, thereby preventing the faulty berthing robot from affecting the user's normal use during the malfunction.

[0130] In one embodiment, based on the vehicle chassis height corresponding to the faulty berthing robot, the control instruction of the response berthing robot control center is converted into a response instruction of the faulty berthing robot, and the faulty berthing robot acts based on the response instruction.

[0131] In one feasible method, the control of the faulty berthing robot can be adjusted to suit the conversion of response instructions. Due to the different positions of the vehicle chassis, the berthing robot needs to be lifted to different heights, and the responding berthing robot only obtains the user's interactive instructions, not all the data of the faulty berthing robot. After the control instructions are converted into response instructions, it can adapt to the vehicle corresponding to the faulty berthing robot.

[0132] This application realizes edge collaboration through multiple adjacent berth robots to complete fault detection of the berth robot, and can use adjacent berth robots to complete the control of the faulty berth robot to ensure that the faulty berth robot can complete the corresponding interactive actions when the user interacts, further avoiding the impact of the fault on the user's use of the berth robot.

[0133] Although the present invention is disclosed in conjunction with the above embodiments, it is not intended to limit the present invention. Anyone familiar with this technology can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention should be based on the definition of the attached claims.

Claims

1. A road berth robot fault perception system based on edge collaboration, characterized in that: include: Parking robot, set in the corresponding parking space; The perception layer identifies whether there is a vehicle in the parking space and matches the vehicle type. The perception layer obtains the vehicle parking information; The control center controls the parking robot based on the vehicle parking information obtained by the perception layer and executes the lifting program; Communication module, which enables long-distance communication and close-range interaction of berth robots; Interaction module, which displays the parking space status and allows users to interact with the parking robot through the interaction module; Collaborative fault perception module: Based on the communication module, it realizes close interaction between the perception layers of the berth robots, establishes a fault perception model, compares the berth robots through the fault perception model, calibrates the fluctuation parameters, and determines the fault information of the berth robots; Dynamic scheduling module, when the user interacts with the faulty berth robot, the dynamic scheduling module controls the faulty berth robot by scheduling the control instructions of the berth robot control center.

2. The road berth robot fault perception system based on edge collaboration according to claim 1 is characterized in that: The fault sensing system also includes: A data management platform for storing vehicle information and billing data, which stores historical fault data and determines repair instructions based on the fault information of the faulty berth robot; The control terminal realizes remote control of the parking robot. The communication module communicates remotely with the control terminal, and the control terminal visualizes the position, status and occupancy rate of all parking robots in real time.

3. The road berth robot fault perception system based on edge collaboration according to claim 2 is characterized in that: The steps for the perception layer to identify parking spaces and vehicles are as follows: Multi-sensor fusion for vehicle presence detection; Obtain vehicle chassis, model, and metal distribution characteristics to output the model ID; Establish a model to be labeled, and input the vehicle chassis, model, and metal distribution characteristics into the model to be labeled for cloud matching; Generate berth information based on vehicle model ID or model to be annotated.

4. The road berth robot fault perception system based on edge collaboration according to claim 3 is characterized in that: The perception layer uploads the model to be annotated to the control end through the interaction module for cloud matching. The control end inputs the model to be annotated into the vehicle model library for similarity search and returns the matching results to the perception layer: If there is a vehicle model ID, the control end sends the vehicle model ID to the perception layer; If there is no matching result, the control end generates a labeling task and pushes the labeling task. After the vehicle ID of the unlabeled model is manually labeled, the control end sends the vehicle ID to the perception layer.

5. The road berth robot fault perception system based on edge collaboration according to claim 4 is characterized in that: The control center determines the vehicle interference range: A coordinate system is established based on the parking robot and the size of the parking space, in which the origin is established based on the position of the parking robot; Obtain vehicle wheel position data, determine the wheelbase based on the chassis features corresponding to the vehicle model ID, and generate the coordinate points of the corresponding wheels; Calculate the vehicle tilt angle based on the wheel coordinate points; Generate the vehicle projection range in the coordinate system and determine the vehicle interference range; When the vehicle interference range exceeds the parking space, the user will be reminded through the interactive module, and the parking robot will not be able to rise; When the vehicle interference range does not exceed the parking space, the control center can control the parking robot to rise.

6. The road berth robot fault perception system based on edge collaboration according to claim 5 is characterized in that: The collaborative fault perception module uses edge short-range communication to locate the fault point in the following steps: The berth robot continuously collects key sensor data of the local robot. When any key sensor data exceeds the local threshold or the residual suddenly increases, the fault perception submodule determines the local warning and generates a warning event, recording the local warning timestamp, sensor type and current data; The fault time is broadcast to robots in the same group or adjacent berths in a multicast or subscription manner through short-range communication. The adjacent berth robots match the fault time in the broadcast list, enter the collaborative sensing mode, and perform the self-test procedure; The corresponding group or subscribed robot establishes a self-test sub-process to detect the components corresponding to the triggering warning parameters and generate a self-test report. The adjacent berth robot sends the self-test report back to the faulty berth robot that triggered the warning through the edge link; The robot at the faulty berth aggregates its own historical data with the self-test reports submitted by all adjacent robots, aligns the timestamps of each node, and removes redundant data caused by communication packet loss or time drift. It then performs denoising, outlier removal, and interpolation to form a parallel comparison dataset. By performing parameter correction on the parallel comparison data set, dynamically updating the parameter warning threshold, and calculating the residual error of the faulty berth robot and the adjacent berth robots on the same physical quantity, the fault area is preliminarily located. The initial fault area is subdivided into subcomponent data, and the dynamic trend of the deviation between the actual and expected signals is determined in combination with the state of the Kalman filter. The sensor or submodule with the most significant deviation is identified and declared as the fault point.

7. The road berth robot fault perception system based on edge collaboration according to claim 6 is characterized in that: After verifying the fault point on multiple machines, determine the recovery strategy: Verify the fault point results with the secondary self-test report of the neighboring machine to determine the fault type and severity level. If a verification conflict occurs, enter the review process, broadcast a confirmation request, and invite a third-party robot to participate in the verification. The final confirmed fault point, fault type, and severity level are reported to the data management platform. The data management platform sends the fault data to the maintenance platform, which triggers the corresponding recovery strategy based on the fault data. The key data, algorithm parameters, and calibration factors in the entire process of collaborative fault location are integrated into training samples and regularly uploaded in batches to the data management platform to participate in offline training and online updates of the fault model.

8. The road berth robot fault perception system based on edge collaboration according to claim 7 is characterized in that: The parallel comparison datasets are used to calculate calibration factors through least squares fitting, modify the sensor deviations of each collection node, and dynamically update warning thresholds and trust weights based on ambient temperature, load level, and historical fluctuations. After each iteration, the latest threshold configuration is distributed to all participating edge collection nodes. After benchmarking the calibrated data, the residuals of the faulty berth robot and the adjacent berth robots on the same physical quantity are calculated, the residual matrix is ​​constructed, and the nodes or sensor channels with prominent abnormalities are identified. If more than one adjacent berth robot gives consistent abnormalities, the fault area is preliminarily located.

9. The road berth robot fault perception system based on edge collaboration according to claim 8 is characterized in that: The dynamic scheduling module controls the faulty berth robot by scheduling the control instructions of the response berth robot control center: The user makes an interaction request to the faulty berth robot through the interaction module; The faulty berth robot determines whether the interaction request interferes with the fault information. If no interference occurs, the faulty berth robot completes the corresponding interaction action according to the interaction request. If interference occurs, the dynamic scheduling model obtains the interaction request and determines the responding berth robot based on the fault perception model. The response berth robot receives the interaction request and uploads the interaction request to the control end for response verification. After completing the response verification, the response berth robot generates a control instruction based on the interaction request and returns it to the dynamic scheduling model; The dynamic scheduling model converts the control instructions into response instructions corresponding to the faulty berth robot and controls the actions of the faulty berth robot.

10. The road berth robot fault perception system based on edge collaboration according to claim 9 is characterized in that: Based on the vehicle chassis height corresponding to the faulty berthing robot, the control instruction of the response berthing robot control center is converted into the response instruction of the faulty berthing robot, and the faulty berthing robot acts based on the response instruction.

Images