Parking risk data acquisition method, device and terminal

Abstract

The invention provides a parking risk data collection method and device and a terminal, and belongs to the technical field of data collection, the method comprises the steps that sensor data collected by a preset vehicle in the parking process is acquired, and the sensor data comprises visual data and ultrasonic data; judging whether the ultrasonic data has a false detection risk or not based on the type of the visual target; and if the ultrasonic data has the error detection risk and the driving parameters of the preset vehicle meet the preset risk condition, recording sensor data with the error detection risk. According to the method, the hysteresis caused by obtaining problem data after sale after collision in a traditional scheme can be avoided, and an accurate data source is provided for subsequent risk analysis and algorithm optimization.

Description

Technical Field

[0001] This application relates to the field of data acquisition technology, and in particular to a parking risk data acquisition method, device and terminal. Background Technology

[0002] Automated parking systems, as a key application in the field of intelligent driving, greatly improve the convenience and safety of vehicle parking, especially for users with limited driving experience who face the challenges of parking in tight spaces and the risk of operational errors. The reliable operation of automated parking systems heavily relies on the accurate detection of the surrounding environment by the perception module. Current mainstream perception solutions generally employ visual sensors, such as monocular fisheye cameras, combined with ultrasonic sensors. The quality of the collaborative work between these two types of sensors and the corresponding arbitration algorithm directly determine the system's accuracy in obstacle recognition, distance measurement, and adaptability to complex scenarios. These are the core technological supports for ensuring the safety and stability of the automated parking process.

[0003] However, the existing multi-sensor fusion detection mechanism of automatic parking systems still has some hardware and software defects. For example, ultrasonic sensors are prone to failure due to physical limitations, and visual sensors have low accuracy in distance detection, which leads to false detections and missed detections in some risky scenarios. This results in poor detection accuracy of potential collision risks in parking scenarios, making it difficult to identify and avoid scrapes and collisions when parking at close range in advance. Therefore, how to collect effective sensor data is particularly important for the iterative optimization of subsequent perception algorithms. Summary of the Invention

[0004] This application provides a parking risk data acquisition method, device, and terminal to solve the problem of low detection accuracy of risk scenarios caused by the defects of multi-sensor fusion detection in the prior art.

[0005] Firstly, this application provides a method for collecting parking risk data, including: Acquire sensor data collected during the parking process of a preset vehicle, the sensor data including visual data and ultrasonic data; the visual data includes visual targets and the types of the visual targets; Based on the type of the visual target, determine whether the ultrasonic data has a risk of false detection; If the ultrasonic data has a risk of false detection and the driving parameters of the preset vehicle meet the preset risk conditions, then the sensor data with the risk of false detection is recorded.

[0006] As can be seen from the above embodiments, this embodiment, by acquiring visual and ultrasonic data, can determine whether there is a risk of false detection in ultrasonic data based on the type of visual target. When there is a risk of false detection in ultrasonic data, non-parking risk scenarios are excluded using vehicle driving parameters, ensuring that only sensor data with a genuine parking collision hazard is recorded. Furthermore, this method can immediately acquire the corresponding sensor data at the moment a parking collision risk occurs. This not only avoids the lag caused by obtaining problematic data after a collision in traditional solutions, but also provides an accurate data source for subsequent risk analysis and algorithm optimization, reducing the false detection rate of risk scenarios and lowering the cost of storing and processing invalid data.

[0007] In one possible implementation, the sensor data includes the target distance; acquiring the sensor data collected by the preset vehicle during parking includes: The system receives sensor data uploaded by the preset vehicle during the parking process, which is taken before and after a preset time point and for a preset duration. The preset time point is the time when the preset vehicle triggers a parking risk condition. The parking risk condition is the detection that a manual intervention condition has been triggered and the presence of a target with a distance less than a first preset distance threshold in the sensor data.

[0008] As can be seen from the above embodiments, in traditional automatic parking collision scenarios, manual review of fragmented sensor data is required after an accident. This not only makes it difficult to fully reconstruct the environment and vehicle state before and after the accident, but also consumes a significant amount of manpower and time for data filtering and reproduction. This embodiment uses manual intervention and close-range target detection as trigger conditions for data acquisition, ensuring the collection of high-risk parking scenario data. Simultaneously, by setting a preset duration, it can completely capture environmental data before and during high-risk parking collision events, achieving a complete reconstruction of the parking collision scenario. This provides complete and accurate data for subsequent algorithm optimization, thereby improving the accuracy of algorithm optimization.

[0009] In one possible implementation, the false detection risk includes the risk of missed detection and the risk of false detection; the ultrasonic data includes the ultrasonic target; The step of determining whether the ultrasonic data has a false detection risk based on the type of the visual target includes: Determine whether the type of the visual target is a risk target type; the risk target type includes at least one of the following: suspended obstacle, slender obstacle, regular geometric obstacle, irregular obstacle, and sharp obstacle; If the type of the visual target is the risk target type, and there is no ultrasonic target that matches the visual target, then the ultrasonic data is determined to have a risk of missed detection. If the type of the visual target is the risk target type, and there is an ultrasonic target that matches the visual target, then the ultrasonic data is determined to have a risk of false detection.

[0010] As can be seen from the above embodiments, this embodiment determines the risk of missed detection / false detection based on the matching relationship between visual targets and ultrasonic targets. This avoids the problem in the prior art that the risk of parking collision cannot be accurately associated with the physical defects of the sensors by manually backtracking to infer the risk. Furthermore, the parking algorithm is optimized in a closed loop by screening sensor data with parking false detection risk, which significantly improves the pertinence and efficiency of the algorithm improvement.

[0011] In one possible implementation, the ultrasonic data includes ultrasonic target distance; the false detection risk includes the risk of false detection of low-lying obstacles; The step of determining whether the ultrasonic data has a false detection risk based on the type of the visual target also includes: If the type of the visual target is not the type of the risk target, and the rate of change of the ultrasonic target distance is greater than the duration of the vehicle speed being greater than a first time threshold, then the ultrasonic data is determined to have a risk of false detection of low obstacles.

[0012] As can be seen from the above embodiments, existing visual sensors cannot effectively detect low-lying obstacles because their field of view cannot effectively cover them. Furthermore, ultrasonic sensors often misinterpret changes in distance to low-lying obstacles as noise, resulting in the complete failure to identify potential collision risks in such scenarios. This embodiment avoids the shortcomings of blind spots. When the visual sensor fails to detect a risky target, it determines whether there is a risk of false detection of a low-lying obstacle by comparing the rate of change of ultrasonic distance with the vehicle's speed. This allows for accurate detection of collision risks from low-lying obstacles, significantly improving the coverage of potential collision scenarios. It provides accurate risk data support for subsequent optimization of low-lying obstacle perception algorithms, preventing close-range parking collisions caused by blind spots.

[0013] In one possible implementation, the driving parameters include vehicle speed; the visual data also includes visual target distance; and the preset risk conditions are: the vehicle speed is less than a preset speed threshold, the visual target distance is less than a second preset distance threshold, and the vehicle's driving direction is towards the visual target.

[0014] As can be seen from the above embodiments, this embodiment determines whether there is a high parking collision risk in the current scenario by the speed at which the vehicle approaches the obstacle target and the distance between the vehicle and the obstacle target. If there is a high parking collision risk, the sensor data with the risk of false detection will be recorded; otherwise, the sensor data will be discarded. This can effectively filter out sensor data that is not related to parking collision risk, improve the accuracy of parking collision risk identification, solve the problem of risk misjudgment and invalid data redundancy caused by the lack of scenario limitations in traditional solutions, and reduce the redundancy cost of subsequent data processing.

[0015] In one possible implementation, the driving parameters include vehicle speed; the sensor data includes target distance; The recorded sensor data that is at risk of false detection includes: The risk level of sensor data at risk of false detection is determined based on the target distance and vehicle speed. Record sensor data that pose a risk of false detection and the corresponding risk level.

[0016] As can be seen from the above embodiments, this embodiment determines the risk level of sensor data based on the target distance and vehicle speed. The closer the target distance and the faster the vehicle speed, the higher the risk level. Sensor data with false detection risk and risk level are recorded simultaneously. This can avoid the problem of delayed response in high-risk scenarios caused by not prioritizing risk data identification in the prior art. It enables R&D and maintenance personnel to prioritize high-risk data, avoid resource misallocation, and improve the efficiency and pertinence of parking collision risk management.

[0017] In one possible implementation, the visual data includes visual target distance; the ultrasonic data includes ultrasonic target distance; The method of determining the risk level of sensor data with a risk of false detection based on target distance and vehicle speed includes: The minimum value between the visual target distance and the ultrasonic target distance corresponding to the same target is taken as the target distance, and a distance factor is determined based on the target distance; the target distance and the distance factor are negatively correlated. A speed factor is determined based on the vehicle speed; the vehicle speed and the speed factor are positively correlated. The risk score of the sensor data is obtained by weighted summation of the distance factor and the velocity factor. Based on the risk score and the level thresholds corresponding to each risk level, the risk level of the sensor data is determined; the risk level includes emergency risk, high risk, and medium risk.

[0018] As can be seen from the above embodiments, this embodiment selects the minimum of the visual target distance and the ultrasonic target distance as the target distance, determines the distance factor according to the negative correlation, and determines the speed factor according to the positive correlation of vehicle speed. Finally, the risk level is obtained by weighted summation. This can solve the problem of inaccurate judgment of static classification rules in the prior art, realize the quantitative assessment of risk, avoid excessive alarms or insufficient alarms caused by misjudging the urgency level, and further improve the accuracy and reliability of risk management.

[0019] In one possible implementation, the risk levels include emergency risk, high risk, and medium risk; After determining the risk level of sensor data with a risk of false detection based on target distance and vehicle speed, the method further includes: If the risk level of the sensor data is emergency risk or high risk, the sensor data and risk level will be pushed to the terminal of the relevant staff. If the risk level of the sensor data is medium risk, then the sensor data will be stored in the log.

[0020] As can be seen from the above embodiments, this embodiment achieves risk-level response to collected sensor data by pushing emergency / high-risk data to staff terminals and storing medium-risk data in logs, thereby improving the processing efficiency of parking collision risk scenarios, avoiding batch accidents, and solving the problem of delayed high-risk response and information overload caused by unified storage of all risk data. The processing of storing medium-risk sensor data in logs can balance the real-time nature and efficiency of the data, without occupying immediate processing resources, and can also cover low- and medium-risk scenarios through analysis the next day, avoiding omissions, ensuring that risk data is not lost, and providing complete data support for subsequent algorithm optimization.

[0021] Secondly, this application provides a parking risk data collection device, comprising: The sensor data acquisition module is used to acquire sensor data collected by a preset vehicle during parking. The sensor data includes visual data and ultrasonic data. The visual data includes visual targets and the types of the visual targets. The false detection risk monitoring module is used to determine whether there is a false detection risk in the ultrasonic data based on the type of the visual target. The risk data acquisition module is used to record sensor data with a risk of false detection if the ultrasonic data has a risk of false detection and the driving parameters of the preset vehicle meet the preset risk conditions.

[0022] In one possible implementation, the sensor data includes the target distance; the sensor data acquisition module includes: The system receives sensor data uploaded by the preset vehicle during the parking process, which is taken before and after a preset time point and for a preset duration. The preset time point is the time when the preset vehicle triggers a parking risk condition. The parking risk condition is the detection that a manual intervention condition has been triggered and the presence of a target with a distance less than a first preset distance threshold in the sensor data.

[0023] In one possible implementation, the false detection risk includes the risk of missed detection and the risk of false detection; the ultrasonic data includes ultrasonic targets; the false detection risk monitoring module includes: A risk target type determination unit is used to determine whether the type of the visual target is a risk target type; the risk target type includes at least one of suspended obstacles, slender obstacles, regular geometric obstacles, irregular obstacles, and sharp obstacles; The missed detection risk monitoring unit is used to determine that the ultrasonic data has a missed detection risk if the type of the visual target is the risk target type and there is no ultrasonic target matching the visual target. The false detection risk monitoring unit is used to determine that the ultrasonic data has a false detection risk if the type of the visual target is the risk target type and there is an ultrasonic target that matches the visual target.

[0024] In one possible implementation, the ultrasonic data includes the ultrasonic target distance; the false detection risk includes the false detection risk of low obstacles; the false detection risk monitoring module further includes: The low obstacle false detection risk monitoring unit is used to determine that the ultrasonic data has a low obstacle false detection risk if the type of the visual target is not the risk target type and the change rate of the ultrasonic target distance is greater than the duration of the vehicle driving speed is greater than a first time threshold.

[0025] In one possible implementation, the driving parameters include vehicle speed; the visual data also includes visual target distance; and the preset risk conditions are: the vehicle speed is less than a preset speed threshold, the visual target distance is less than a second preset distance threshold, and the vehicle's driving direction is towards the visual target.

[0026] In one possible implementation, the driving parameters include vehicle speed; the sensor data includes target distance; the risk data acquisition module includes: The risk level calculation unit is used to determine the risk level of sensor data that may have a false detection risk based on the target distance and vehicle speed. The data recording unit is used to record sensor data that poses a risk of false detection and the corresponding risk level.

[0027] In one possible implementation, the visual data includes visual target distance; the ultrasonic data includes ultrasonic target distance; the risk level calculation unit includes: The minimum value between the visual target distance and the ultrasonic target distance corresponding to the same target is taken as the target distance, and a distance factor is determined based on the target distance; the target distance and the distance factor are negatively correlated. A speed factor is determined based on the vehicle speed; the vehicle speed and the speed factor are positively correlated. The risk score of the sensor data is obtained by weighted summation of the distance factor and the velocity factor. Based on the risk score and the level thresholds corresponding to each risk level, the risk level of the sensor data is determined; the risk level includes emergency risk, high risk, and medium risk.

[0028] In one possible implementation, the risk levels include emergency risk, high risk, and medium risk; the parking risk data acquisition device further includes a graded response module for: If the risk level of the sensor data is emergency risk or high risk, the sensor data and risk level will be pushed to the terminal of the relevant staff. If the risk level of the sensor data is medium risk, then the sensor data will be stored in the log.

[0029] Thirdly, this application provides a terminal 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 the steps of the method as described in the possible implementations of the first aspect above.

[0030] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method as described in any possible implementation of the first aspect above. Attached Figure Description

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0032] Figure 1 This is a flowchart illustrating the implementation of the parking risk data collection method provided in this application embodiment; Figure 2This is a schematic diagram of the interaction process between the two ends of the parking risk data collection method provided in the embodiments of this application; Figure 3 This is a schematic diagram of the parking risk data acquisition device provided in the embodiments of this application; Figure 4 This is a schematic diagram of the terminal provided in the embodiments of this application. Detailed Implementation

[0033] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0034] To make the objectives, technical solutions, and advantages of this application clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0035] The existing multi-sensor fusion detection mechanism of automatic parking systems has significant flaws, resulting in insufficient detection accuracy in complex real-world parking scenarios and difficulty in effectively detecting potential collision risks. From the perspective of sensor performance, ultrasonic sensors are prone to failure due to physical limitations. For example, when facing suspended objects such as fire hydrants or charging piles, the lack of an effective reflective surface at the bottom prevents sound waves from penetrating and forming an echo, leading to missed detections. For regular geometric structures such as right-angled walls and columns, when the ultrasonic beam is not incident perpendicularly, the sound waves undergo specular reflection according to the law that the angle of incidence equals the angle of reflection, causing the echo path to deviate from the receiver. If the angle between the ultrasonic sensor and the wall exceeds 30°, the echo signal attenuates by more than 70%, and the system is prone to misjudging it as an unobstructed object. For irregular surfaces such as shrubs and vegetation, or sharp objects such as the angle iron of trucks, sound waves are scattered, energy is absorbed, or the concentrated reflection points are misjudged as noise by the filtering algorithm, leading to problems such as ranging jumps or excessively large deviations in distance values ​​from the actual values. Visual sensors are typically monocular fisheye cameras, which also have inherent detection biases. Specifically, due to limitations in installation location and viewing angle, monocular fisheye cameras create blind spots for obstacles such as road shoulders and low mounds with a height of less than 30cm. Secondly, in low-contrast scenes where the obstacle and the ground are similar in color, texture features disappear, causing visual algorithms to fail to extract effective contours, and the ranging deviation of a single eye often exceeds ±10cm. These accuracy deficiencies of the multi-sensor fusion detection mechanism directly lead to poor detection accuracy of potential collision risks in parking scenarios, making it difficult to identify and avoid scrapes and collisions when parking close to the vehicle.

[0036] This embodiment addresses the problem of poor accuracy in detecting potential collision risks in the aforementioned parking scenarios by providing a parking risk data acquisition method. This method can collect sensor data in real time under parking collision risk scenarios, generate closed-loop optimization labels to drive parking algorithm iteration, thereby improving the coverage of potential collision risk scenarios and the efficiency of algorithm iteration.

[0037] See Figure 1 The document illustrates a flowchart of the parking risk data collection method provided in this embodiment. The method is executed in the cloud, and is described in detail below: S101: Acquire sensor data collected by a preset vehicle during parking, the sensor data including visual data and ultrasonic data; the visual data includes visual targets and the type of the visual targets.

[0038] In this embodiment, visual data includes the visual target, the visual target distance, and the visual target type. Ultrasonic data includes the ultrasonic target, the ultrasonic target distance, and the ultrasonic target type.

[0039] Specifically, in a parking scenario, the vehicle uses visual sensors to collect near-range visual images and ultrasonic sensors to collect near-range ultrasonic data. The onboard terminal uses the visual images to determine if any obstacles exist within the detection range, and identifies these obstacles as visual targets. The type of the visual target and the distance between the vehicle and the visual target (i.e., the visual target distance) are then determined. Specifically, this embodiment can pre-train the visual algorithm using a dataset of visual images labeled with target types to ensure accurate identification of risky targets. The visual algorithm can be a traditional machine learning algorithm or a deep learning algorithm.

[0040] The vehicle-mounted terminal is also used to determine whether there are obstacle targets within the detection range based on ultrasonic data, and to use the obstacle targets determined based on ultrasonic data as obstacle targets, determine the type of obstacle targets and the distance between the vehicle and the ultrasonic targets, i.e., the ultrasonic target distance.

[0041] When the vehicle terminal detects that the vehicle is in a parking collision risk scenario, it can upload the sensor data of the current parking collision risk scenario to the cloud.

[0042] In this embodiment, while uploading sensor data of the current parking collision risk scenario to the cloud, the vehicle terminal can also collect the vehicle's driving parameters and basic parameters at this time. The driving parameters include vehicle speed, gear, steering wheel angle, and acceleration / deceleration pedal opening. The basic parameters include the vehicle's unique identifier, the trigger time of the parking collision risk event, and the time corresponding to the collected sensor data.

[0043] After receiving the aforementioned sensor data, vehicle form parameters, and basic parameters, the vehicle terminal can perform format verification on the acquired data, such as checking whether vehicle identification is missing or whether the sensor data is valid. Then, it can perform data cleaning to remove obvious outliers and ensure data availability.

[0044] S102: Determine whether there is a risk of false detection in the ultrasonic data based on the type of the visual target.

[0045] In this embodiment, the visual images acquired by the visual sensor can usually identify the type of obstacle target relatively accurately. For example, the visual images can accurately identify slender obstacles, suspended obstacles, pillars, right-angle walls, sharp obstacles, bushes and grass, and other irregular obstacles.

[0046] The risks of false detection in ultrasonic data include the risk of missed detection and the risk of false detection. In this embodiment, the effectiveness of ultrasonic data can be verified by using the visual target type as a reference standard. Since vision is more intuitive in recognizing the type of obstacle, while ultrasonic waves are prone to failure due to physical characteristics, this embodiment determines whether there is a risk of missed detection or false detection in ultrasonic data by comparing the target type, target distance, and target presence detected by ultrasonic data and visual data.

[0047] S103: If the ultrasonic data has a risk of false detection and the driving parameters of the preset vehicle meet the preset risk conditions, then the sensor data with the risk of false detection is recorded.

[0048] In this embodiment, the vehicle's driving parameters can be used to determine whether the vehicle faces a high collision risk. If the vehicle faces a high collision risk and the ultrasonic data has a risk of false detection, it is necessary to collect the sensor data corresponding to the high collision risk scenario. This will allow the staff to perform closed-loop optimization of the parking algorithm after obtaining the sensor data of the high collision risk scenario during the parking process, so as to avoid the occurrence of collision events when facing the scenario in the future.

[0049] As can be seen from the above embodiments, this embodiment, by acquiring visual and ultrasonic data, can determine whether there is a risk of false detection in ultrasonic data based on the type of visual target. When there is a risk of false detection in ultrasonic data, vehicle driving parameters are used to exclude non-parking risk scenarios, ensuring that only sensor data with genuine parking collision hazards are recorded. This method can immediately acquire the corresponding sensor data at the moment a parking collision risk occurs. This not only avoids the lag caused by obtaining problematic data after a collision in traditional solutions, but also provides an accurate data source for subsequent risk analysis and algorithm optimization, laying a data foundation for closed-loop optimization of potential parking collision scenarios, while reducing the cost of storing and processing invalid data.

[0050] In one possible implementation, the sensor data includes the target distance; Figure 2 This diagram illustrates the interaction flow between the two ends of the parking risk data collection method. (Refer to...) Figure 2 The specific implementation process of S101 includes: The system receives sensor data uploaded by the preset vehicle during the parking process, which is taken before and after a preset time point and for a preset duration. The preset time point is the time when the preset vehicle triggers a parking risk condition. The parking risk condition is the detection that a manual intervention condition has been triggered and the presence of a target with a distance less than a first preset distance threshold in the sensor data.

[0051] In this embodiment, the parking risk condition includes two sub-conditions. The vehicle terminal will only upload sensor data to the cloud when both sub-conditions are triggered, in order to avoid uploading invalid data.

[0052] Specifically, one of the sub-conditions is the manual takeover condition, which includes steering wheel takeover, gear shift takeover, and brake takeover. When the vehicle terminal detects any one of these manual takeover events, it is considered that the manual takeover condition has been triggered. Since manual takeover is usually an operation performed after the user perceives a system abnormality, it is a direct signal of potential system failure. Therefore, determining whether the manual takeover condition has been triggered can effectively determine whether there is a parking collision risk that the system has not detected.

[0053] Another sub-condition is whether the target distance of the detected obstacle is less than a first preset distance threshold. Since the risk of parking collisions in daily life is usually within 1m, the first preset distance threshold can be set to 0.5m to 1.5m to determine that the uploaded sensor data is sensor data from a high-risk close-range scenario. When both of the above conditions are met, the vehicle terminal determines that there is a parking collision risk. Therefore, the time point at which the parking collision risk is detected can be used as a preset time point. Using the preset time point as the midpoint, the sensor data, vehicle driving parameters and basic parameters for a preset duration before and after the preset time point are packaged and sent to the cloud. This avoids meaningless full-time data collection, reduces the amount of cloud computing, and improves the collection efficiency of sensor data with the risk of false detection.

[0054] For example, the preset duration can be 20 seconds, which means uploading sensor data 20 seconds before and 20 seconds after the preset time point to the cloud.

[0055] As can be seen from the above embodiments, in the case of a collision in the traditional automatic parking system, where manual review of scattered sensor data is required after the accident, it is not only difficult to fully reconstruct the environment and vehicle state before and after the accident, but also consumes a lot of manpower and time for data screening and reproduction. This embodiment uses manual intervention and close-range target detection as the trigger conditions for data collection to ensure that high-risk parking scenario data is collected. At the same time, by setting a preset duration, it can fully capture the environmental data before the high-risk parking collision event and the state at the time of the event, realizing a complete reconstruction of the parking collision accident scenario. This provides complete and accurate data for subsequent algorithm optimization, thereby improving the accuracy of algorithm optimization.

[0056] In one possible implementation, the false detection risk includes the risk of missed detection and the risk of false detection; the ultrasonic data includes the ultrasonic target; the specific implementation process of S102 includes: Determine whether the type of the visual target is a risk target type; the risk target type includes at least one of the following: suspended obstacle, slender obstacle, regular geometric obstacle, irregular obstacle, and sharp obstacle; If the type of the visual target is the risk target type, and there is no ultrasonic target that matches the visual target, then the ultrasonic data is determined to have a risk of missed detection. If the type of the visual target is the risk target type, and there is an ultrasonic target that matches the visual target, then the ultrasonic data is determined to have a risk of false detection.

[0057] In this embodiment, reference Figure 2 After acquiring sensor data, the cloud first needs to determine whether the visual target is a risky target. Specifically, the cloud can first determine whether the visual target is a suspended obstacle, such as a fire hydrant, charging pile, or truck bed. Since ultrasound relies on sound wave reflection, but suspended objects such as fire hydrants, charging piles, and truck beds have no reflective surface at their bottom, sound waves cannot echo back after penetration, leading to missed detections by ultrasonic sensors. Therefore, the accuracy of ultrasonic sensors in identifying suspended obstacles is very low. To simplify the process and improve data acquisition efficiency, this embodiment can directly determine the risk of missed detection when a visual target is detected as a suspended obstacle, without going through the matching detection step between the visual target and the ultrasonic target.

[0058] Secondly, after determining that the visual target is not a suspended obstacle, the system can further determine whether the visual target is a special obstacle such as a thin pole or a parking lock, or a regular geometric obstacle, an irregular obstacle, or a sharp obstacle. Among these, special obstacles such as thin poles and parking locks are prone to missed detection due to their small ultrasonic wave reflection cross-section; irregular obstacles, including bushes and vegetation, cause sound wave scattering and are prone to false detection; regular geometric obstacles, such as garage pillars and wall corners, often exhibit specular reflection and are prone to false detection; and sharp obstacles, such as truck angle irons and metal sharp corners, are prone to false detection due to concentrated reflection points. If the visual target falls into any of these categories, it is considered that there may be a risk of false detection using ultrasonic waves. Therefore, the ultrasonic data is further examined to see if there is an ultrasonic target matching the visual target. If so, the ultrasonic data is determined to have a risk of false detection; otherwise, it is determined that the ultrasonic data has a risk of missed detection.

[0059] In this embodiment, when matching visual and ultrasonic targets, the time of the visual and ultrasonic targets is first calibrated using the vehicle CAN bus timestamp. Then, the two types of data are converted to the same coordinate system at the same time. If there is a target in the two types of data at the same time with a spatial overlap greater than or equal to a preset overlap threshold, it is considered that there is an ultrasonic target in the ultrasonic data that matches the visual target. The preset overlap can be 80%.

[0060] As can be seen from the above embodiments, this embodiment accurately determines the type of false detection based on the matching relationship between visual and ultrasonic targets. This avoids the problem in existing technologies where manual backtracking to infer parking collision risks cannot accurately correlate with sensor physical defects. Furthermore, by using the sensor data identified as having parking false detection risks, the parking algorithm is optimized in a closed loop, significantly improving the targeting and efficiency of algorithm improvement. In addition, the sensor data provided in this embodiment can also accurately locate the types of falsely detected obstacles, providing a clear direction for subsequent algorithm optimization.

[0061] In one possible implementation, the ultrasonic data includes the ultrasonic target distance; the false detection risk includes the false detection risk of low-lying obstacles; reference Figure 2 The specific implementation process of S102 also includes: If the type of the visual target is not the type of the risk target, and the rate of change of the ultrasonic target distance is greater than the duration of the vehicle's driving speed being greater than a first time threshold, then the ultrasonic data is determined to have a risk of false detection of low obstacles.

[0062] Specifically, since low-lying obstacles are often missed by visual sensors because they are outside the field of view of a monocular fisheye camera, this embodiment specifically addresses the risk of missed detection of low-lying obstacles caused by visual blind spots. When the visual sensor does not detect any risky targets, that is, there is no risk of missed or false detection of ultrasonic data, the presence of a risk of false detection of low-lying obstacles in the ultrasonic data is determined by detecting whether the distance of the ultrasonic target is abnormal.

[0063] Specifically, this embodiment can determine whether the ultrasonic data is abnormal by detecting abnormal jumps in the ultrasonic target distance. Normally, the ultrasonic target distance changes at a rate that varies with the vehicle's speed. When the rate of change in the ultrasonic target distance is too large relative to the vehicle's speed, it indicates an abnormal jump in the ultrasonic target distance, such as a jump from 80cm to 0, or from 0 to 80cm. This suggests a situation where an obstacle suddenly appears without the vehicle actively approaching, possibly indicating a low-lying obstacle entering the ultrasonic detection range but not being detected by the visual sensor. If this abnormal situation persists for more than a first time threshold, it is determined that there is a risk of false detection of low-lying obstacles in the ultrasonic data. The first time threshold can be 0.5 seconds.

[0064] As can be seen from the above embodiments, existing visual sensors cannot effectively cover low-lying obstacles due to their limited field of view, thus failing to detect such targets effectively. Furthermore, ultrasonic sensors often misinterpret changes in distance to low-lying obstacles as noise, resulting in the complete failure to identify potential collision risks in such scenarios. This embodiment cleverly avoids the shortcomings of visual blind spots. When visual sensors fail to detect risky targets, it determines the existence of false detection risks of low-lying obstacles by comparing the rate of change of ultrasonic distance with the vehicle's speed. This enables accurate detection of collision risks from low-lying obstacles, significantly improving the coverage of potential collision scenarios and providing accurate risk data support for subsequent optimization of low-lying obstacle perception algorithms, thus avoiding close-range parking collisions caused by visual blind spots.

[0065] In one possible implementation, the driving parameters include vehicle speed; the visual data also includes visual target distance; and the preset risk conditions are: the vehicle speed is less than a preset speed threshold, the visual target distance is less than a second preset distance threshold, and the vehicle's driving direction is towards the visual target.

[0066] This embodiment uses a combination of three dimensions—vehicle speed, visual target distance, and driving direction—to ensure that only scenario data with a genuine risk of parking collision is recorded, thus avoiding the collection of invalid data.

[0067] Specifically, regarding vehicle speed, the cloud can determine whether the vehicle is in a parking scenario by judging whether the vehicle speed is less than a preset speed threshold. When the vehicle speed is greater than the preset speed threshold, the vehicle is in a driving state rather than a parking operation state. Even if there is a sensor misdetection, the user can brake quickly, and the risk of collision is low.

[0068] Specifically, in this embodiment, real-time vehicle speed can be obtained through the onboard ABS (Anti-lock Braking System) system or wheel speed sensors.

[0069] Regarding the target distance condition, the cloud can determine whether there is a risk of near-collision by judging whether the visual target distance is less than a second preset distance threshold. When the visual target distance is less than the second preset distance threshold, even a slight deviation in parking operation may cause a scratch. Therefore, the target distance condition can be used to filter out sensor data with a high risk of collision. When making this judgment, the visual target distance can be selected as the judgment parameter, or the minimum value between the visual target distance and the ultrasonic target distance can be selected as the target distance, and then it can be judged whether this target distance is less than the second preset distance threshold. The second preset distance threshold is less than or equal to 30cm.

[0070] Regarding the vehicle's direction of travel, there is obviously no collision risk when the vehicle is moving away from the obstacle target. Therefore, this embodiment requires detecting that the vehicle is moving closer to the visual target before determining that there is a parking collision risk. Specifically, the relative angle between the obstacle target and the vehicle's direction of travel can be used to determine whether the vehicle is about to approach the obstacle target.

[0071] Taking into account the above three conditions, when the vehicle speed is less than the preset speed threshold, the visual target distance is less than the second preset distance threshold, and the vehicle's driving direction is towards the visual target, it is determined that there is a high risk of parking collision. Therefore, it is necessary to record the sensor data that has been filtered out and has the risk of false detection.

[0072] As can be seen from the above embodiments, this embodiment determines whether there is a parking collision risk in the current scenario by the speed at which the vehicle approaches the obstacle target and the distance between the vehicle and the obstacle target. If there is a parking collision risk, the sensor data with the risk of false detection will be recorded; otherwise, the sensor data will be discarded. This can effectively filter out sensor data that is not related to parking collision risk, improve the accuracy of parking collision risk identification, solve the problem of risk misjudgment and invalid data redundancy caused by the lack of scenario limitations in traditional solutions, and reduce the redundancy cost of subsequent data processing.

[0073] In one possible implementation, the driving parameters include vehicle speed; the sensor data includes target distance; the specific implementation process of S103 includes: S201: Determine the risk level of sensor data that poses a risk of false detection based on the target distance and vehicle speed.

[0074] S202: Record sensor data that poses a risk of false detection and the corresponding risk level.

[0075] In this embodiment, the faster the vehicle speed and the closer the target distance, the shorter the reaction time left for the user or system, and the higher the risk of parking collision. Therefore, this embodiment can calculate the risk level of the selected sensor data by combining the target distance and vehicle speed.

[0076] Finally, sensor data with potential false detection risks, along with the risk level, false detection type, and parking risk assessment criteria, are recorded and stored as risk data. These criteria include target distance and vehicle speed.

[0077] As can be seen from the above embodiments, this embodiment determines the risk level of sensor data based on the target distance and vehicle speed. The closer the target distance and the faster the vehicle speed, the higher the risk level. Sensor data with false detection risk and risk level are recorded simultaneously. This can avoid the problem of delayed response in high-risk scenarios caused by not prioritizing risk data identification in the prior art. It enables R&D and maintenance personnel to prioritize high-risk data, avoid resource misallocation, and improve the efficiency and pertinence of parking collision risk management.

[0078] In one possible implementation, the visual data includes the visual target distance; the ultrasonic data includes the ultrasonic target distance; the specific implementation process of S201 includes: The minimum value between the visual target distance and the ultrasonic target distance corresponding to the same target is taken as the target distance, and a distance factor is determined based on the target distance; the target distance and the distance factor are negatively correlated. A speed factor is determined based on the vehicle speed; the vehicle speed and the speed factor are positively correlated. The risk score of the sensor data is obtained by weighted summation of the distance factor and the velocity factor. Based on the risk score and the level thresholds corresponding to each risk level, the risk level of the sensor data is determined; the risk level includes emergency risk, high risk, and medium risk.

[0079] In this embodiment, the target distance can be converted into a distance factor between 0 and 1, and the vehicle speed can be converted into a speed factor between 0 and 1. Then, the distance factor and the speed factor are weighted and summed to obtain a risk score between 0 and 1.

[0080] When the risk score is greater than the first score, the sensor data is classified as having an urgent risk level. When the risk score is not greater than the first score but is greater than the second score, the sensor data is classified as having a high risk level. When the risk score is not greater than the second score, the sensor data is classified as having a medium risk level. The first score can be 0.7 and the second score can be 0.4.

[0081] As can be seen from the above embodiments, this embodiment selects the minimum value between the visual target distance and the ultrasonic target distance as the target distance, determines the distance factor according to the negative correlation, and determines the speed factor according to the positive correlation of vehicle speed. Finally, the risk level is obtained by weighted summation. This can solve the problems of coarse classification and inaccurate risk level determination in traditional static rules, realize the quantitative assessment of risk, avoid excessive or insufficient alarms caused by misjudging the urgency, and further improve the accuracy and reliability of risk management.

[0082] In one possible implementation, the risk levels include emergency risk, high risk, and medium risk; after S201, the method provided in this embodiment further includes: If the risk level of the sensor data is emergency risk or high risk, the sensor data and risk level will be pushed to the terminal of the relevant staff. If the risk level of the sensor data is medium risk, then the sensor data will be stored in the log.

[0083] In this embodiment, emergency risks require immediate response and therefore need to be pushed to relevant staff immediately. High risks have a slightly shorter response time than emergency risks, but staff still need to be notified. Medium risks can be processed in batches the next day to avoid resource misallocation.

[0084] Specifically, emergency risks can be immediately sent to the mobile devices of relevant staff via SMS or email. High-risk risks can be sent to the relevant work devices, such as work computers, via push notification software. Medium-risk risks are directly stored in the log, and staff can view the risk data in the log report the following day.

[0085] In this embodiment, the false detection types of each risk data can also be statistically analyzed to determine the frequency of occurrence of each false detection type. Then, for risk data of the same risk level, the user is instructed to prioritize optimizing the detection algorithm for false detection data with high occurrence frequency, thereby avoiding the spread of risk.

[0086] In this embodiment, risk data of different risk levels can also be stored in layers. Emergency and high-risk data are stored on a high-speed cloud disk, while medium-risk data is stored on an archive cloud disk, balancing access efficiency and storage costs. Simultaneously, the retention period for emergency and high-risk data can be set to be longer than that for medium-risk data to avoid wasting storage resources.

[0087] As can be seen from the above embodiments, this embodiment achieves risk-level response to collected sensor data by pushing emergency / high-risk data to staff terminals and storing medium-risk data in logs, thereby improving the processing efficiency of parking collision risk scenarios, avoiding batch accidents, and solving the problem of delayed high-risk response and information overload caused by unified storage of all risk data. This ensures that high-risk parking collision scenarios are detected in a timely manner, and staff can immediately download data packages to reproduce the problem and optimize algorithms to prevent the spread of risk. The processing of medium-risk sensor data by storing it in logs can balance the real-time nature and efficiency of the data. It does not occupy immediate processing resources and can also cover low- and medium-risk scenarios through analysis the next day, avoiding omissions and ensuring that risk data is not lost, providing complete data support for subsequent algorithm optimization.

[0088] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0089] The following are device embodiments of this application. For details not described in detail, please refer to the corresponding method embodiments described above.

[0090] Figure 3 A schematic diagram of the parking risk data acquisition device provided in an embodiment of this application is shown. For ease of explanation, only the parts related to the embodiment of this application are shown, and are described in detail below: like Figure 3 As shown, the parking risk data acquisition device 100 includes: The sensor data acquisition module 110 is used to acquire sensor data collected by a preset vehicle during parking. The sensor data includes visual data and ultrasonic data. The visual data includes visual targets and the types of the visual targets. The false detection risk monitoring module 120 is used to determine whether there is a false detection risk in the ultrasonic data based on the type of the visual target. The risk data acquisition module 130 is used to record sensor data with a risk of false detection if the ultrasonic data has a risk of false detection and the driving parameters of the preset vehicle meet the preset risk conditions.

[0091] In one possible implementation, the sensor data includes the target distance; the sensor data acquisition module 110 includes: The system receives sensor data uploaded by the preset vehicle during the parking process, which is taken before and after a preset time point and for a preset duration. The preset time point is the time when the preset vehicle triggers a parking risk condition. The parking risk condition is the detection that a manual intervention condition has been triggered and the presence of a target with a distance less than a first preset distance threshold in the sensor data.

[0092] In one possible implementation, the false detection risk includes the risk of missed detection and the risk of false detection; the ultrasonic data includes ultrasonic targets; the false detection risk monitoring module 120 includes: A risk target type determination unit is used to determine whether the type of the visual target is a risk target type; the risk target type includes at least one of suspended obstacles, slender obstacles, regular geometric obstacles, irregular obstacles, and sharp obstacles; The missed detection risk monitoring unit is used to determine that the ultrasonic data has a missed detection risk if the type of the visual target is the risk target type and there is no ultrasonic target matching the visual target. The false detection risk monitoring unit is used to determine that the ultrasonic data has a false detection risk if the type of the visual target is the risk target type and there is an ultrasonic target that matches the visual target.

[0093] In one possible implementation, the ultrasonic data includes the ultrasonic target distance; the false detection risk includes the false detection risk of low obstacles; the false detection risk monitoring module 120 further includes: The low obstacle false detection risk monitoring unit is used to determine that the ultrasonic data has a low obstacle false detection risk if the type of the visual target is not the risk target type and the change rate of the ultrasonic target distance is greater than the duration of the vehicle driving speed is greater than a first time threshold.

[0094] In one possible implementation, the driving parameters include vehicle speed; the visual data also includes visual target distance; and the preset risk conditions are: the vehicle speed is less than a preset speed threshold, the visual target distance is less than a second preset distance threshold, and the vehicle's driving direction is towards the visual target.

[0095] In one possible implementation, the driving parameters include vehicle speed; the sensor data includes target distance; the risk data acquisition module 130 includes: The risk level calculation unit is used to determine the risk level of sensor data that may have a false detection risk based on the target distance and vehicle speed. The data recording unit is used to record sensor data that poses a risk of false detection and the corresponding risk level.

[0096] In one possible implementation, the visual data includes visual target distance; the ultrasonic data includes ultrasonic target distance; the risk level calculation unit includes: The minimum value between the visual target distance and the ultrasonic target distance corresponding to the same target is taken as the target distance, and a distance factor is determined based on the target distance; the target distance and the distance factor are negatively correlated. A speed factor is determined based on the vehicle speed; the vehicle speed and the speed factor are positively correlated. The risk score of the sensor data is obtained by weighted summation of the distance factor and the velocity factor. Based on the risk score and the level thresholds corresponding to each risk level, the risk level of the sensor data is determined; the risk level includes emergency risk, high risk, and medium risk.

[0097] In one possible implementation, the risk levels include emergency risk, high risk, and medium risk; the parking risk data acquisition device 100 further includes a graded response module for: If the risk level of the sensor data is emergency risk or high risk, the sensor data and risk level will be pushed to the terminal of the relevant staff. If the risk level of the sensor data is medium risk, then the sensor data will be stored in the log.

[0098] This application also provides a computer program product having program code that, when run in a corresponding processor, controller, computing device, or terminal, executes the steps in any of the above-described parking risk data acquisition method embodiments, for example... Figure 1 Steps S101 to S103 are shown. Those skilled in the art will understand that the methods and apparatus proposed in the embodiments of this application can be implemented in various forms, including hardware, software, firmware, dedicated processors, or combinations thereof. Dedicated processors may include application-specific integrated circuits (ASICs), reduced instruction set computers (RISCs), and / or field-programmable gate arrays (FPGAs). The proposed methods and apparatus are preferably implemented as a combination of hardware and software. The software is preferably installed as an application program on a program storage device. This is typically based on a machine with a computer platform, such as one or more central processing units (CPUs), random access memory (RAM), and one or more input / output (I / O) interfaces. An operating system is also typically installed on the computer platform. The various processes and functions described herein may be part of an application program, or a portion thereof may be executed by an operating system.

[0099] Figure 4 This is a schematic diagram of the terminal provided in an embodiment of this application. For example... Figure 4 As shown, the terminal 4 in this embodiment includes: a processor 40, a memory 41, and a computer program 42 stored in the memory 41 and executable on the processor 40. When the processor 40 executes the computer program 42, it implements the steps described in the various parking risk data acquisition method embodiments above, for example... Figure 1 Steps S101 to S103 are shown. Alternatively, when the processor 40 executes the computer program 42, it implements the functions of each module / unit in the above-described device embodiments, for example... Figure 3 The functions of modules 110 to 130 are shown.

[0100] For example, the computer program 42 may be divided into one or more modules / units, which are stored in the memory 41 and executed by the processor 40 to complete / implement the solution provided in this application. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program 42 in the terminal 4.

[0101] The terminal 4 can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. The terminal 4 may include, but is not limited to, a processor 40 and a memory 41. Those skilled in the art will understand that... Figure 4 This is merely an example of terminal 4 and does not constitute a limitation on terminal 4. It may include more or fewer components than shown, or combine certain components, or different components. For example, the terminal may also include input / output devices, network access devices, buses, etc.

[0102] The processor 40 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0103] The memory 41 can be an internal storage unit of the terminal 4, such as a hard disk or memory of the terminal 4. The memory 41 can also be an external storage device of the terminal 4, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the terminal 4. Furthermore, the memory 41 can include both internal storage units and external storage devices of the terminal 4. The memory 41 is used to store the computer program and other programs and data required by the terminal. The memory 41 can also be used to temporarily store data that has been output or will be output.

[0104] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0105] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0106] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0107] In the embodiments provided in this application, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0108] 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 units can be selected to achieve the purpose of this embodiment according to actual needs.

[0109] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0110] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above-described embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various parking risk data collection method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0111] Furthermore, the features of the embodiments shown in the accompanying drawings or the various embodiments mentioned in this specification should not be construed as independent embodiments. Rather, each feature described in one example of an embodiment can be combined with one or more other desired features from other embodiments to produce other embodiments not described in words or with reference to the accompanying drawings.

[0112] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A method for collecting parking risk data, characterized in that, include: Acquire sensor data collected during the parking process of a preset vehicle, the sensor data including visual data and ultrasonic data; the visual data includes visual targets and the types of the visual targets; Based on the type of the visual target, determine whether the ultrasonic data has a risk of false detection; If the ultrasonic data has a risk of false detection and the driving parameters of the preset vehicle meet the preset risk conditions, then the sensor data with the risk of false detection is recorded.

2. The parking risk data collection method according to claim 1, characterized in that, The sensor data includes the target distance; the acquisition of sensor data collected by the preset vehicle during parking includes: The system receives sensor data uploaded by the preset vehicle during the parking process, which is taken before and after a preset time point and for a preset duration. The preset time point is the time when the preset vehicle triggers a parking risk condition. The parking risk condition is the detection that a manual intervention condition has been triggered and the presence of a target with a distance less than a first preset distance threshold in the sensor data.

3. The parking risk data collection method according to claim 1, characterized in that, The risk of false detection includes the risk of missed detection and the risk of false detection; the ultrasonic data includes ultrasonic targets; The step of determining whether the ultrasonic data has a false detection risk based on the type of the visual target includes: Determine whether the type of the visual target is a risk target type; the risk target type includes at least one of the following: suspended obstacle, slender obstacle, regular geometric obstacle, irregular obstacle, and sharp obstacle; If the type of the visual target is the risk target type, and there is no ultrasonic target that matches the visual target, then the ultrasonic data is determined to have a risk of missed detection. If the type of the visual target is the risk target type, and there is an ultrasonic target that matches the visual target, then the ultrasonic data is determined to have a risk of false detection.

4. The parking risk data collection method according to claim 3, characterized in that, The ultrasonic data includes the ultrasonic target distance; the false detection risk also includes the risk of false detection by low obstacles; The step of determining whether the ultrasonic data has a false detection risk based on the type of the visual target also includes: If the type of the visual target is not the type of the risk target, and the rate of change of the ultrasonic target distance is greater than the duration of the vehicle speed being greater than a first time threshold, then the ultrasonic data is determined to have a risk of false detection of low obstacles.

5. The parking risk data collection method according to claim 1, characterized in that, The driving parameters include vehicle speed; the visual data also includes visual target distance; the preset risk conditions are: the vehicle speed is less than a preset speed threshold, the visual target distance is less than a second preset distance threshold, and the vehicle's driving direction is towards the visual target.

6. The parking risk data collection method according to claim 1, characterized in that, The driving parameters include vehicle speed; the sensor data includes target distance; The recorded sensor data that is at risk of false detection includes: The risk level of sensor data at risk of false detection is determined based on the target distance and vehicle speed. Record sensor data that pose a risk of false detection and the corresponding risk level.

7. The parking risk data collection method according to claim 6, characterized in that, The visual data includes the visual target distance; the ultrasonic data includes the ultrasonic target distance. The method of determining the risk level of sensor data with a risk of false detection based on target distance and vehicle speed includes: The minimum value between the visual target distance and the ultrasonic target distance corresponding to the same target is taken as the target distance, and a distance factor is determined based on the target distance; the target distance and the distance factor are negatively correlated. A speed factor is determined based on the vehicle speed; the vehicle speed and the speed factor are positively correlated. The risk score of the sensor data is obtained by weighted summation of the distance factor and the velocity factor. Based on the risk score and the level thresholds corresponding to each risk level, the risk level of the sensor data is determined; the risk level includes emergency risk, high risk, and medium risk.

8. The parking risk data collection method according to claim 6, characterized in that, The risk levels include emergency risk, high risk, and medium risk. After determining the risk level of sensor data with a risk of false detection based on target distance and vehicle speed, the method further includes: If the risk level of the sensor data is emergency risk or high risk, the sensor data and risk level will be pushed to the terminal of the relevant staff. If the risk level of the sensor data is medium risk, then the sensor data will be stored in the log.

9. A parking risk data acquisition device, characterized in that, include: The sensor data acquisition module is used to acquire sensor data collected by a preset vehicle during parking. The sensor data includes visual data and ultrasonic data. The visual data includes visual targets and the types of the visual targets. The false detection risk monitoring module is used to determine whether there is a false detection risk in the ultrasonic data based on the type of the visual target. The risk data acquisition module is used to record sensor data with a risk of false detection if the ultrasonic data has a risk of false detection and the driving parameters of the preset vehicle meet the preset risk conditions.

10. A terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the parking risk data acquisition method as described in any one of claims 1 to 8.

Images