A sensor fusion positioning detection calibration method based on an unmanned sweeper

Abstract

The application relates to the field of unmanned positioning technology, and particularly discloses a sensor fusion positioning detection and calibration method based on an unmanned sweeper vehicle. The method triggers calibration after obtaining a positioning result, compares the positioning result with a position based on environment observation, judges whether the positioning result is consistent with the position based on environment observation, evaluates position deviation and road condition interference degree if the positioning result is inconsistent with the position based on environment observation, calculates a correlation ratio between positioning misalignment and road condition interference, determines that the abnormality is caused by road condition interference if the correlation ratio is within a reasonable range, calls similar historical correction data to correct the position, recalculates the correlation ratio and confirms whether the correlation ratio is caused by sensor failure if the correlation ratio is beyond the reasonable range, and gives a warning. The application can distinguish the source of positioning abnormality to a certain extent, reduces misjudgment, improves the accuracy and robustness of positioning calibration by combining road condition modeling and historical data correction, and is suitable for high-precision positioning requirements of the unmanned sweeper vehicle in a complex environment.

Description

Technical Field

[0001] This invention relates to the field of unmanned driving positioning technology, and specifically to a sensor fusion positioning detection and calibration method based on an unmanned sweeping vehicle. Background Technology

[0002] With the rapid development of autonomous driving technology, unmanned sweeping vehicles have been widely used in urban sanitation and park cleaning. Accurate and stable positioning capabilities are the core prerequisites for unmanned sweeping vehicles to achieve autonomous navigation and safe operation. Currently, mainstream positioning methods mostly employ multi-sensor fusion technology, such as combining wheel encoders with inertial measurement units (IMUs) to achieve continuous positioning through trajectory extrapolation. However, this type of method is susceptible to factors such as wheel slippage, uneven road surfaces, and sensor drift during long-term operation, leading to the accumulation of positioning errors and consequently affecting the sweeping vehicle's operational accuracy and safety.

[0003] To suppress accumulated errors, existing technologies typically incorporate environmental observation and positioning methods, such as lidar and visual sensors, to achieve positioning correction by matching with environmental maps. However, existing methods still have the following shortcomings in practical applications: First, the positioning calibration mechanism lacks the ability to distinguish the source of anomalies, making it difficult to determine whether positioning inaccuracies are caused by road interference or sensor malfunctions, leading to frequent misjudgments.

[0004] Secondly, the assessment of road condition disturbances relies on empirical thresholds and lacks dynamic modeling and quantitative analysis methods.

[0005] Third, there is a lack of an adaptive correction mechanism based on historical data after the positioning is inaccurate. Summary of the Invention

[0006] To address the above problems, this invention proposes a sensor fusion positioning detection and calibration method based on an unmanned sweeping vehicle. The specific technical solution is as follows: A sensor fusion positioning detection and calibration method based on an unmanned sweeping vehicle includes the following steps: Step S1: After the positioning system obtains the positioning result of the sweeping vehicle at the current moment, it triggers positioning calibration.

[0007] Step S2: Obtain the current position of the sweeper based on environmental observation and compare it with the positioning result in step S1. If they match, send the position and continue the positioning calibration at the next moment. If they do not match, pause the position sending and recalibrate. If the position still fails to meet the criteria, trigger a positioning anomaly warning and proceed to step S3. Otherwise, send the corresponding position.

[0008] Step S3: Assess the current position deviation and evaluate the degree of road interference based on speed data and road surface data. Further calculate the correlation ratio between positioning inaccuracy and road interference. If the correlation ratio is within a reasonable range, the anomaly is caused by road interference and the process jumps to step S4. Otherwise, the anomaly is caused by sensor failure and the process jumps to step S5.

[0009] Step S4: Retrieve location correction data from a previous time that is the same or similar to the current time's road conditions, correct the current time's location accordingly, and send it.

[0010] Step S5: Recalculate the correlation ratio. If it still exceeds the reasonable range, confirm the fault and issue a prompt. Otherwise, return to step S4 to correct the location and send the data.

[0011] Compared with existing technologies, the sensor fusion positioning detection and calibration method based on unmanned sweeping vehicles described in this invention has the following beneficial effects: 1. By introducing a correlation ratio calculation mechanism between positioning inaccuracy and road condition interference, this invention can, to a certain extent, distinguish whether positioning anomalies are caused by external road condition interference or internal sensor faults, thereby reducing misjudgments and improving the diagnostic accuracy of positioning anomalies; at the same time, setting a reasonable range for dynamic correlation ratios further improves the robustness and scientific nature of judging the source of anomalies.

[0012] 2. This invention constructs a road condition interference assessment model based on speed data and road surface data, overcoming the limitations of traditional methods that rely solely on empirical thresholds for judgment, and achieving dynamic and quantitative assessment of road condition interference, thereby improving the accuracy and adaptability of the assessment.

[0013] 3. When positioning inaccuracies are caused by road condition interference, this invention can correct the current position based on historical correction data under the same or similar road conditions, thereby establishing a correction mechanism that helps improve the speed and accuracy of positioning recovery. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0016] Figure 2 This is a structural block diagram of the present invention.

[0017] Figure 3 This is a logic block diagram for evaluating the degree of road condition interference according to the present invention. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a sensor fusion positioning detection and calibration method based on an unmanned sweeping vehicle proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0020] The following describes in detail, with reference to the accompanying drawings, a specific scheme of the sensor fusion positioning detection and calibration method based on an unmanned sweeping vehicle provided by the present invention.

[0021] This invention achieves calibration by comparing the real-time positioning results acquired by the sweeper's positioning system with the location based on environmental observations. It introduces a correlation ratio calculation mechanism between positioning inaccuracies and road condition interference, combined with a determined reasonable range of dynamic correlation ratios, to distinguish whether positioning anomalies are caused by external road condition interference or internal sensor malfunctions. If the anomaly is caused by road condition interference, similar historical correction data is retrieved for position correction; if it is caused by sensor malfunction, an early warning is issued. This method can, to a certain extent, distinguish the source of positioning anomalies, reduce misjudgments, and improve the accuracy and robustness of positioning calibration.

[0022] Please see Figure 1 and Figure 2 As shown, the present invention provides a sensor fusion positioning detection and calibration method based on an unmanned sweeping vehicle, which includes the following steps S1 to S5.

[0023] In one embodiment of the present invention, to ensure the accuracy and reliability of the unmanned sweeper's positioning, a trigger-based positioning calibration process needs to be designed. First, the positioning system calculates the sweeper's current pose in real time, and after obtaining the positioning result, triggers subsequent calibration judgments. Considering that the positioning system may accumulate errors during long-term operation or in complex environments, environmental observation information needs to be introduced for verification, laying the foundation for subsequent comparison analysis and anomaly handling.

[0024] Step S1, obtaining positioning results, specifically includes the following steps: After the positioning system obtains the positioning result of the sweeper at the current moment, it triggers positioning calibration. The positioning system consists of a wheel encoder and an inertial measurement unit. The method for obtaining the positioning result of the sweeper at the current moment includes: fusing the relative displacement data measured by the wheel encoder with the angular velocity and acceleration data measured by the inertial measurement unit to calculate the sweeper's driving trajectory.

[0025] Based on the driving trajectory and the previous position of the sweeper, the current positioning result of the sweeper is output, which includes two-dimensional coordinates and orientation.

[0026] Once the current location result is obtained, the location result can be verified, that is, compared with the location observed in the environment to evaluate its accuracy.

[0027] Considering that positioning systems (such as wheel encoders and IMUs) may have accumulated errors or instantaneous errors in their calculations, while location information obtained based on environmental observations (such as matching lidar point clouds with global maps) is ideally more absolute and accurate, comparing the two is a direct means of identifying positioning inaccuracies. By comparing their consistency, the reliability of the positioning result is initially assessed; if inconsistent, a recalibration and anomaly warning process is initiated. This aims to promptly identify problems and perform initial anomaly source filtering, preventing the use of erroneous location information in subsequent control and planning.

[0028] Step S2, Real-time Positioning Calibration, specifically includes the following steps: S2-1, Obtaining the current position of the sweeper based on environmental observation, specifically including: S2-1-1, Collecting environmental data of the environment around the sweeper at the current moment through environmental perception sensors, identifying environmental features based on environmental data and selecting reference environmental features, wherein the method for selecting reference environmental features is: S2-1-1-1, Obtaining the attributes of each environmental feature, filtering out all static environmental features, wherein the attributes include static and dynamic.

[0029] S2-1-1-2. Obtain the point cloud clarity of each static environmental feature and the relative distance between each static environmental feature and the sweeper.

[0030] S2-1-1-3. Sort each static environmental feature in order of point cloud sharpness from high to low to obtain the first sequence, and sort them in order of relative distance from near to far to obtain the second sequence.

[0031] S2-1-1-4. For each static environmental feature, sum its index in the first sequence and its index in the second sequence to obtain the corresponding sum value.

[0032] S2-1-1-5. Sort each static environmental feature according to the cumulative value in ascending order to form an environmental feature recommendation queue.

[0033] S2-1-1-6. Select the static environmental feature that ranks first in the environmental feature recommendation queue as the reference environmental feature.

[0034] It should be noted that environmental perception sensors include, but are not limited to, LiDAR, visual cameras, etc.

[0035] S2-1-2. Based on environmental data, obtain the relative position information between the current sweeper and the reference environmental features, whereby the relative position information includes relative orientation and relative distance.

[0036] S2-1-3. Based on the reference environmental features and the relative position information between the sweeper and the reference environmental features, match the high-precision global map pre-stored in the database to determine the position of the sweeper in the global map, and record it as the position of the sweeper based on environmental observation at the current moment.

[0037] S2-2. Compare the location based on environmental observation with the location result in S1 to determine whether they are consistent. The determination method includes: S2-2-1. Calculate the distance between the two-dimensional coordinates in the location result and the two-dimensional coordinates in the location based on environmental observation using the distance formula between two points, and use it as the coordinate offset distance.

[0038] S2-2-2. Obtain the angle between the orientation in the positioning result and the orientation in the location based on environmental observation, and use it as the orientation offset angle.

[0039] S2-2-3. Based on the positioning accuracy requirements pre-stored in the database, obtain the allowable range of coordinate offset distance and orientation offset angle.

[0040] S2-2-4. If either the coordinate offset distance or the orientation offset angle at the current moment exceeds the corresponding allowable range, the position based on environmental observation is determined to be inconsistent with the positioning result; otherwise, it is determined to be consistent.

[0041] S2-3. If the two are consistent, the positioning result is sent as the final position, and the positioning calibration process for the next moment is initiated.

[0042] S2-4. If they are inconsistent, pause the location transmission and perform a recalibration. If they are consistent during the recalibration, send the location at the time of the recalibration.

[0043] If the recalibration is still inconsistent, a positioning anomaly is determined, an early warning is triggered, and the process proceeds to step S3.

[0044] It should be noted that the recalibration process is the same as the current positioning calibration process. Specifically, the positioning system reacquires the current positioning result of the sweeper and compares the reacquired positioning result with the position based on environmental observation to determine whether the two are consistent.

[0045] If, after recalibration, the location based on environmental observations is still inconsistent with the positioning results, it is determined that the positioning is inaccurate at the current time. In this case, it is necessary to conduct an in-depth analysis of the root cause of the inaccuracy, whether it is due to road condition interference or sensor malfunction.

[0046] Once a positioning error is confirmed, it is crucial to distinguish whether the cause is external environmental interference or internal sensor malfunction in order to take appropriate action. External environmental factors, such as slippery or bumpy roads, can affect the slippage of the wheel encoder or the measurement accuracy of the IMU, thus introducing positioning errors; while internal sensor malfunctions may cause data jumps or interruptions. This invention quantifies the severity of the error by assessing the position deviation at the current moment, while simultaneously combining speed and road surface data to assess the degree of road condition interference, and then calculates the correlation ratio between the two. If the correlation ratio indicates a high correlation between the position deviation and road condition interference, it can be inferred that the error is caused by road conditions; otherwise, it is more likely to point to sensor malfunction. This differentiation mechanism provides a decision-making basis for achieving accurate correction or fault handling.

[0047] Step S3, position misalignment diagnosis, specifically includes the following steps: S3-1, assess the position deviation at the current moment. The specific method is to calculate the ratio between the coordinate offset distance at the current moment and its corresponding upper limit of the allowable range, as the coordinate relative offset ratio, and calculate the ratio between the orientation offset angle and its corresponding upper limit of the allowable range, as the orientation relative offset ratio.

[0048] Based on the relative offset ratio of coordinates and the relative offset ratio of orientation, and combined with the preset weights of coordinate offset and orientation offset, the position deviation at the current moment is obtained through linear weighted fusion analysis.

[0049] It should be noted that location usually has a more direct impact on the safety of unmanned sweeping vehicles than orientation, therefore coordinate offset weighting is often higher. In a typical example that prioritizes location safety, the coordinate offset weight is 0.7, and the orientation offset weight is 0.3.

[0050] S3-2. Assess the degree of road condition interference based on speed and road surface data, where, see... Figure 3 As shown, the method for assessing the degree of road condition interference is as follows: S3-2-1, obtain the speed data of the sweeper and the road surface data of the road section it has passed within the time window formed by the current time and the previous time. The speed data is the average vehicle speed, and the road surface data includes the road surface humidity and the road surface smoothness.

[0051] S3-2-2. Through historical experimental testing, the critical values ​​for road surface moisture and road surface smoothness that interfere with the sweeper's positioning system were determined, and denoted as follows: .

[0052] S3-2-3. Input the average vehicle speed, road surface moisture, and road surface smoothness into the road condition interference analysis model to calculate the road condition interference level at the current moment. The road condition interference analysis model is as follows: .

[0053] In the formula, Indicates the degree of road condition interference. These represent road surface moisture, road surface smoothness, and average vehicle speed, respectively. These represent the preset weights for road surface moisture and road surface smoothness, respectively. This represents the influence factor corresponding to the preset unit vehicle speed.

[0054] It should be noted that road surface moisture can be detected by non-contact optical sensors mounted on the sweeper, such as laser road surface condition sensors or spectral analyzers; road surface smoothness can be detected by three-dimensional imaging and vision technology, such as scanning the road surface outline with a line laser and using a high-speed camera to capture the deformation of the laser line, generating a three-dimensional digital model of the road surface based on triangulation, so as to realize the detection of road surface smoothness.

[0055] S3-3. Calculate the correlation ratio between positioning inaccuracy and road condition interference. The specific method is to calculate the ratio between the current position deviation and the degree of road condition interference, and use it as the correlation ratio between positioning inaccuracy and road condition interference.

[0056] S3-4. Determine the reasonable range of dynamic correlation ratio. Specific methods include: S3-4-1. Based on the degree of road interference and the corresponding deviation of the sweeper's positioning position under various historical road conditions test scenarios, statistically analyze multiple sets of data on the degree of road interference and the position deviation, and construct a training set.

[0057] S3-4-2. Based on the training set, machine learning algorithms are used to analyze the mapping relationship between the degree of road condition interference and the degree of position deviation in order to build a position deviation prediction model.

[0058] S3-4-3. Input the current level of road condition interference into the position deviation prediction model to obtain the predicted position deviation.

[0059] S3-4-4 Calculate the ratio between the predicted position deviation at the current moment and the actual road condition interference level. Use this ratio as the upper limit of the correlation ratio between the current positioning inaccuracy and road condition interference. Determine the reasonable range of the correlation ratio at the current moment based on the upper limit of the correlation ratio.

[0060] S3-5. Compare the correlation ratio with the corresponding reasonable range. If the correlation ratio is within the reasonable range, the anomaly is caused by road condition interference and the process jumps to S4. Otherwise, the anomaly is caused by sensor failure and the process jumps to S5.

[0061] It should be noted that the method for determining whether the correlation ratio at the current moment exceeds the reasonable range is as follows: if the correlation ratio is greater than the upper limit of the correlation ratio, it is determined to be beyond the reasonable range; otherwise, it is determined to be within the reasonable range.

[0062] If the correlation analysis determines that the current location anomaly is mainly caused by road condition interference, then the process can proceed to the location correction process using historical experience data.

[0063] When positioning inaccuracies are determined to be caused by road condition interference, it's important to consider that the error patterns of positioning systems often exhibit repetitiveness and regularity under similar road conditions. Therefore, successful correction experiences (i.e., position correction data) accumulated by the sweeper in historical tasks, especially when encountering similar road conditions in the current task, can be used to correct the current inaccurate position. This method can quickly and effectively compensate for positioning deviations caused by road condition interference without requiring immediate stopping or triggering more complex fault handling, ensuring the continuity and efficiency of the sweeping task.

[0064] Step S4: Dynamic position correction. Retrieve position correction data from previous moments that are the same or similar to the current moment's road conditions, correct the current moment's position accordingly, and send it. Specifically, this includes the following steps: S4-1: Obtain the road condition data and corresponding position correction data from previous moments before the current moment during the sweeper's execution of this task. The road condition data includes speed data and road surface data, and the position correction data includes two-dimensional coordinate correction and orientation correction.

[0065] It should be noted that both the two-dimensional coordinate correction and orientation correction are signed.

[0066] S4-2. Iterate through the traffic data of each previous time step to find the previous time step with the same traffic conditions as the current time step.

[0067] S4-3. If the query is successful, the position correction data corresponding to the previous time with the same road conditions as the current time will be used as the position correction data for the current time, and the position of the sweeper will be corrected accordingly.

[0068] S4-4. If the query fails, calculate the road condition similarity between the previous time and the current time based on the road condition data deviation. The method for calculating the road condition similarity includes: comparing the road condition data of the previous time and the current time to obtain the deviation of each sub-item in the road condition data; calculating the ratio between the deviation of each sub-item and the measured value of the corresponding road condition data sub-item at the current time as the deviation ratio of each sub-item; accumulating the deviation ratios of each sub-item; and taking the reciprocal of the accumulated value as the road condition similarity.

[0069] The earlier timestamps with road condition similarity greater than a preset threshold are selected as the earlier timestamps of similar road conditions.

[0070] S4-5. Based on the position correction data corresponding to similar road conditions at previous times, correct the position of the sweeper at the current time.

[0071] It should be noted that when multiple previous times with the same or similar road conditions are found, the average of the location correction data corresponding to each previous time is taken as the final location correction data.

[0072] It should be noted that if no previous time with the same or similar road conditions as the current time can be found during this task, the search scope will be expanded to find the location correction data corresponding to the same or similar road conditions in the historical task data.

[0073] To prevent misjudgments caused by momentary interference or computational fluctuations, this embodiment of the invention includes a fault confirmation step. When S3 initially determines that the anomaly may be caused by a sensor malfunction, or when a final confirmation of the fault is required at a certain node in the S4 correction process, the process proceeds to S5. By recalculating the correlation ratio, if the result still exceeds a reasonable range, a sensor malfunction can be confirmed with high confidence, and a timely alert can be issued for shutdown inspection or maintenance. If the correlation ratio returns to a reasonable range after recalculation, it indicates that the previous judgment may have been interfered with, and the process can return to S4 to attempt position correction using historical data. This step provides a final guarantee for the robustness and security of the system.

[0074] S5. Fault / Abnormality Re-verification: Recalculate the correlation ratio. If it still exceeds the reasonable range, confirm the fault and provide a prompt. Otherwise, return to S4 to perform location correction and transmission.

[0075] It should be noted that the process of recalculating the correlation ratio is the same as the process of calculating the correlation ratio between the current positioning inaccuracy and road condition interference. Specifically, the positioning system reacquires the current positioning result of the sweeper, and recalculates the current position deviation based on the reacquired positioning result, thereby recalculating the correlation ratio between the current positioning inaccuracy and road condition interference.

[0076] It should also be noted that if the number of recalculations exceeds the upper limit and the system still cannot be stably corrected, it will be forcibly identified as a sensor malfunction and an alarm will be issued, thereby breaking the potential infinite loop and ensuring the robustness and security of the system.

[0077] In this embodiment, the present invention also introduces an environmental feature optimization mechanism, which comprehensively selects reference environmental features based on multiple factors such as point cloud clarity and distance, thereby improving the reliability and matching efficiency of environmental observation and positioning.

[0078] In this embodiment, the present invention further verifies fault judgment through a recalculation mechanism, reduces false alarm rate, and enhances system stability and practicality.

[0079] In summary, this invention triggers calibration after acquiring the positioning result. It compares the result with the location based on environmental observation to determine consistency. If inconsistent, it assesses the position deviation and the degree of road interference, calculating the correlation ratio between positioning inaccuracy and road interference. If the correlation ratio is within a reasonable range, the anomaly is determined to be caused by road interference, and similar historical correction data is retrieved for position correction. If the correlation ratio exceeds a reasonable range, the correlation ratio is recalculated, and it is confirmed whether it is a sensor malfunction, issuing an early warning. This invention, through its correlation ratio mechanism, can distinguish the source of positioning anomalies to a certain extent, reducing misjudgments. Combined with road condition modeling and historical data correction, it improves the accuracy and robustness of positioning calibration, making it suitable for the high-precision positioning needs of unmanned sweeping vehicles in complex environments.

[0080] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0081] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0082] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0083] Those skilled in the art will recognize that the modules 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.

[0084] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0085] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0086] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A sensor fusion positioning, detection, and calibration method based on an unmanned sweeping vehicle, characterized in that, Includes the following steps: S1. After the positioning system obtains the current positioning result of the sweeper, it triggers positioning calibration; S2. Obtain the current position of the sweeper based on environmental observation and compare it with the positioning result in S1. If they match, send the position and continue the positioning calibration at the next moment. If they do not match, pause the position sending and recalibrate. If the position still fails to meet the criteria, trigger a positioning anomaly warning and proceed to S3. Otherwise, send the corresponding position. S3. Assess the current position deviation and evaluate the degree of road interference based on speed data and road surface data. Calculate the correlation ratio between positioning inaccuracy and road interference. If the correlation ratio is within a reasonable range, the anomaly is caused by road interference and jumps to S4. Otherwise, the anomaly is caused by sensor failure and jumps to S5. S4. Retrieve location correction data from a previous time that is the same or similar to the current time's traffic conditions, correct the current time's location accordingly, and send it. S5. If the correlation ratio is still outside the reasonable range after recalculation, the fault is confirmed and a prompt is made; otherwise, return to S4 to perform position correction and transmission. The method for calculating the correlation ratio and determining a reasonable range for the correlation ratio is as follows: Calculate the ratio between the current position deviation and the degree of road condition interference, and use it as the correlation ratio between positioning inaccuracy and road condition interference; Based on the degree of road condition interference and the corresponding deviation of the sweeper's positioning position under various historical road condition test scenarios, a training set was constructed by statistically analyzing multiple sets of data on the degree of road condition interference and the position deviation. Based on the training set, machine learning algorithms are used to analyze the mapping relationship between the degree of road condition interference and the degree of position deviation, so as to build a position deviation prediction model. Input the current level of road condition interference into the position deviation prediction model to obtain the predicted position deviation. Calculate the ratio between the predicted position deviation at the current moment and the actual road condition interference level, and use it as the upper limit of the correlation ratio between the current positioning inaccuracy and road condition interference. Determine the reasonable range of the correlation ratio at the current moment based on the upper limit of the correlation ratio. The specific analysis process of step S4 is as follows: Acquire road condition data and corresponding position correction data for each previous moment before the current moment during the execution of this task by the sweeper. The road condition data includes speed data and road surface data, and the position correction data includes two-dimensional coordinate correction and orientation correction. Iterate through the traffic data from previous moments to find the previous moments with the same traffic conditions as the current moment; If the query is successful, the location correction data corresponding to the previous time with the same road conditions as the current time will be used as the location correction data for the current time, and the position of the sweeper will be corrected accordingly. If the query fails, the similarity between the road conditions at previous times and the current time is calculated based on the deviation of the road condition data, and the previous times with a road condition similarity greater than a preset threshold are selected as the previous times with similar road conditions. The position of the sweeper is corrected at the current moment based on the position correction data corresponding to similar road conditions at the previous time. If multiple previous times with the same or similar road conditions as the current time are found, the average of the location correction data corresponding to each previous time is taken as the final location correction data. If no previous times with the same or similar road conditions as the current time can be found during the current task, the search scope is expanded to search for location correction data corresponding to historical times with the same or similar road conditions as the current time from historical task data.

2. The sensor fusion positioning, detection, and calibration method based on an unmanned sweeping vehicle according to claim 1, characterized in that: The positioning system consists of a wheel encoder and an inertial measurement unit.

3. The sensor fusion positioning, detection, and calibration method based on an unmanned sweeping vehicle according to claim 1, characterized in that: The method for obtaining the current location result of the sweeper includes: The relative displacement data measured by the wheel encoder is fused with the angular velocity and acceleration data measured by the inertial measurement unit to calculate the sweeper's trajectory. Based on the driving trajectory and the previous position of the sweeper, the current positioning result of the sweeper is output.

4. The sensor fusion positioning, detection, and calibration method based on an unmanned sweeping vehicle according to claim 1, characterized in that: The method for obtaining the current position of the sweeper based on environmental observation includes: The sweeper collects environmental data of the surrounding environment at the current moment through environmental perception sensors, identifies environmental features based on the environmental data, and selects reference environmental features from them. Based on environmental data, obtain the relative position information between the current sweeper and the reference environmental features, where the relative position information includes relative orientation and relative distance; Based on the reference environmental features and the relative position information between the sweeper and the reference environmental features, the sweeper is matched with the high-precision global map pre-stored in the database to determine the position of the sweeper in the global map, and recorded as the position of the sweeper based on environmental observation at the current moment.

5. The sensor fusion positioning and detection calibration method based on an unmanned sweeping vehicle according to claim 4, characterized in that: The method for selecting the characteristics of the reference environment is as follows: Obtain the attributes of each environmental feature and filter out all static environmental features, where the attributes include both static and dynamic attributes; Obtain the point cloud clarity of each static environmental feature and the relative distance between each static environmental feature and the sweeper; The static environmental features are sorted in order of point cloud sharpness from high to low to obtain the first sequence, and then sorted in order of relative distance from near to far to obtain the second sequence. For each static environmental feature, its index in the first sequence and its index in the second sequence are summed to obtain the corresponding sum value; The static environmental features are sorted in ascending order of their accumulated values ​​to form an environmental feature recommendation queue. The static environmental feature that ranks first in the environmental feature recommendation queue is selected as the reference environmental feature.

6. The sensor fusion positioning and detection calibration method based on an unmanned sweeping vehicle according to claim 1, characterized in that: The method for determining whether the location based on environmental observation is consistent with the positioning result is as follows: The distance between the two-dimensional coordinates in the positioning result and the two-dimensional coordinates in the location based on environmental observation are calculated using the distance formula between two points, and this distance is used as the coordinate offset distance. The angle between the orientation in the positioning result and the orientation in the location based on environmental observation is obtained as the orientation offset angle. Based on the positioning accuracy requirements pre-stored in the database, obtain the allowable range of coordinate offset distance and orientation offset angle; If either the coordinate offset distance or the orientation offset angle at the current moment exceeds the corresponding allowable range, the position based on environmental observation is determined to be inconsistent with the positioning result; otherwise, it is determined to be consistent.

7. The sensor fusion positioning and detection calibration method based on an unmanned sweeping vehicle according to claim 6, characterized in that: The method for assessing positional deviation is as follows: Calculate the ratio between the current coordinate offset distance and its corresponding upper limit of the allowable range, as the relative coordinate offset ratio; and calculate the ratio between the orientation offset angle and its corresponding upper limit of the allowable range, as the relative orientation offset ratio. Based on the relative offset ratio of coordinates and the relative offset ratio of orientation, and combined with the preset weights of coordinate offset and orientation offset, the position deviation at the current moment is obtained through linear weighted fusion analysis.

8. The sensor fusion positioning and detection calibration method based on an unmanned sweeping vehicle according to claim 1, characterized in that: The method for assessing the degree of road condition interference is as follows: The system acquires the speed data of the sweeper and the road surface data of the road segment it has traveled within the time window formed by the current moment and the previous moment. The speed data is the average vehicle speed, and the road surface data includes the road surface moisture and road surface smoothness. Through historical experimental testing, the critical values ​​for road surface moisture and road surface smoothness that interfere with the sweeper's positioning system were determined, and denoted as follows: ; Average vehicle speed, road surface moisture, and road surface smoothness are input into the road condition disturbance level analysis model to calculate the road condition disturbance level at the current time. The road condition disturbance level analysis model is as follows: ； In the formula, Indicates the degree of road condition interference. These represent road surface moisture, road surface smoothness, and average vehicle speed, respectively. These represent the preset weights for road surface moisture and road surface smoothness, respectively. This represents the influence factor corresponding to the preset unit vehicle speed.

Images