Positioning calibration method and system

Abstract

A positioning and calibration method and system. The method comprises the following steps: providing a preset travel path; based on the preset travel path, obtaining machine calibration data from multiple anchor points of a target object in a preset acquisition posture by a data acquisition device equipped with a first short-distance communication device, and obtaining manual calibration data from multiple anchor points of the target object by a real person equipped with a second short-distance communication device; comparing the machine calibration data with the manual calibration data to obtain a feature comparison result; adjusting the acquisition posture of the data acquisition device based on the feature comparison result; based on the preset travel path and the adjusted acquisition posture, obtaining simulated calibration data from multiple anchor points of the target object by the data acquisition device; determining the positioning accuracy and calibration parameters of the positioning algorithm through the simulated calibration data. The method enables the data acquisition device to obtain simulated calibration data, and maximizes the collection efficiency and collection accuracy of the calibration data.

Description

Technical Field

[0001] This application relates to the field of short-range communication technology, and in particular to a calibration method and system for a positioning algorithm. Background Technology

[0002] Short-range wireless communication is a communication technology that uses radio waves to transmit information over short distances. Bluetooth Low Energy (BLE) and Ultra Wide Band (UWB) technologies are important components of short-range wireless communication, widely used in the Internet of Things (IoT) short-range positioning field due to their advantages of low power consumption or high positioning accuracy. However, with technological advancements, the requirements for positioning accuracy are constantly changing. Positioning algorithms based on specific application scenarios may struggle to adapt to new business needs, necessitating the recalibration of key parameters in these algorithms.

[0003] In the entire process of training and calibrating positioning algorithm models, the most common method currently is to manually measure data and calibrate the algorithm parameters based on a pre-defined algorithm model. For example, Bluetooth technology often uses three-point or four-point positioning as the basis for positioning algorithm models. However, at the same location, the Bluetooth signals of different brands, models, or types of terminals may vary. If the distance-to-signal-strength (RSSI) model remains unchanged, the positioning results will be inaccurate. Therefore, to ensure consistent positioning algorithm results for each terminal, the algorithm parameters used should differ for terminals with varying RSSI signals. This requires manual on-site measurements of different brands, models, or types of terminals, combined with prior experience, to calibrate the algorithm parameters. Finally, repeated testing is necessary before deployment. However, the product update cycle for terminals is very short, and the repetitive manual work involved in the calibration process reduces calibration efficiency and accuracy.

[0004] Regarding existing technologies for positioning calibration in short-range wireless communication, patent number CN202410815299.X discloses a method, apparatus, device, and storage medium for UWB-assisted BLE positioning calibration. This method uses UWB technology to accurately locate a ranging module, thereby obtaining the relative position between the ranging module and a UWB anchor point. Since the positions of the UWB anchor point and the Bluetooth anchor point are relatively fixed, the distance between the ranging module and the Bluetooth anchor point and the positioning area are obtained. Signal calibration is then performed on the distance and positioning area to obtain the corresponding signal strength calibration value. This signal calibration value is input into the Bluetooth positioning system to obtain the Bluetooth positioning system's ranging value and positioning area. Finally, the UWB ranging value is compared with the Bluetooth ranging value, and the UWB positioning area is compared with the Bluetooth positioning range to calibrate the Bluetooth positioning result.

[0005] In the prior art, patent number CN202311557931.7 discloses a PEPS Bluetooth positioning self-calibration system and method. In this system, after a vehicle enters a new environment, the master Bluetooth transmitter sends a Bluetooth signal to the slave Bluetooth transmitter. The influence of current environmental factors can be calculated from the signal strength value received from the slave Bluetooth transmitter. A mobile terminal is placed at a preset position inside the vehicle, and its gain value is obtained. Finally, the influence of environmental factors is combined with the gain value of the mobile terminal to correct the strength value of the Bluetooth signal received from the mobile terminal.

[0006] In the prior art, patent number CN202110497439.X discloses a signal positioning calibration method and system for a digital key system. The calibration method includes the following steps: obtaining the antenna parameters of all unlocking terminals to be calibrated, and randomly selecting one of the unlocking terminals as a reference object, with the remaining unlocking terminals as comparison objects; calculating the standard signal strength difference of each comparison object relative to the reference object based on the antenna parameters; obtaining the reference calibration parameters corresponding to the reference object based on measured data; correcting the reference calibration parameters according to each standard signal strength difference to obtain corrected calibration parameters adapted to each comparison object; in the positioning terminal, using the reference calibration parameters to perform signal positioning calibration on the reference object, and using each corrected calibration parameter to perform signal positioning calibration on the comparison object adapted to it.

[0007] In the prior art, patent number CN202110451103.X discloses a positioning calibration system, method, and storage medium, including a mobile platform and a back-end processing device; the mobile platform includes a mobile device and a data processing device; the calibration module includes at least two different types of positioning sensors, which are used to acquire the original calibration information of the mobile platform in real time; the data processing device is used to acquire the wireless positioning information of the mobile platform in real time and acquire the original calibration information; both the wireless positioning information and the original calibration information include the location coordinates of the mobile platform and the corresponding timestamp; the back-end processing device includes a storage module and a positioning trajectory demonstration module; the storage module is used to store the original calibration information and the wireless positioning information; the positioning trajectory demonstration module is used to generate a dynamic trajectory of the mobile platform on a virtual map based on the original calibration information and the wireless positioning information.

[0008] There is still no effective solution to the problem of low positioning and calibration efficiency in short-range wireless communication in related technologies. Summary of the Invention

[0009] Therefore, it is necessary to provide a positioning calibration method and system that can improve the efficiency of positioning calibration in response to the above-mentioned technical problems.

[0010] Firstly, this application provides a positioning and calibration method, which includes the following steps:

[0011] S10: Provides a preset travel path;

[0012] S20: Based on the preset travel path, a data acquisition device equipped with a first short-range communication device acquires machine calibration data from multiple anchor points of the target object in a preset acquisition pose. A human being equipped with a second short-range communication device acquires manual calibration data from multiple anchor points of the target object. Each anchor point includes at least one third short-range communication device, thereby communicating with the first short-range communication device of the data acquisition device and the second short-range communication device of the human being.

[0013] S30: Compare the machine calibration data and the manual calibration data to obtain the feature comparison result;

[0014] S40: Adjust the acquisition pose of the data acquisition device based on the feature comparison results;

[0015] S50: Based on the preset travel path and the adjusted acquisition pose, obtain simulated calibration data for multiple anchor points of the target object through a data acquisition device;

[0016] S60: Determine the positioning accuracy and calibration parameters of the positioning algorithm using the simulated calibration data.

[0017] In one embodiment, the machine calibration data includes: the sampling position of the data acquisition device at the sampling time in the preset travel path; and

[0018] Based on the acquisition location, the data acquisition device receives the communication signals of the plurality of anchor points in the acquisition pose;

[0019] The manually calibrated data includes:

[0020] The sampling position of the real person at the sampling time in the preset travel path; and

[0021] Based on the acquisition location, the real person receives communication signals from the multiple anchor points.

[0022] In one embodiment, the preset travel path includes at least one of a radial path centered on the target object, a circumferential path, and a combination of the radial path and the circumferential path.

[0023] In one embodiment, the acquisition pose includes the height, horizontal angle, and vertical angle of the first short-range communication device of the data acquisition equipment.

[0024] In one embodiment, step S30 includes: extracting data features from the machine calibration data and the manual calibration data respectively to obtain machine data features and manual data features; comparing the machine data features and the manual data features to determine feature difference parameters.

[0025] In one embodiment, the short-range communication device includes Bluetooth Low Energy (BLE) and / or UWB, and the data features include at least one of the following features:

[0026] Signal strength distribution characteristics of multiple anchor points at each acquisition location;

[0027] The signal strength variation trend characteristics of each anchor point as the communication distance changes;

[0028] The signal time-of-flight distribution characteristics of the multiple anchor points at each acquisition location;

[0029] The characteristics of the signal flight time variation trend of each anchor point as the communication distance changes;

[0030] The trend characteristics of the signal flight time variation at each of the anchor points as the acquisition location changes;

[0031] The distribution characteristics of the measured distances of the multiple anchor points at each of the said acquisition locations;

[0032] The trend characteristics of the measured distance change of each anchor point with the communication distance; and

[0033] The trend characteristics of the measurement distance change of each anchor point as the acquisition location changes.

[0034] In one embodiment, step S50 includes:

[0035] The decision tree model is trained and validated based on the simulated calibration data to obtain the positioning algorithm with a preset positioning accuracy and the calibration parameters.

[0036] In one embodiment, step S60 is followed by:

[0037] S70: Provides a preset verification path;

[0038] S80: Based on a preset verification path, obtain simulated positioning information and manual positioning information.

[0039] S90: Based on the error of the simulated positioning information and the manual positioning information, determine the accuracy of the positioning algorithm and the calibration parameters.

[0040] In one embodiment, step S80 further includes:

[0041] S81: Based on the preset verification path and the adjusted acquisition pose, the simulated positioning information of the target object is obtained from the multiple anchor points of the target object through the data acquisition device;

[0042] S82: Based on the preset verification path, obtain the artificial positioning information through the multiple anchor points of the target object by the real person.

[0043] Secondly, this application also provides a positioning and calibration system, including a robot, a smart terminal, and a computing unit. The computing unit is used to perform calculations involved in the positioning and calibration method as described in any of the embodiments of the first aspect above. The robot moves according to the preset travel path. The smart terminal is disposed on the robot and acquires the machine calibration data in the preset acquisition pose.

[0044] The aforementioned positioning calibration method and system provide a preset travel path, enabling the data acquisition device to collect machine calibration data along the preset path, and for a human to collect manual calibration data along the preset path. Based on the feature comparison results of the machine calibration data and the manual calibration data, the acquisition pose of the data acquisition device is adjusted. Then, simulated calibration data is obtained by moving the data acquisition device along the preset travel path again. The positioning accuracy and calibration accuracy of the positioning algorithm are determined based on the simulated calibration data. Compared with related technologies that rely on manual calibration data collected by humans for positioning algorithm calibration, the positioning calibration method of this application corrects the acquisition pose of the data acquisition device based on the feature comparison results of machine calibration data and manual calibration data. This allows the data acquisition device to simulate manual collection to obtain simulated marker data, achieving automated collection of calibration data and improving the efficiency and accuracy of positioning calibration. Attached Figure Description

[0045] Figure 1 This is a diagram illustrating the application environment of the positioning and calibration method in one embodiment.

[0046] Figure 2 This is a flowchart illustrating a positioning and calibration method in one embodiment;

[0047] Figure 3 This is a schematic diagram of a loop path in one embodiment;

[0048] Figure 4 This is a schematic diagram of the central radiation path in another embodiment;

[0049] Figure 5 This is a flowchart illustrating the verification process of the localization algorithm and calibration parameters in one embodiment;

[0050] Figure 6 This is a flowchart illustrating the process of obtaining simulated location information and manual location information based on a preset verification path in one embodiment.

[0051] Figure 7 This is a schematic diagram of the data acquisition architecture in one embodiment;

[0052] Figure 8 This is a flowchart illustrating the positioning and calibration method in another embodiment;

[0053] Figure 9 This is a schematic diagram of the adaptive algorithm training and parameter fitting process in one embodiment;

[0054] Figure 10 This is a schematic diagram of the functional test range in one embodiment. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0056] The positioning and calibration method provided in this application embodiment can be applied to, for example, Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed in the cloud or on other network servers. Terminal 102 can collect manually calibrated data by being held or worn by a person, or by being transported by a self-moving robot or intelligent vehicle. Server 104 is used to store both machine calibration data and manually calibrated data. The computing power of the terminal or server is used to perform calculations and analyses on the machine calibration data and manually calibrated data to ultimately determine the calibration parameters of the positioning algorithm. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers and smart in-vehicle devices. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted devices. Server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0057] In one embodiment, such as Figure 2 As shown, this application provides a positioning calibration method, which is applied to... Figure 1 Taking the application environment in the example, this positioning and calibration method includes the following steps:

[0058] Step S10: Provide a preset travel path.

[0059] Specifically, the preset travel path is a data acquisition path planned manually and / or by computer equipment based on the data acquisition requirements and site conditions during the positioning and calibration process. The preset travel path can be configured into the data acquisition device in real time via wireless communication or pre-imported.

[0060] In one embodiment, the preset travel path can be a radial path centered on the target object, a circular path centered on the target object, or a combination of a radial path and a circular path centered on the target object.

[0061] Specifically, such as Figure 3As shown, taking a robot equipped with a smart terminal as an example, the robot's preset travel path can be set through a path planning algorithm. By adding a circular route and setting travel parameters, a circular preset travel path can be generated. This path can then be imported into the robot or controlled through real-time navigation.

[0062] In one embodiment, such as Figure 4 To accurately reflect the movement trajectory around the entire vehicle body, a preset travel path composed of a central radial route and parallel routes is provided. The closest and furthest points of this preset travel path relative to the vehicle body can be adapted to the performance of the short-range communication equipment and the corresponding service requirements for the positioning function. For example, when the primary service requirement for testing is vehicle unlocking, the furthest point can be set to 3 meters around the vehicle body; when the primary service requirement is vehicle locking, the furthest point can be set to 5 meters around the vehicle body; and when the primary service requirement is a welcoming function, the furthest point can be set to 8 to 10 meters around the vehicle body.

[0063] In one embodiment, a path planning program for a preset travel path is also provided. Through this program, the user can set the size of the target object, and thereby set the path shape, travel speed, path repetition count, stopping position, number of stops, and stopping time of the preset travel path.

[0064] Step S20: Based on the preset travel path, a data acquisition device equipped with a first short-range communication device acquires machine calibration data from multiple anchor points of the target object in a preset acquisition pose. A human operator equipped with a second short-range communication device acquires manual calibration data from the multiple anchor points of the target object. Each anchor point includes at least one third short-range communication device, thereby communicating with the first short-range communication device of the data acquisition device and the second short-range communication device of the human operator.

[0065] Specifically, short-range communication technology refers to technologies that can achieve wireless data transmission within a range of a few meters to hundreds of meters. Common short-range communication devices include Bluetooth devices, Wi-Fi, ZigBee, infrared data transmission, near-field communication, and ultra-wideband (UWB). In this embodiment, the first, second, and third short-range communication devices are one or more of the corresponding devices for implementing the aforementioned short-range communication technologies. Data acquisition equipment is an automated device used to collect machine calibration data, such as a robot equipped with a short-range communication device or a smart terminal, an automated guided vehicle (AGV), or a combination of a robot and an AGV. Furthermore, data acquisition equipment can also be a combination of a robot or AGV and a smart terminal equipped with a short-range communication device. Smart terminals include, but are not limited to, smartphones, smartwatches, smart bracelets, smart glasses, and Bluetooth headsets. During the movement and start-stop process along a preset path in a preset acquisition posture, the data acquisition device communicates with multiple anchor points of the target object to obtain machine calibration data. Furthermore, the real person communicates with multiple anchor points of the target object while moving and starting / stopping along the same preset travel path by wearing or holding a second short-range communication device, thereby obtaining artificial calibration data.

[0066] In one embodiment, machine calibration data includes the sampling position of a data acquisition device at a sampling time along a preset travel path; and, based on the sampling position, communication signals from multiple anchor points received by the data acquisition device in a sampling pose. Human calibration data includes: the sampling position of a human at a sampling time along a preset travel path; and, based on the sampling position, communication signals from multiple anchor points received by the human.

[0067] In one specific embodiment, the target object is a sensing target equipped with a third short-range communication device, such as a car equipped with multiple Bluetooth and / or ultra-wideband (UWB) devices. Anchor points are components on the car equipped with Bluetooth and / or UWB devices, such as the center console, rearview mirror, doors, engine compartment, and trunk. The data acquisition device is a robot, AGV, or a combination of a robot and AGV equipped with a smart terminal. The smart terminal includes, but is not limited to, smartphones, smartwatches, smart bracelets, smart glasses, and Bluetooth headsets equipped with a first short-range communication device. The smart terminal is placed on the data acquisition device according to a preset acquisition posture, which includes the installation height, horizontal angle, and vertical angle of the first short-range communication device on the data acquisition device. The data acquisition device moves along a pre-planned preset path. During this movement, the position information of the data acquisition device itself, and the communication signals collected by the first short-range communication device from the third short-range communication devices at each anchor point on the target object at the corresponding position, together constitute the machine calibration data. Manual calibration data refers to calibration data collected by a human holding or wearing a smart terminal, following the same preset path as the data acquisition device, around the target object, and at the same data acquisition location for each anchor point, through a second short-range communication device.

[0068] Specifically, the sampling mode can be customized based on the performance of the short-range communication device. Taking a robot with a car as the target object and a smart terminal installed as the data acquisition device as an example, after the preset travel path is imported into the robot, the robot automatically travels along the preset travel path with the assistance of its internal sensors such as GPS and LiDAR, and sends its own position in real time to the smart terminal bound to the robot via Wi-Fi, data communication network, or Bluetooth. At the same time, the smart terminal also receives the raw measurement data of the anchor points on the car. Preferably, the smart terminal receives the anchor point position data, the BLE Bluetooth signal strength of the anchor point, and / or the UWB ultra-wideband distance data of the anchor point at a sampling rate of 10Hz. During the robot's movement and data collection process, the data collection methods include, but are not limited to, the smart terminal collecting data once at each position on the robot's travel path, or pausing at each position for a period of time and collecting data multiple times.

[0069] In one embodiment, after obtaining the original machine calibration data and manual calibration data, the original data of the machine calibration data and manual calibration data are preprocessed, such as sampling, missing value processing, outlier processing, filtering processing, etc., and the preprocessed data is used as the corresponding machine calibration data and / or manual calibration data.

[0070] Step S30: Compare the machine-calibrated data and the manually calibrated data to obtain the feature comparison results.

[0071] Specifically, the goal of machine calibration is to simulate manual calibration, thereby saving labor costs and improving data acquisition efficiency. Therefore, after acquiring machine calibration data, it is compared with manually calibrated data to determine their differences. In this embodiment, features can be extracted from both machine and manually calibrated data separately, and then the features of the two data are compared to obtain feature comparison results. In some embodiments, the robot calibration data and manually calibrated data can be directly compared to obtain the difference data; then, feature extraction is performed on the difference data using statistical knowledge to obtain feature comparison results.

[0072] In one embodiment, step S30 includes: extracting data features from machine calibration data and manual calibration data respectively to obtain machine data features and manual data features; comparing machine data features and manual data features to determine feature difference parameters.

[0073] In one embodiment, the short-range communication device includes Bluetooth Low Energy (BLE) and / or Ultra Wideband (UWB), and the data characteristics include at least one of the following features:

[0074] Signal strength distribution characteristics at each anchor point at each acquisition location;

[0075] The trend characteristics of Bluetooth signal strength variation at each anchor point with communication distance;

[0076] The trend characteristics of Bluetooth signal strength variation at each anchor point as the acquisition location changes;

[0077] The time-of-flight distribution characteristics of UWB signals at multiple anchor points at each acquisition location;

[0078] Characteristics of the flight time variation of UWB signals at each anchor point as a function of communication distance;

[0079] Characteristics of the flight time variation of UWB signals at each anchor point as the acquisition location changes;

[0080] The distribution characteristics of the measured distances at multiple anchor points at each acquisition pose;

[0081] The trend characteristics of the measured distance change of each anchor point with the communication distance; and

[0082] The trend characteristics of the measurement distance change of each anchor point as the collection location changes.

[0083] Step S40: Adjust the acquisition pose of the data acquisition device based on the feature comparison results.

[0084] Specifically, after obtaining the feature comparison results, the acquisition posture of the data acquisition device is adjusted based on the relevant information fed back by the feature comparison results, such as signal strength distribution and signal change trends. For example, the installation height, horizontal angle, and vertical angle of the first short-range communication device on the data acquisition device are adjusted. Preferably, the travel route and travel speed of the data acquisition device can also be adjusted based on the feature comparison results.

[0085] Step S50: Based on the preset travel path and the adjusted acquisition pose, acquire simulated calibration data from multiple anchor points of the target object through the data acquisition device.

[0086] Specifically, in real-world scenarios, the way a person holds or wears the smart terminal varies when manually collecting calibration data. For example, it may be held with the left hand, worn on the left wrist, held with the right hand, worn on the right wrist, worn on the chest, or worn on the head. Furthermore, there may be obstructions around the vehicle, such as parking lot pillars or pedestrians and vehicles obstructing the view. This can lead to differences in features between the data collected manually and the data collected by the robot, resulting in discrepancies in the positioning algorithm models trained based on the machine calibration data and the human calibration data. Therefore, the positioning calibration method in this embodiment is adapted to real-world scenarios. First, a human collects data using the same trajectory as the robot, and then extracts their respective features to analyze the differences between the two data sets. This allows for the setting of rules for the robot's data collection, such as the position and angle of the smart terminal on the robot, to simulate a real-world scenario. This ensures that the data collected by the robot is simulated data collected by a human, i.e., realistic calibration data.

[0087] In one embodiment, step S50 includes: training and validating the decision tree model based on simulated calibration data to obtain a positioning algorithm with a preset positioning accuracy and calibration parameters.

[0088] Specifically, the decision tree model is a type of machine learning model. Since each node in a decision tree represents a judgment of an attribute, and each branch represents the output of a judgment result, the data from each anchor point can be used as decision nodes to define the decision boundaries, thereby making decisions based on the anchor point data in the actual positioning scenario to achieve positioning. Based on these characteristics, this embodiment designs a positioning algorithm using a decision tree model based on information acquired via Bluetooth BLE and / or Ultra Wideband (UWB).

[0089] Step S60: Perform positioning calibration based on simulated calibration data to determine the calibration parameters of the positioning algorithm.

[0090] Specifically, after obtaining the simulated calibration data, the positioning algorithm model is trained and optimized using the data, and the calibration parameters of the positioning algorithm are determined based on the algorithm model and the simulated calibration data.

[0091] In the above positioning and calibration method, after obtaining the feature comparison difference between machine calibration and manual calibration data, the machine acquisition rules are optimized based on the difference, so that the machine calibration data is closer to the manual calibration data, thereby increasing the amount and efficiency of calibration data acquisition, realizing automated acquisition of calibration data, reducing manual intervention in the calibration process, effectively solving the problems of time-consuming and labor-intensive manual intervention, and saving time and human resources costs.

[0092] In one embodiment, such as Figure 5 As shown, after step S60, the following steps are also included:

[0093] Step S70: Provide a preset verification path.

[0094] Step S80: Based on the preset verification path, obtain simulated positioning information and manual positioning information.

[0095] Step S90: Based on the error judgment of simulated positioning information and manual positioning information, determine the accuracy of the positioning algorithm and calibration parameters.

[0096] In one embodiment, such as Figure 6 As shown, step S80 further includes:

[0097] Step S81: Based on the preset verification path and the adjusted acquisition pose, obtain simulated positioning information of multiple anchor points of the target object through the data acquisition device.

[0098] Step S82: Based on the preset verification path, obtain human positioning information by using multiple anchor points of the target object through real people.

[0099] The specific significance of steps S70 to S90 above is that, after the positioning calibration and positioning algorithm are completed, the accuracy of the positioning algorithm and calibration parameters can be judged based on the actual test results before deployment and application. For example, during real vehicle testing, the positioning accuracy can be judged based on the activation or deactivation of relevant functions. For example, when a person holds a smart terminal and approaches the vehicle, if the welcoming function is activated when they reach the welcoming range, it indicates that the accuracy of the current positioning information meets the standard, and the positioning algorithm and calibration parameters meet the design requirements. Another example is when a person holds a smart terminal and approaches the vehicle, if the door is unlocked when they reach the unlocking range, it indicates that the accuracy of the current positioning information meets the standard. Yet another example is when a person holds a smart terminal and moves away from the vehicle, if the door is locked when they enter the locking range, it indicates that the current positioning is accurate.

[0100] The positioning and calibration method of this application realizes the automated collection of calibration data by a robot simulating a real person. The calibration parameters are then processed by an algorithm model, which can reduce the human intervention in the entire calibration process and effectively solve the problems of slow speed, high cost, and data distortion between machine calibration and real person calibration.

[0101] This application also provides a positioning and calibration system, including a robot, an intelligent terminal, and a computing unit. The robot moves according to a preset travel path; the intelligent terminal is set on the robot and acquires machine calibration data in a preset acquisition pose; the computing unit is used to execute the steps described in any embodiment of the positioning and calibration method above.

[0102] In one embodiment, the data acquisition architecture of the positioning and calibration system is as follows: Figure 7 As shown, the preset travel path is accessed through the robot's (data acquisition device) API via the path planning app and then imported into the robot (data acquisition device). Subsequently, the robot (data acquisition device) and a mobile phone (first short-range communication device) equipped with the data calibration app are connected via Wi-Fi. The mobile phone (first short-range communication device) is then bound to the robot. The data calibration app loads a map through the robot's API and can obtain the robot's position in real time. The mobile phone is then connected to the car (target object) via Bluetooth. The data calibration app can obtain measurement data in real time at a sampling frequency of 10Hz. Finally, the robot's position information and the corresponding data are preprocessed and uploaded to the server for positioning calibration.

[0103] In one specific embodiment, a positioning calibration method is provided, applied to a positioning calibration system, such as... Figure 8 As shown, the method includes the following steps:

[0104] Step S100, data acquisition.

[0105] Specifically, before data collection, a preset travel path for the robot must first be set. This preset path can be set using a path planning program. Once the preset path is set, it can be imported into the robot (data acquisition device).

[0106] With the assistance of its internal sensors such as GPS and LiDAR, the robot (data acquisition device) automatically moves along a preset path and sends its real-time location to a mobile phone bound to the robot via Wi-Fi and map functions. The data calibration app receives the data and simultaneously receives the original measurement data of the car (target object) anchor point. This original measurement data consists of the robot's (data acquisition device's) position, Bluetooth signal strength, and / or UWB distance data received at a sampling rate of 10Hz.

[0107] After preprocessing the raw data, it is uploaded to the server as input for training the algorithm model. Preprocessing includes, but is not limited to, sampling, handling of missing and outlier values, and filtering.

[0108] Step S200: Adaptive algorithm training and parameter fitting.

[0109] Specifically, step S200 also includes Figure 9 The content shown is as follows:

[0110] 1. Data Analysis

[0111] Obtain machine calibration data, i.e., machine data. Obtain real-person data collected by test personnel moving along the robot's preset path, i.e., human calibration data. Extract the respective features to analyze the differences between the two data, and set rules for robot data collection, such as the position and angle of the mobile phone on the robot, to simulate real-person scenarios. After that, the data collected by the robot are all simulated real-person data, i.e., simulated calibration data. The extracted features may include: (1) the trend of Bluetooth or UWB data with distance from the vehicle; (2) the strength distribution of data at each anchor point around the vehicle body; (3) the trend of data mean, variance, and unit time.

[0112] 2. Perform model building, validation, and optimization.

[0113] Based on the collected data and data analysis results, a preliminary algorithm model is constructed using the decision tree algorithm in machine learning. The algorithm model is then trained using human calibration data as input, or the decision boundaries are defined using human calibration data to test the algorithm's localization results. Machine calibration data is then used as input to obtain the algorithm's localization results. If the localization performance differs between the two datasets based on the same algorithm and parameters, the calibration parameters in the algorithm are optimized, or the robot data is adjusted based on the characteristic differences between the machine calibration data and the human calibration data. This makes the machine calibration data more realistically simulate the human calibration data, ensuring that the trained algorithm model and calibration parameters better match the real-world scenario.

[0114] 3. Calibration parameter fitting

[0115] After determining the localization algorithm, simulated calibration data is collected again using a robot. This simulated calibration data includes the robot's position and the data of each anchor point on the car at the corresponding position. Then, based on the constructed localization algorithm, calibration parameters, i.e., the decision boundary, are fitted. The fitting process includes the following steps:

[0116] Input simulated calibration data, and based on the robot's position data, filter out measurement data at fixed locations and calculate statistical values, such as average and minimum values, as the starting point of the decision boundary for that fixed location. The selection of the fixed location depends on the functional requirements, such as 3 meters for the car unlocking range.

[0117] Then, using the decision boundary as the initial value for iteration, the decision boundary is iteratively fitted. The iteration direction can be either positive or negative, and the step size depends on the specific data; for example, it can be set to 1dBm for Bluetooth signals. The decision boundary and simulated calibration data are used as inputs to the algorithm model. The algorithm model is executed, and the algorithm's positioning accuracy is output. The simulated calibration data remains unchanged. The decision boundary is iteratively fitted, and the positioning accuracy ACC of the input decision boundary is calculated. The positioning accuracy ACC can be obtained using the following method: for example, when the unlock boundary is set to 3m, the number of data points within and outside 3m of the algorithm's positioning result obtained from the input decision boundary is counted, and then the positioning accuracy ACC is obtained using the following formula.

[0118]

[0119] In the above formula, The number of actual locations identified within 3m; The number of actual locations identified at a distance of 3m; The number of incorrect locations identified within 3m; This represents the number of incorrect locations identified at a distance of 3m.

[0120] By combining this positioning accuracy with appropriate iteration termination conditions, such as terminating after 10 iterations and selecting the decision boundary with the highest accuracy as the optimal decision boundary, or terminating when the algorithm's positioning accuracy corresponding to the current input decision boundary is lower than the previous accuracy, and taking the previous decision boundary as the optimal decision boundary, the iteration strategy for the above decision boundary will converge to a certain decision boundary after a finite number of iterations. By iteratively fitting this optimal decision boundary, the decision boundary parameters can be calibrated.

[0121] The calibration parameters are uploaded to the calibration app and used together with the algorithm model as input for real vehicle testing to verify the reliability of the calibration parameters.

[0122] Step S300: Deploy the application.

[0123] Specifically, after positioning and calibration are completed, testing is conducted on a real vehicle to deploy the application. First, one or more short-range communication devices, such as Bluetooth BLE or UWB, are installed in the vehicle. A mobile phone is connected to the in-vehicle device, and a person holds the phone and continuously moves it closer to and further away from the car. The test is conducted to see if the car responds accordingly when the phone reaches a certain boundary, such as unlocking, locking, or welcoming. Figure 10 As shown, along the direction closest to the vehicle, the defined boundaries are, in sequence, the welcoming boundary, the locking boundary, and the unlocking boundary. The positioning algorithm model containing calibration parameters is deployed to an actual vehicle for real-time accuracy testing in real-vehicle areas, including functions such as contactless unlocking, locking, and welcoming.

[0124] The unlocking test involves approaching the vehicle from a distance and walking to the unlocking boundary when it is 3 meters away from the vehicle, and testing whether the vehicle unlocks automatically. The locking test involves approaching the vehicle from a distance and walking to the locking boundary when it is 5 meters away from the vehicle, and testing whether the vehicle locks automatically. The welcoming test involves approaching the vehicle from a distance and walking to the vicinity of the welcoming boundary when it is 8 to 10 meters away from the vehicle, and testing whether the vehicle exhibits a welcoming behavior.

[0125] After the real vehicle test is completed, check whether the real vehicle test results and parameter accuracy are consistent. If they are consistent, the algorithm model and calibrated parameters can be deployed to the actual vehicle. If they are inconsistent, analyze the robot data to simulate the human body and the reliability of the calibration accuracy until the real vehicle test results and parameter accuracy results are consistent.

[0126] In this embodiment, the robot can collect a large amount of simulated calibration data according to a preset travel path and collection rules, replacing manual labor in repetitive data collection work, thus achieving high efficiency. Furthermore, the robot data and human data were analyzed. Considering the way humans hold mobile phones, the robot data can be used to simulate human data to train an algorithm model applicable to real-world scenarios. The trained model exhibits good generalization and robustness. Finally, the calibration parameters are tested on a real vehicle to ensure the reliability of the trained algorithm model and the calibrated parameters in real-world scenarios. The automated calibration method and system proposed in this embodiment can achieve rapid calibration of algorithm parameters for different needs, saving resources and improving efficiency.

[0127] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0128] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0129] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0130] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A positioning calibration method, characterized in that: The following steps are involved: S10: providing a preset travel path; S20: Based on the preset travel path, a data acquisition device equipped with a first short-range communication device acquires machine calibration data for multiple anchor points of a target object in a preset acquisition posture, and a real person equipped with a second short-range communication device acquires manual calibration data for the multiple anchor points of the target object, each of the anchor points includes at least one third short-range communication device, thereby communicating with the first short-range communication device of the data acquisition device and the second short-range communication device of the real person; S30: comparing the machine calibration data with the manual calibration data to obtain a feature comparison result; S40: adjusting the acquisition posture of the data acquisition device based on the feature comparison result; S50: Based on the preset travel path and the adjusted acquisition posture, obtaining simulation calibration data for multiple anchor points of the target object through a data acquisition device; S60: Determine the positioning accuracy and calibration parameters of the positioning algorithm through the simulated calibration data.

2. The positioning calibration method according to claim 1, characterized in that: In step S20, the machine calibration data includes: The data collection device has a collection position at a sampling time in the preset travel path; and Based on the acquisition position, the data acquisition device receives communication signals of the plurality of anchor points in the acquisition posture; The manual calibration data includes: The collection position of the real person at the sampling moment in the preset travel path; and Based on the acquisition position, the real person receives communication signals from the multiple anchor points.

3. The positioning calibration method according to claim 2, characterized in that: The preset travel path includes at least one of a radial path centered on the target object, a surrounding path, and a combination of the radial path and the surrounding path.

4. The positioning calibration method according to claim 2, characterized in that: The acquisition posture includes the height, horizontal angle and vertical angle of the first short-range communication device of the data acquisition equipment.

5. The positioning calibration method according to claim 2, characterized in that: Step S30 includes: Extracting data features from the machine calibration data and the manual calibration data respectively to obtain machine data features and manual data features; The machine data feature is compared with the manual data feature to determine feature difference parameters.

6. The positioning calibration method according to claim 5, characterized in that: The short-range communication device includes low-power Bluetooth BLE and / or UWB, and the data characteristics include at least one of the following characteristics: At each of the acquisition positions, signal strength distribution characteristics of the plurality of anchor points; The signal strength variation trend characteristics of each anchor point as the communication distance changes; The signal strength variation trend characteristics of each anchor point as the acquisition position changes; At each of the acquisition positions, the signal flight time distribution characteristics of the plurality of anchor points; The signal flight time variation trend characteristics of each anchor point as the communication distance changes; The signal flight time variation trend characteristics of each anchor point as the acquisition position changes; At each of the acquisition positions, distribution characteristics of the measured distances of the plurality of anchor points; The measured distance variation trend characteristics of each anchor point as the communication distance changes; as well as The measured distance variation trend characteristics of each of the anchor points as the acquisition position changes.

7. The positioning calibration method according to claim 1, characterized in that: Step S50 includes: Model training and model verification are performed on the decision tree model based on the simulated calibration data to obtain the positioning algorithm with preset positioning accuracy and the calibration parameters.

8. The positioning calibration method according to claim 1, characterized in that: After step S60, the following steps are also included: S70: providing a preset verification path; S80: Based on the preset verification path, obtain simulated positioning information and manual positioning information; S90: Determine the accuracy of the positioning algorithm and the calibration parameters based on the errors of the simulated positioning information and the manual positioning information.

9. The positioning calibration method according to claim 8, characterized in that: Step S80 also includes: S81: Based on the preset verification path and the adjusted acquisition posture, acquiring the simulated positioning information for the multiple anchor points of the target object through a data acquisition device; S82: Based on the preset verification path, the manual positioning information is obtained by the real person for the multiple anchor points of the target object.

10. A positioning and calibration system, comprising a robot, an intelligent terminal and a computing unit, characterized in that: The computing unit is used to execute the positioning and calibration method according to any one of claims 1 to 9; The robot moves according to the preset travel path; The intelligent terminal is arranged on the robot to obtain the machine calibration data in the preset acquisition posture.

Images