Physical parameter estimation method and device based on sensor

Abstract

The embodiment of the invention provides a physical parameter estimation method and device based on a sensor. Relates to the technical field of data processing. According to the specific scheme, the method comprises the following steps: obtaining a first observation value of a first sensor obtained based on collected data and a second observation value obtained based on data collected by a second sensor; calculating a difference value between the first observation value and the second observation value; according to the difference value and observation noise of a target sensor, a normalization error is obtained, and the target sensor is a first sensor or a second sensor; adjusting the first observation noise of the first sensor and the second observation noise of the second sensor according to the normalization error; and estimating a first predicted value of the target physical parameter according to the first observation noise and the second observation noise. By applying the scheme provided by the embodiment of the invention, the accuracy of the determined physical parameters can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a sensor-based physical parameter estimation method and device. Background Art

[0002] In the fields of aviation, aerospace, unmanned driving, and robotics, it is necessary to process data based on the data collected by the sensors mounted on the equipment to determine the physical parameters of the equipment, and then the equipment can perform control functions such as equipment positioning, path planning, controlling the equipment's posture, and controlling the movement of the equipment based on the physical parameters. For example, a drone can process data based on the data collected by the gyroscope mounted on the drone to obtain the heading angle of the drone as the physical parameter of the equipment, and then control the heading of the drone based on the obtained heading angle. The accuracy of the heading angle will affect the accuracy of controlling the heading of the drone. It can be seen that the accuracy of the physical parameters determined based on the data collected by the sensors is very important.

[0003] In the prior art, the device processes data based on data collected by a single sensor to obtain physical parameters. However, a single sensor often has problems such as limited accuracy and noise interference. Therefore, the accuracy of the data collected by the sensor will affect the accuracy of the physical parameters of the device, resulting in low accuracy of the determined physical parameters. Summary of the invention

[0004] The purpose of the embodiments of the present invention is to provide a method and device for estimating physical parameters based on sensors, so as to improve the accuracy of the determined physical parameters. The specific technical solution is as follows:

[0005] According to one aspect of an embodiment of the present invention, a sensor-based physical parameter estimation method is provided, the method comprising:

[0006] Acquire a first observation value of a first sensor and a second observation value of a second sensor provided on the mobile device; wherein the first observation value is: an observation value of a target physical parameter obtained based on data collected by the first sensor, and the second observation value is: an observation value of the target physical parameter obtained based on data collected by the second sensor;

[0007] calculating a difference between the first observation and the second observation;

[0008] Obtaining a normalized error according to the difference and an observation noise of a target sensor, wherein the target sensor is: the first sensor or the second sensor;

[0009] adjusting the first observation noise and the second observation noise according to the normalized error;

[0010] A first predicted value of a target physical parameter is estimated based on the first observation noise and the second observation noise.

[0011] In one embodiment of the present invention, adjusting the first observation noise and the second observation noise according to the normalized error includes:

[0012] If the normalized error is greater than a preset error threshold, adding the first observation noise and the second observation noise to obtain a first adjusted noise of the first sensor and a second adjusted noise of the second sensor;

[0013] If the normalized error is less than or equal to a preset error threshold, the first observation noise and the second observation noise are reduced to obtain a third adjusted noise of the first sensor and a fourth adjusted noise of the second sensor.

[0014] In one embodiment of the present invention, the adding the first observation noise and the second observation noise to obtain a first adjusted noise of the first sensor and a second adjusted noise of the second sensor includes:

[0015] Increasing the first observed noise by a preset noise growth step to obtain an increased first observed noise; and obtaining a first adjusted noise based on the increased first observed noise;

[0016] The second observed noise is increased by a preset noise growth step to obtain an increased second observed noise; and based on the increased second observed noise, a second adjusted noise is obtained.

[0017] In one embodiment of the present invention, obtaining the first adjusted noise based on the increased first observed noise includes:

[0018] If the increased first observed noise is greater than or equal to the preset noise upper limit threshold, the preset noise upper limit threshold is used as the first adjusted noise; if the increased first observed noise is less than the preset noise upper limit threshold, the increased first observed noise is used as the first adjusted noise.

[0019] In one embodiment of the present invention, the preset error threshold is determined in the following manner: the preset error threshold is determined from a preset chi-square distribution table according to preset degrees of freedom and a preset significance level value.

[0020] In one embodiment of the present invention, the observation noise of the sensor is represented by an observation noise matrix, wherein the observation noise matrix is ​​a square matrix, the order of the observation noise matrix is ​​the same as the number of sensors arranged on the mobile device, and each diagonal element in the observation noise matrix corresponds to the observation noise of each sensor;

[0021] The adding of the first observation noise and the second observation noise comprises:

[0022] Increasing the values ​​of the diagonal elements in the observation noise matrix corresponding to the first sensor, and increasing the values ​​of the diagonal elements in the observation noise matrix corresponding to the second sensor;

[0023] The reducing the first observation noise and the second observation noise comprises:

[0024] The values ​​of the diagonal elements corresponding to the first sensor in the observation noise matrix are reduced, and the values ​​of the diagonal elements corresponding to the second sensor in the observation noise matrix are reduced.

[0025] In one embodiment of the present invention, obtaining a normalized error according to the difference and the observation noise of the target sensor includes:

[0026] Determine, according to the difference, a difference representation value between the first observation value and the second observation value;

[0027] Determine a normalization basis according to the observation noise of the target sensor;

[0028] A normalized error is obtained based on the gap characterization value and a normalized basis.

[0029] In one embodiment of the present invention, the method further comprises:

[0030] Predicting a target physical parameter of the mobile device based on data collected by a sensor disposed on the mobile device to obtain a second predicted value of the target physical parameter;

[0031] Determine the prior state covariance based on the posterior state covariance and the process noise covariance;

[0032] After adjusting the first observation noise and the second observation noise according to the normalized error, the method further includes:

[0033] For each sensor, determine a gain coefficient of the sensor based on the adjusted observation noise of the sensor and the prior state covariance, wherein the gain coefficient of the sensor represents a degree of change in noise uncertainty of an observation value obtained based on data collected by the sensor;

[0034] Determining an updated a posteriori state covariance based on the a priori state covariance and the determined gain coefficients of each sensor;

[0035] The estimating a first predicted value of a target physical parameter according to the first observation noise and the second observation noise includes:

[0036] A first predicted value of the target physical parameter is estimated based on the determined gain coefficients of the sensors and the measurement residuals between the observed values ​​and the second predicted value.

[0037] In one embodiment of the present invention, the target physical parameter is: heading angle;

[0038] The step of predicting a target physical parameter of the mobile device based on data collected by a sensor disposed on the mobile device to obtain a second predicted value of the target physical parameter includes:

[0039] Predict the heading angle change according to the roll angle, the pitch angle, the angular velocity of the roll angle and the angular velocity of the pitch angle collected by the sensor mounted on the mobile device;

[0040] Based on the predicted heading angle change, a second predicted value of the heading angle is obtained.

[0041] According to another aspect of an embodiment of the present invention, a sensor-based physical parameter estimation device is provided, the device comprising:

[0042] An observation value acquisition module, used to acquire a first observation value of a first sensor and a second observation value of a second sensor provided on the mobile device; wherein the first observation value is: an observation value of a target physical parameter obtained based on data collected by the first sensor, and the second observation value is: an observation value of the target physical parameter obtained based on data collected by the second sensor;

[0043] A difference calculation module, used for calculating the difference between the first observation value and the second observation value;

[0044] a normalized error acquisition module, configured to acquire a normalized error according to the difference and an observation noise of a target sensor, wherein the target sensor is: the first sensor or the second sensor;

[0045] an observation noise adjustment module, configured to adjust the first observation noise and the second observation noise according to the normalized error;

[0046] A numerical estimation module is used to estimate a first predicted value of a target physical parameter based on the first observation noise and the second observation noise.

[0047] In one embodiment of the present invention, the observation noise adjustment module is specifically used to: if the normalized error is greater than a preset error threshold, increase the first observation noise and the second observation noise to obtain a first adjusted noise of the first sensor and a second adjusted noise of the second sensor; if the normalized error is less than or equal to the preset error threshold, reduce the first observation noise and the second observation noise to obtain a third adjusted noise of the first sensor and a fourth adjusted noise of the second sensor.

[0048] In one embodiment of the present invention, the observation noise adjustment module is specifically used to: increase the first observation noise by a preset noise growth step to obtain an increased first observation noise; obtain a first adjustment noise based on the increased first observation noise; increase the second observation noise by a preset noise growth step to obtain an increased second observation noise; and obtain a second adjustment noise based on the increased second observation noise.

[0049] In one embodiment of the present invention, the observation noise adjustment module is specifically used to: if the increased first observation noise is greater than or equal to a preset noise upper limit threshold, then use the preset noise upper limit threshold as the first adjustment noise; if the increased first observation noise is less than the preset noise upper limit threshold, then use the increased first observation noise as the first adjustment noise.

[0050] In one embodiment of the present invention, the preset error threshold is determined in the following manner: the preset error threshold is determined from a preset chi-square distribution table according to preset degrees of freedom and a preset significance level value.

[0051] In one embodiment of the present invention, the observation noise of the sensor is represented by an observation noise matrix, wherein the observation noise matrix is ​​a square matrix, the order of the observation noise matrix is ​​the same as the number of sensors arranged on the mobile device, and each diagonal element in the observation noise matrix corresponds to the observation noise of each sensor;

[0052] The observation noise adjustment module is specifically used to: increase the values ​​of the diagonal elements corresponding to the first sensor in the observation noise matrix, and increase the values ​​of the diagonal elements corresponding to the second sensor in the observation noise matrix; reduce the values ​​of the diagonal elements corresponding to the first sensor in the observation noise matrix, and reduce the values ​​of the diagonal elements corresponding to the second sensor in the observation noise matrix.

[0053] In one embodiment of the present invention, the observation noise adjustment module is specifically used to determine a gap characterization value between the first observation value and the second observation value based on the difference; determine a normalization basis based on the observation noise of the target sensor; and obtain a normalized error based on the gap characterization value and the normalization basis.

[0054] In one embodiment of the present invention, the device further comprises:

[0055] A parameter prediction module, configured to predict a target physical parameter of the mobile device based on data collected by a sensor disposed on the mobile device, and obtain a second predicted value of the target physical parameter;

[0056] A covariance determination module, used for determining a priori state covariance based on a posterior state covariance and a process noise covariance;

[0057] A gain coefficient determination module, configured to determine, for each sensor, a gain coefficient of the sensor based on the adjusted observation noise of the sensor and the prior state covariance, wherein the gain coefficient of the sensor represents a degree of change in noise uncertainty of an observation value obtained based on data collected by the sensor;

[0058] A covariance updating module, configured to determine an updated a posteriori state covariance based on the a priori state covariance and the determined gain coefficients of each sensor;

[0059] The observation noise adjustment module is specifically used to estimate the first predicted value of the target physical parameter based on the determined gain coefficients of each sensor and the measurement residuals between each observation value and the second predicted value.

[0060] In one embodiment of the present invention, the target physical parameter is: heading angle;

[0061] The parameter prediction module is specifically used to: predict the heading angle change according to the roll angle, pitch angle, angular velocity of the roll angle and angular velocity of the pitch angle collected by a sensor loaded on the mobile device; and obtain a second predicted value of the heading angle based on the predicted heading angle change.

[0062] According to another aspect of an embodiment of the present invention, there is provided an electronic device, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;

[0063] Memory, used to store computer programs;

[0064] The processor is used to implement any of the above-mentioned sensor-based physical parameter estimation methods when executing the program stored in the memory.

[0065] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, any of the above-mentioned sensor-based physical parameter estimation methods is implemented.

[0066] According to another aspect of the embodiments of the present invention, a computer program product comprising instructions is provided, which, when executed on a computer, enables the computer to execute any of the above-mentioned sensor-based physical parameter estimation methods.

[0067] Beneficial effects of the embodiments of the present invention:

[0068] In the sensor-based physical parameter estimation method provided by the embodiment of the present invention, the observation noise of each sensor can characterize the noise with uncertainty contained in the data collected by the sensor. By adjusting the observation noise of each sensor, the gain coefficient of each sensor can be adjusted, and the weight of the observation value of each sensor in the physical parameter estimation can be indirectly adjusted. According to the normalized error obtained by the difference between the first observation value and the second observation value and the observation noise of the target sensor, the uncertainty of the data collected by the first sensor and the second sensor can be determined. Therefore, adjusting the observation noise according to the above normalized error can be regarded as adjusting the observation noise according to the uncertainty of the data collected by the sensor. In this way, the weight of the observation value of each sensor in the physical parameter estimation can be flexibly adjusted according to the uncertainty of the data collected by the sensor, thereby reducing the influence of the data collected by the sensor with high uncertainty on the accuracy of the parameter estimation, improving the influence of the data collected by the sensor with low uncertainty on the accuracy of the parameter estimation, and improving the accuracy of the first predicted value of the estimated target physical parameter.

[0069] Of course, it is not necessary to achieve all of the advantages described above at the same time to implement any product or method of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other embodiments can also be obtained based on these drawings.

[0071] Figure 1 A schematic diagram of a flow chart of a sensor-based physical parameter estimation method provided in an embodiment of the present invention;

[0072] Figure 2 A schematic diagram of a flow chart of an observation noise adjustment method provided by an embodiment of the present invention;

[0073] Figure 3A schematic flow chart of another sensor-based physical parameter estimation method provided in an embodiment of the present invention;

[0074] Figure 4 A schematic flow chart of another sensor-based physical parameter estimation method provided in an embodiment of the present invention;

[0075] Figure 5 A schematic diagram of the structure of a sensor-based physical parameter estimation device provided in an embodiment of the present invention;

[0076] Figure 6 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0077] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field based on the present invention belong to the scope of protection of the present invention.

[0078] The execution subject and application scenario of the embodiment of the present invention are described below.

[0079] The solution provided by the embodiment of the present invention can be applied to various types of autonomous mobile devices, for example, autonomous mobile devices such as aircraft, vehicles and robots, and specifically, can be applied to devices such as drones and autonomous driving vehicles. The above autonomous mobile devices can be widely used to perform different types of navigation tasks, such as unmanned driving, aerospace, robot navigation and other navigation tasks. For the convenience of description, the execution subjects provided by the embodiments of the present invention are collectively referred to as mobile devices.

[0080] The frames are described below.

[0081] The scheme provided by the embodiment of the present invention can be executed periodically. For example, if the first physical parameter estimation is performed and the first predicted value of the target physical parameter is obtained, the first predicted value obtained can be referred to as: the first predicted value of the target physical parameter of the first frame, if the physical parameter estimation is performed again, that is, the second physical parameter estimation is performed, and the first predicted value of the target physical parameter is obtained again, the first predicted value obtained again can be referred to as: the first predicted value of the target physical parameter of the second frame, in this case, in the process of performing the second physical parameter estimation, the first predicted value of the target physical parameter of the first frame can be referred to as: the first predicted value of the target physical parameter of the previous frame, and similarly, the data obtained in the process of performing the first physical parameter estimation can also become the data of the previous frame, for example, the second predicted value of the previous frame, the observed value of the previous frame, the observed noise of the previous frame, and the gain coefficient of the previous frame, etc., that is, the data of the previous frame represents the data obtained when the physical parameter estimation was performed last time. Similarly, when performing the second physical parameter estimation, the first predicted value of the target physical parameter of the second frame can be called: the first predicted value of the target physical parameter of the current frame. Similarly, the data obtained in the process of performing the second physical parameter estimation can also become the data of the current frame, for example, the second predicted value of the current frame, the observed value of the current frame, the observed noise of the current frame, the gain coefficient of the current frame, and the fusion value of the current frame, etc. For example, the following steps S101-S106 can be executed cyclically, and each execution of steps S101-S106 can obtain the first predicted value of the target physical parameter of this frame.

[0082] The solution provided by the embodiment of the present invention is described in detail below.

[0083] In one embodiment of the present invention, see Figure 1 A flow chart of a sensor-based physical parameter estimation method is provided, the method comprising the following steps S101-S106.

[0084] Step S101: Acquire a first observation value of a first sensor and a second observation value of a second sensor provided on a mobile device.

[0085] The first observation value is: the observation value of the target physical parameter obtained based on the data collected by the first sensor, and the second observation value is: the observation value of the target physical parameter obtained based on the data collected by the second sensor.

[0086] The target physical parameters are physical parameters of the mobile device. For example, the target physical parameters may include navigation positioning parameters such as heading angle, roll angle, pitch angle, angular velocity, speed, acceleration, mileage, and position. The number of sensors loaded on the mobile device is at least two. The sensors loaded on the mobile device may include sensors of the same type or sensors of at least two different sensor types. The number of sensors of each sensor type may be one or more. The situation of sensors of different sensor types is described in the following embodiments and will not be described in detail here.

[0087] Sensors of different sensor types can collect different types of data, and the methods for obtaining the observed values ​​of the target physical parameters of the sensors for different types of data are also different. Moreover, the methods for obtaining the observed values ​​of the target physical parameters of the sensors for different types of target physical parameters are also different.

[0088] The following uses different types of target physical parameters as examples to illustrate how to obtain the observed values ​​of the target physical parameters.

[0089] 1. The target physical parameters are: speed

[0090] The following describes a method for obtaining the observed value of the target physical parameter by taking the target physical parameter as speed as an example.

[0091] If the sensor is an accelerometer, then the mobile device may integrate the acceleration collected by the accelerometer to obtain an observed value of the speed of the mobile device.

[0092] If the sensor is a visual sensor, then the mobile device can use the optical flow method to obtain the observed value of the speed of the mobile device for the image data collected by the visual sensor.

[0093] 2. The target physical parameters are: heading angle

[0094] If the sensor is a gyroscope, then the mobile device may integrate the angular velocity of the heading angle collected by the gyroscope to obtain an observed value of the heading angle of the mobile device.

[0095] If the sensor is a visual sensor, then the mobile device can use the optical flow method to obtain the observation value of the heading angle of the mobile device for the image data collected by the visual sensor.

[0096] For each sensor, the observed value of the target physical parameter can be obtained in the above manner. That is, based on the data collected by the first sensor, the first observed value of the target physical parameter can be obtained. Similarly, based on the data collected by the second sensor, the second observed value of the target physical parameter can be obtained.

[0097] Step S102: Calculate the difference between the first observation value and the second observation value.

[0098] Specifically, the absolute value of the difference between the first observation value and the second observation value may be used as the difference between the first observation value and the second observation value.

[0099] Step S103: Obtain a normalized error according to the difference and the observation noise of the target sensor.

[0100] The target sensor is: the first sensor or the second sensor. Each sensor is provided with observation noise corresponding to the sensor.

[0101] In one implementation, the mobile device may obtain a normalized error between the first observation value and the second observation value based on the difference and the observation noise of the first sensor.

[0102] In another implementation, the mobile device may obtain a normalized error between the first observation value and the second observation value based on the difference and the observation noise of the second sensor.

[0103] In another implementation, the mobile device may obtain the normalized error between the first observation value and the second observation value based on the above-mentioned "one implementation" and "another implementation". In this case, the number of normalized errors between the first observation value and the second observation value obtained is two. Then, the observation noise adjustment may be performed for each normalized error obtained. For example, assuming that the normalized errors between the two first observation values ​​and the second observation value obtained are normalized error A and normalized error B, then step S104 may be performed for normalized error A, and the observation noise adjustment may be performed for the observation noise of the first sensor and the observation noise of the second sensor, and then step S104 may be performed for normalized error B, and the observation noise adjustment may be performed for the observation noise of the first sensor and the observation noise of the second sensor.

[0104] Specifically, in the above implementation, the mobile device may obtain the normalized error between the first observation value and the second observation value according to the following steps A to C:

[0105] Step A: Determine a difference characterization value between the first observation value and the second observation value based on the difference.

[0106] In one implementation, the mobile device may calculate the square of the difference as a value representing the gap between the first observation value and the second observation value.

[0107] Step B: Determine the normalization basis based on the observation noise of the target sensor.

[0108] In one implementation, the mobile device may calculate the square value of the observation noise of the target sensor as the normalization basis.

[0109] Step C: Obtain a normalized error based on the gap characterization value and the normalized basis.

[0110] In one implementation, the mobile device may calculate the quotient of the difference representation value and the normalized basis as the normalized error. For example, the normalized error may be obtained by the following expression:

[0111] ;

[0112] in, is the normalized error, is the above difference, is the observation noise of the target sensor, is the gap characterization value, In this way, the normalized error between the first observation value and the second observation value can be calculated more accurately to more accurately determine whether there is a significant difference between the first observation value of the first sensor and the second observation value of the second sensor, thereby more accurately adjusting the observation noise of the sensor and improving the accuracy of the target physical parameter estimation.

[0113] Step S104: adjusting the first observation noise of the first sensor and the second observation noise of the second sensor according to the normalized error.

[0114] The specific implementation of step S104 of adjusting the observed noise is described in the following embodiments and will not be described in detail here.

[0115] In addition, when the number of sensors provided on the mobile device is three or more, the mobile device may perform steps S101-S104 for each group of sensors to adjust the observation noise of the group of sensors until the observation noise is adjusted for each group of sensors, wherein a group of sensors includes two different sensors among the sensors provided on the mobile device. The mobile device may loop through each group of sensors, for example, through a double loop, for example, based on (for (int i = 0; ...; i++) and for (int j = i + 1; ...; j++)).

[0116] Step S105: Estimate a first predicted value of the target physical parameter according to the first observation noise and the second observation noise.

[0117] The specific implementation of the first prediction value estimation in step S105 is described in the following embodiments and will not be described in detail here.

[0118] In the sensor-based physical parameter estimation method provided by the embodiment of the present invention, the observation noise of each sensor can characterize the noise with uncertainty contained in the data collected by the sensor. By adjusting the observation noise of each sensor, the gain coefficient of each sensor can be adjusted, and the weight of the observation value of each sensor in the physical parameter estimation can be indirectly adjusted. According to the normalized error obtained by the difference between the first observation value and the second observation value and the observation noise of the target sensor, the uncertainty of the data collected by the first sensor and the second sensor can be determined. Therefore, adjusting the observation noise according to the above normalized error can be regarded as adjusting the observation noise according to the uncertainty of the data collected by the sensor. In this way, the weight of the observation value of each sensor in the physical parameter estimation can be flexibly adjusted according to the uncertainty of the data collected by the sensor, thereby reducing the influence of the data collected by the sensor with high uncertainty on the accuracy of the parameter estimation, improving the influence of the data collected by the sensor with low uncertainty on the accuracy of the parameter estimation, and improving the accuracy of the first predicted value of the estimated target physical parameter.

[0119] In one embodiment of the present invention, the sensors loaded on the mobile device include: sensors of at least two different sensor types.

[0120] For different target physical parameters, the sensor types used may also be different. For example, if the target physical parameter is heading angle, the sensor types of the sensor may include: magnetometer, RTK (Real-time Kinematic) antenna, visual sensor, gyroscope, and ESKF (Error State Kalman Filter). If the target physical parameter is mileage, the sensor types of the sensor may include: accelerometer, odometer, and GNSS (Global Navigation Satellite System).

[0121] A single type of sensor may have limitations for different working environments of mobile devices.

[0122] The limitations of different types of sensors are explained below.

[0123] The magnetometer is greatly affected by changes in the environmental magnetic field, especially in complex environments such as cities and underground. For example, the presence of various facilities in cities that easily cause large changes in the magnetic field makes the magnetometer easily interfered by the complex environment, which in turn leads to a large error in the heading angle obtained based on the data collected by the magnetometer.

[0124] RTK antennas have the advantage of high precision, but they have strict requirements on the working environment conditions of mobile devices. For example, if the working environment of the mobile device needs to have signal shielding, the accuracy of the data collected by the RTK antenna will be greatly affected. Or the accuracy of the data collected by the RTK antenna will also be affected by factors such as satellite line of sight, making it difficult for the RTK antenna to provide accurate data in harsh working environments.

[0125] Visual sensors rely on the clarity of visual features. In low-light and complex scene working environments, it is difficult for visual sensors to obtain clear visual data. If the clarity of visual features is low, it will be difficult to obtain accurate data based on the data collected by the visual sensors.

[0126] The gyroscope can provide continuous angular velocity data, but when used for a long time, it will produce a drift effect due to the characteristics of the gyroscope, which will cause the errors in the data collected by the gyroscope to gradually accumulate, making it difficult to obtain accurate data.

[0127] The use of sensors of multiple different sensor types can adjust the observation noise of sensors with lower uncertainty in the data collected by the sensors in the current working environment in order to deal with changes in sensor noise in a dynamically changing and complex working environment, thereby increasing the weight of the observation values ​​of the above sensors in the estimation of physical parameters, and reducing the weight of the observation values ​​of sensors with lower uncertainty in the data collected by the sensors in the estimation of physical parameters, thereby improving the accuracy of the first predicted value of the target physical parameter.

[0128] Using sensors of multiple different sensor types can effectively reduce the interference and noise problems that a single sensor type sensor may be subject to from the working environment. When a sensor of one sensor type is affected by the interference and noise of the working environment, sensors of other sensor types may be less affected by the interference and noise of the working environment. For sensors of other sensor types, the observed values ​​of the target physical parameters obtained based on the data collected by sensors of other sensor types are still relatively accurate, which can improve the accuracy of the first predicted value of the obtained target physical parameters. In this way, the observation noise and gain coefficient can be automatically adjusted according to the uncertainty of the data of different sensors, so as to improve the stability and accuracy of the mobile device in different working environments, and improve the robustness and reliability of the mobile device. Combining the advantages of sensors of multiple sensor types, the accuracy of the target physical parameter estimation is significantly improved, and the dependence on high-precision sensors is reduced. In the case of interference in a variable environment such as electromagnetic interference, lighting conditions, multipath effects in urban environments, and signal shielding in mountainous areas, the sensor-based physical parameter estimation method provided in the embodiment of the present invention can provide a more stable and accurate target physical parameter estimation.

[0129] The specific implementation of the observation noise adjustment in step S104 is described below.

[0130] In one embodiment of the present invention, see Figure 2 A flow chart of an observation noise adjustment method is provided, wherein the method comprises the following steps S201-S203 for adjusting the observation noise.

[0131] Step S201: determine whether the normalized error is greater than a preset error threshold.

[0132] If the judgment is yes, that is, the normalized error is greater than the preset error threshold, step S202 is executed; if the judgment is no, that is, the normalized error is less than or equal to the preset error threshold, step S203 is executed.

[0133] Step S202: adding the first observation noise and the second observation noise to obtain a first adjusted noise of the first sensor and a second adjusted noise of the second sensor.

[0134] The following describes a method of adding the first observation noise to obtain the first adjustment noise.

[0135] The mobile device may increase the first observed noise by a preset noise growth step to obtain an increased first observed noise, obtain a first adjusted noise based on the increased first observed noise, and use the obtained first adjusted noise as the adjusted first observed noise.

[0136] For example, the noise growth step can be set to 0.1f, where f is the value of the observation noise of the first sensor before adjustment. Then, the value of the observation noise of the first sensor after increase is: f+0.1f, and f+0.1f is used as the adjusted first observation noise.

[0137] Furthermore, a noise upper threshold may be set to set the maximum value of the observed noise. The mobile device may obtain the first adjusted noise based on the increased first observed noise in the following manner:

[0138] If the increased first observed noise is greater than or equal to the preset noise upper limit threshold, the preset noise upper limit threshold is used as the first adjustment noise. If the increased first observed noise is less than the preset noise upper limit threshold, the increased first observed noise is used as the first adjustment noise.

[0139] Similarly, the mobile device can add the second observation noise to obtain the second adjustment noise in the manner of adding the first observation noise to obtain the first adjustment noise:

[0140] The mobile device may increase the second observed noise by a preset noise growth step to obtain an increased second observed noise, obtain a second adjusted noise based on the increased second observed noise, and use the obtained second adjusted noise as the adjusted second observed noise.

[0141] For example, the noise growth step can be set to 0.1f, where f is the value of the observation noise of the second sensor before adjustment. Then, the value of the observation noise of the second sensor after the increase is: f+0.1f, and f+0.1f is used as the adjusted second observation noise. The noise growth step represents the step size of each increase in the observation noise. By setting different noise growth steps, the sensitivity of increasing the observation noise can be adjusted. Different noise growth steps can be selected according to different actual conditions, which can improve the flexibility of noise adjustment.

[0142] Furthermore, a noise upper limit threshold may be set to set the maximum value of the observed noise. The mobile device may obtain the second adjusted noise based on the increased second observed noise in the following manner:

[0143] If the increased second observed noise is greater than or equal to the preset noise upper limit threshold, the preset noise upper limit threshold is used as the second adjusted noise. If the increased second observed noise is less than the preset noise upper limit threshold, the increased second observed noise is used as the second adjusted noise.

[0144] Specifically, the mobile device can obtain the first adjusted noise according to the following expression: R=min(MAX_R, f+0.1f), where R is the first observed noise and MAX_R is the preset noise upper limit threshold. Similarly, the second adjusted noise can also be obtained according to the above expression, and when the second adjusted noise is obtained, R is the second observed noise.

[0145] In this way, by setting the upper noise threshold of the observation noise, the adjustment range of the observation noise is limited so that the observation noise will not be too large, thereby reducing the situation that causes instability in the physical parameter estimation method and improving the stability of physical parameter estimation in mobile devices.

[0146] Step S203: reducing the first observation noise and the second observation noise to obtain a third adjusted noise of the first sensor and a fourth adjusted noise of the second sensor.

[0147] The following describes a method of reducing the first observation noise to obtain the third adjustment noise.

[0148] The mobile device can reduce the first observed noise by a preset noise reduction step to obtain a reduced first observed noise, obtain a third adjusted noise based on the reduced first observed noise, and use the obtained third adjusted noise as the adjusted first observed noise.

[0149] For example, the noise reduction step size can be set to 0.05f, where f is the value of the observation noise of the first sensor before adjustment. Then, the value of the observation noise of the first sensor after reduction is: f-0.05f, and f-0.05f is used as the adjusted first observation noise. The noise reduction step size represents the step size of each increase in the observation noise. By setting different noise reduction step sizes, the sensitivity of reducing the observation noise can be adjusted. Different noise reduction step sizes can be selected according to different actual conditions, which can improve the flexibility of noise adjustment.

[0150] Furthermore, a noise lower limit threshold may be set to set a minimum value of the observed noise. The mobile device may obtain the third adjusted noise based on the reduced first observed noise in the following manner:

[0151] If the reduced first observed noise is greater than or equal to the preset noise upper limit threshold, the preset noise upper limit threshold is used as the third adjusted noise. If the reduced first observed noise is less than the preset noise upper limit threshold, the reduced first observed noise is used as the third adjusted noise.

[0152] Similarly, the mobile device can reduce the second observation noise to obtain the fourth adjustment noise in the manner of reducing the first observation noise to obtain the third adjustment noise:

[0153] The mobile device can reduce the second observed noise by a preset noise reduction step to obtain a reduced second observed noise, obtain a fourth adjusted noise based on the reduced second observed noise, and use the obtained fourth adjusted noise as the adjusted second observed noise.

[0154] For example, the noise reduction step can be set to 0.05f, where f is the value of the observation noise of the second sensor before adjustment. Then, the value of the observation noise of the second sensor after reduction is: f-0.05f, and f-0.05f is used as the adjusted second observation noise.

[0155] Furthermore, a noise lower limit threshold may be set to set a minimum value of the observed noise. The mobile device may obtain the fourth adjusted noise based on the reduced second observed noise in the following manner:

[0156] If the reduced second observed noise is greater than or equal to the preset noise lower limit threshold, the reduced second observed noise is used as the fourth adjusted noise. If the reduced second observed noise is less than the preset noise lower limit threshold, the preset noise lower limit threshold is used as the fourth adjusted noise.

[0157] The mobile device can obtain the third adjusted noise according to the following expression: R=max(MIN_R, f-0.05f), where R is the first observed noise and MIN_R is the preset noise lower limit threshold. Similarly, the fourth adjusted noise can also be obtained according to the above expression, and when the fourth adjusted noise is obtained, R is the second observed noise.

[0158] In this way, by setting the noise lower limit threshold of the observation noise, the adjustment range of the observation noise is limited so that the observation noise will not be too small, thereby reducing the situation that causes instability in the physical parameter estimation method and improving the stability of physical parameter estimation in mobile devices.

[0159] As can be seen from the above, according to the consistency of the readings of the first sensor and the second sensor, that is, according to the difference between the first observation value and the second observation value, the uncertainty of the data collected by the first sensor and the data collected by the second sensor can be determined by the obtained normalized error. When the difference between the readings of the first sensor and the second sensor is too large, it means that the first sensor and the second sensor may be faulty, interfered or have low accuracy. In other words, the uncertainty of the data collected by the first sensor and the data collected by the second sensor is high and the reliability is low. The observation noise of the first sensor and the observation noise of the second sensor can be increased, thereby reducing the weight of the first observation value and the second observation value in the estimation of physical parameters. On the contrary, when the difference between the readings of the first sensor and the second sensor is small, it means that the credibility of the data collected by the first sensor and the second sensor is high. In other words, the uncertainty of the data collected by the first sensor and the data collected by the second sensor is low and the reliability is high. The observation noise of the first sensor and the observation noise of the second sensor can be reduced, thereby increasing the weight of the first observation value and the second observation value in the estimation of physical parameters. By comparing the normalized error with the preset error threshold, the uncertainty and reliability of the data of the first sensor and the second sensor are determined. In this way, according to the real-time changes of the data collected by the sensor, the observation noise is adaptively and dynamically adjusted, which can increase the influence of the observation value of the sensor with low uncertainty, reduce the influence of the observation value of the sensor with high uncertainty, and improve the accuracy of the first predicted value of the obtained target physical parameter. Based on such a physical parameter estimation method, the mobile device can adapt to different working environments and operating conditions.

[0160] In one embodiment of the present invention, the preset error threshold is: the preset error threshold is determined in the following manner: the preset error threshold is determined from a preset chi-square distribution table according to preset degrees of freedom and preset significance level values.

[0161] The preset error threshold is a statistical threshold, which represents the probability that there is a significant difference between the observed values ​​of the target physical parameter obtained by the data collected by different sensors at a preset confidence level. For example, for the assumption that "when the normalized error between the observed values ​​of the target parameter obtained by the data collected by two different sensors is greater than or equal to the preset error threshold, there is a significant difference between the two observed values", the degree of freedom can be determined to be 1. For another example, the preset confidence level can be: 95%. For a chi-square distribution with a degree of freedom of 1, the corresponding preset error threshold of 95% can be found from the preset chi-square distribution table as: 3.8415.

[0162] In this way, it is possible to accurately determine whether there are significant differences between multiple sensor readings, that is, to determine whether there is a significant difference in the error between the first observation value and the second observation value. The observation noise of each sensor can be adjusted more accurately to improve the accuracy of the first predicted value of the target physical parameter.

[0163] In one embodiment of the present invention, the observation noise of each sensor is represented by an observation noise matrix.

[0164] Among them, the observation noise matrix is ​​a square matrix, the order of the observation noise matrix is ​​the same as the number of sensors, and each diagonal element in the observation noise matrix corresponds to the observation noise of each sensor.

[0165] The above step S104 can adjust the observation noise of the first sensor and the second sensor in the following manner: adjust the values ​​of the diagonal elements corresponding to the first sensor in the observation noise matrix, and adjust the values ​​of the diagonal elements corresponding to the second sensor in the observation noise matrix.

[0166] In one implementation, when increasing the observation noise of the first sensor and the observation noise of the second sensor, the mobile device may increase the values ​​of the diagonal elements corresponding to the first sensor in the observation noise matrix, and increase the values ​​of the diagonal elements corresponding to the second sensor in the observation noise matrix. When reducing the observation noise of the first sensor and the observation noise of the second sensor, the values ​​of the diagonal elements corresponding to the first sensor in the observation noise matrix are reduced, and the values ​​of the diagonal elements corresponding to the second sensor in the observation noise matrix are reduced.

[0167] For example, if the number of sensors is 3, then the observation noise matrix is ​​a 3×3 square matrix. Assuming that the observation noise A of sensor A is 0.1, the observation noise B of sensor B is 0.5, and the observation noise C of sensor C is 0.3, the observation noise matrix can be expressed as follows: For example, if the first sensor and the second sensor are sensor A and sensor B respectively, when adjusting the first observation noise of the first sensor, the elements in the above observation noise matrix are When the second observation noise of the second sensor is adjusted, the elements in the above observation noise matrix are adjusted. Make adjustments.

[0168] In addition, if there is correlation between the observation noises of various sensors, then when adjusting the observation noise, elements other than the diagonal elements in the above observation noise matrix may also be targeted.

[0169] In this way, by using the observation noise matrix to represent the noise of each sensor, the observation noise corresponding to each sensor can be adjusted more conveniently by adjusting the diagonal elements in the observation noise matrix, thereby improving the flexibility of adjusting the observation noise.

[0170] In one embodiment of the present invention, see Figure 3 A flow chart of another method for estimating physical parameters based on a sensor is provided, wherein the method comprises the following steps S301-S307, and the above step S104 can be implemented by the following step S302:

[0171] Step S301: predicting a value of a target physical parameter of the mobile device based on data collected by a sensor disposed on the mobile device to obtain a second predicted value of the target physical parameter.

[0172] For different types of target physical parameters, different prediction methods may be used to predict the second predicted value of the target physical parameter. The following description will be made by taking the target physical parameter as mileage or heading angle as an example.

[0173] Case 1: The target physical parameter is mileage

[0174] The mobile device can predict the mileage of the mobile device based on the first predicted value of the mileage of the previous frame obtained by fusing data collected by the sensor loaded on the mobile device of the previous frame, and obtain a second predicted value of the mileage.

[0175] Specifically, the mobile device may calculate the product of a preset state transfer matrix and a first predicted value of the mileage of the previous frame as the second predicted value of the mileage.

[0176] Case 2: The target physical parameter is the heading angle

[0177] The mobile device predicts the heading angle change according to the roll angle, the pitch angle, the angular velocity of the roll angle and the angular velocity of the pitch angle collected by the sensor on the mobile device. The historical heading angle is updated based on the predicted heading angle change to obtain a second predicted value of the heading angle.

[0178] The mobile device can obtain the roll angle based on the roll angle acquisition system of the mobile device, and can obtain the pitch angle based on the roll angle acquisition system of the mobile device. The angular velocity of the roll angle and the angular velocity of the pitch angle can be obtained using a gyroscope mounted on the mobile device. The historical heading angle may include a first predicted value of the heading angle of the previous frame.

[0179] In this way, the heading angle prediction can be performed without directly using the heading angle and the angular velocity of the heading angle. A different method than using sensor data to obtain the heading angle can be used to predict the heading angle, thereby reducing the impact of the sensor data on the second predicted value. This can improve the accuracy of the second predicted value obtained, and further improve the accuracy of the first predicted value of the heading angle obtained.

[0180] Specifically, the mobile device predicts the heading angle change according to the following expression:

[0181] ;

[0182] in, is the heading angle change, is the roll angle, is the pitch angle, is the angular velocity of the pitch angle, is the angular velocity of the roll angle.

[0183] Then, after obtaining the heading angle change, the mobile device may add the obtained heading angle change to the historical heading angle to obtain a second predicted value of the heading angle.

[0184] In this way, using the above expression to obtain the heading angle change and then predicting the second predicted value of the heading angle can further improve the accuracy of the obtained second predicted value.

[0185] Further, after obtaining the second predicted value of the heading angle, the second predicted value may be normalized to limit the second predicted value to the range of [-π, π]. For example, a preset normalization function may be used to implement the normalization of the second predicted value.

[0186] In addition, the data collected by the sensor on the mobile device used to obtain the second predicted value may be data of a different type from the data used to obtain the observed value of the target physical parameter. For example, when obtaining the observed value of the heading angle, the mobile device may obtain the observed value of the heading angle of the mobile device through the angular velocity of the heading angle collected by the gyroscope, and when obtaining the second predicted value of the heading angle of the mobile device, the data used may include: the roll angle, the pitch angle, the angular velocity of the roll angle collected by the sensor mounted on the mobile device, and the angular velocity of the pitch angle, which does not include the angular velocity of the heading angle.

[0187] Step S302: Determine the priori state covariance based on the posterior state covariance and the process noise covariance.

[0188] Specifically, the mobile device may determine the prior state covariance of the current frame based on the a priori state covariance of the previous frame obtained by the last physical parameter estimation and the preset process noise covariance. The mobile device may determine the prior state covariance of the current frame according to the following expression:

[0189] ;

[0190] in, The current frame is the a priori estimated covariance, is the preset state covariance matrix, is the posterior estimated covariance of the previous frame, is the transposed matrix of the preset state covariance matrix, is the preset process noise covariance matrix.

[0191] Step S303: Acquire a first observation value of a first sensor and a second observation value of a second sensor provided on the mobile device.

[0192] Step S304: Calculate the difference between the first observation value and the second observation value; obtain a normalized error according to the difference and the observation noise of the target sensor, and adjust the first observation noise and the second observation noise according to the normalized error.

[0193] The target sensor is: the first sensor or the second sensor.

[0194] Steps S303-S304 are the same as the above steps S101-S104 and will not be described in detail here.

[0195] Step S305: for each sensor, determine the gain coefficient of the sensor based on the adjusted observation noise of the sensor and the priori state covariance.

[0196] Among them, the gain coefficient of the sensor represents the degree of change of the noise uncertainty of the observation value obtained based on the data collected by the sensor. If the value of the gain coefficient of the sensor is large, the degree of change of the noise uncertainty of the observation value obtained based on the data collected by the sensor becomes smaller, then the credibility of the observation value obtained based on the data collected by the sensor is higher, and the credibility of the second prediction value is lower. On the contrary, if the value of the gain coefficient of the sensor is large, the degree of change of the noise uncertainty of the observation value obtained based on the data collected by the sensor becomes larger, then the credibility of the observation value obtained based on the data collected by the sensor is lower, and the credibility of the second prediction value is higher.

[0197] Specifically, the mobile device may determine the gain coefficient of the current frame based on the adjusted observation noises and the prior state covariance of the current frame in the following manner:

[0198] ;

[0199] in is the gain coefficient of the current frame, The current frame is the a priori estimated covariance, is the identity matrix, is the transposed matrix of the identity matrix, is the adjusted observation noise of the current frame.

[0200] Step S306: Determine an updated a posteriori state covariance based on the a priori state covariance and the determined gain coefficients of each sensor.

[0201] Among them, the updated posterior state covariance can be the posterior state covariance of the current frame

[0202] ;

[0203] in, is the posterior state covariance of the current frame, is the identity matrix.

[0204] Step S307: Estimate a first predicted value of the target physical parameter based on the determined gain coefficients of the sensors and the measurement residuals between the observed values ​​and the second predicted values.

[0205] Specifically, the mobile device may subtract the second predicted value from the observed value of each sensor to obtain the measurement residual between the observed value of the sensor and the second predicted value, and then multiply the gain coefficient of each sensor in matrix form by the measurement residual of each sensor in matrix form to obtain the first predicted value

[0206] In step S307, the first predicted value of the target physical parameter of the mobile device can be obtained according to the following expression:

[0207] ;

[0208] in, is the second predicted value of the target physical parameter of the current frame, is the observed value of the target physical parameter of each sensor in the current frame, is the measured residual between each observed value and the second predicted value.

[0209] Step S307 is the same as the above step S105 and will not be described in detail here.

[0210] In addition, the first predicted value, the second predicted value, and the observed value may be the actual value of the target physical parameter, or the value change of the target physical parameter. For example, when the target physical parameter is the heading angle, the first predicted value calculated by the mobile device may be: the heading angle change. In this case, the mobile device may update the actual value of the heading angle finally obtained in the previous frame based on the first predicted value of the current frame to obtain the actual value of the heading angle of the current frame. For example, the update may be performed according to the following expression:

[0211] ;

[0212] in, is the first prediction value of the current frame, is the actual value of the heading angle obtained in the previous frame. The actual value of the heading angle of the current frame.

[0213] In summary, combining the influence of the uncertainty of the second predicted value and the uncertainty of the observed value to reduce the uncertainty of the second predicted value and the uncertainty of the observed value, comprehensively considering the second predicted value and the observed value, and re-determining the posterior state covariance can reduce the uncertainty of state estimation, improve the accuracy of the subsequent calculation of the gain coefficient that characterizes the degree of change of the noise uncertainty of the observed value, and thereby improve the accuracy of the first predicted value of the target physical parameter.

[0214] In this way, the use of adaptive Kalman filtering technology can dynamically adjust the sensor's observation noise, adapt to different environmental changes and different sensor qualities, reduce the error accumulation of a single sensor, and improve the accuracy of the first predicted value of the estimated target physical parameter. In addition, by using adaptive Kalman filtering technology to fuse data collected by multiple sensors, the data collected by multiple sensors can be effectively combined to estimate physical parameters, reducing the computational bottleneck that may be encountered in traditional methods, and supporting more efficient and faster target physical parameter estimation, thereby improving the efficiency of target physical parameter estimation.

[0215] In addition, before step S101, if the initialization setting has not been performed, the initialization setting needs to be performed. For example, when the physical parameter estimation is performed for the first time, the initialization setting needs to be performed. If the initialization setting has been performed when the physical parameter estimation is performed subsequently, the above step S101 can be executed directly. For example, for the case one of the above step S101: when the target physical parameter is mileage, the initialization setting may include setting the initial mileage and setting other parameters. For the case two of the above step S101: when the target physical parameter is the heading angle, the initialization setting may include setting the initial heading angle and setting other parameters. Among them, setting other parameters may include setting the initial posterior estimation covariance, the unit matrix H, the initial observation noise of each sensor, and the process noise covariance, etc.

[0216] For example, see Figure 4 A flow chart of another method for estimating physical parameters based on a sensor is provided, wherein the method comprises the following steps S401-S408:

[0217] Step S401: Determine whether initialization settings have been performed.

[0218] If the judgment is yes, execute step S403, if the judgment is no, execute step S402.

[0219] Step S402: Perform initialization settings.

[0220] After executing step S402, execute step S403.

[0221] Step S403: Acquire a first observation value of a first sensor and a second observation value of a second sensor provided on the mobile device.

[0222] Step S404: Calculate the difference between the first observation value and the second observation value.

[0223] Step S405: Obtain a normalized error according to the difference and the observation noise of the target sensor.

[0224] Step S406: adjusting the first observation noise of the first sensor and the second observation noise of the second sensor according to the normalized error.

[0225] Step S407: Estimate a first predicted value of the target physical parameter according to the first observation noise and the second observation noise.

[0226] The above steps S403-S407 are the same as steps S101-S105 and will not be described in detail here.

[0227] Corresponding to the above-mentioned physical parameter estimation method based on sensors, an embodiment of the present invention further provides a physical parameter estimation device based on sensors. Figure 5 A schematic diagram of the structure of a sensor-based physical parameter estimation device is provided, the device comprising:

[0228] The observation value acquisition module 501 is used to acquire a first observation value of a first sensor and a second observation value of a second sensor provided on the mobile device; wherein the first observation value is: an observation value of a target physical parameter obtained based on data collected by the first sensor, and the second observation value is: an observation value of the target physical parameter obtained based on data collected by the second sensor;

[0229] A difference calculation module 502, configured to calculate a difference between the first observation value and the second observation value;

[0230] A normalized error acquisition module 503 is used to acquire a normalized error according to the difference and the observation noise of the target sensor, wherein the target sensor is: the first sensor or the second sensor;

[0231] An observation noise adjustment module 504, configured to adjust the first observation noise and the second observation noise according to the normalized error;

[0232] The numerical estimation module 505 is used to estimate a first predicted value of a target physical parameter according to the first observation noise and the second observation noise.

[0233] In the sensor-based physical parameter estimation method provided by the embodiment of the present invention, the observation noise of each sensor can characterize the noise with uncertainty contained in the data collected by the sensor. By adjusting the observation noise of each sensor, the gain coefficient of each sensor can be adjusted, and the weight of the observation value of each sensor in the physical parameter estimation can be indirectly adjusted. According to the normalized error obtained by the difference between the first observation value and the second observation value and the observation noise of the target sensor, the uncertainty of the data collected by the first sensor and the second sensor can be determined. Therefore, adjusting the observation noise according to the above normalized error can be regarded as adjusting the observation noise according to the uncertainty of the data collected by the sensor. In this way, the weight of the observation value of each sensor in the physical parameter estimation can be flexibly adjusted according to the uncertainty of the data collected by the sensor, thereby reducing the influence of the data collected by the sensor with high uncertainty on the accuracy of the parameter estimation, improving the influence of the data collected by the sensor with low uncertainty on the accuracy of the parameter estimation, and improving the accuracy of the first predicted value of the estimated target physical parameter.

[0234] In one embodiment of the present invention, the observation noise adjustment module is specifically used to: if the normalized error is greater than a preset error threshold, increase the first observation noise and the second observation noise to obtain a first adjusted noise of the first sensor and a second adjusted noise of the second sensor; if the normalized error is less than or equal to the preset error threshold, reduce the first observation noise and the second observation noise to obtain a third adjusted noise of the first sensor and a fourth adjusted noise of the second sensor.

[0235] As can be seen from the above, according to the consistency of the readings of the first sensor and the second sensor, that is, according to the difference between the first observation value and the second observation value, the uncertainty of the data collected by the first sensor and the data collected by the second sensor can be determined by the obtained normalized error. When the difference between the readings of the first sensor and the second sensor is too large, it means that the first sensor and the second sensor may be faulty, interfered or have low accuracy. In other words, the uncertainty of the data collected by the first sensor and the data collected by the second sensor is high and the reliability is low. The observation noise of the first sensor and the observation noise of the second sensor can be increased, thereby reducing the weight of the first observation value and the second observation value in the estimation of physical parameters. On the contrary, when the difference between the readings of the first sensor and the second sensor is small, it means that the credibility of the data collected by the first sensor and the second sensor is high. In other words, the uncertainty of the data collected by the first sensor and the data collected by the second sensor is low and the reliability is high. The observation noise of the first sensor and the observation noise of the second sensor can be reduced, thereby increasing the weight of the first observation value and the second observation value in the estimation of physical parameters. By comparing the normalized error with the preset error threshold, the uncertainty and reliability of the data of the first sensor and the second sensor are determined. In this way, according to the real-time changes of the data collected by the sensor, the observation noise is adaptively and dynamically adjusted, which can increase the influence of the observation value of the sensor with low uncertainty, reduce the influence of the observation value of the sensor with high uncertainty, and improve the accuracy of the first predicted value of the obtained target physical parameter. Based on such a physical parameter estimation method, the mobile device can adapt to different working environments and operating conditions.

[0236] In one embodiment of the present invention, the observation noise adjustment module is specifically used to: increase the first observation noise by a preset noise growth step to obtain an increased first observation noise; obtain a first adjustment noise based on the increased first observation noise; increase the second observation noise by a preset noise growth step to obtain an increased second observation noise; and obtain a second adjustment noise based on the increased second observation noise.

[0237] The noise growth step size represents the step size of each increase in the observed noise. By setting different noise growth step sizes, the sensitivity of increasing the observed noise can be adjusted. Different noise growth step sizes can be selected according to different actual conditions, which can improve the flexibility of noise adjustment.

[0238] In one embodiment of the present invention, the observation noise adjustment module is specifically used to: if the increased first observation noise is greater than or equal to a preset noise upper limit threshold, then use the preset noise upper limit threshold as the first adjustment noise; if the increased first observation noise is less than the preset noise upper limit threshold, then use the increased first observation noise as the first adjustment noise.

[0239] In this way, by setting the upper noise threshold of the observation noise, the adjustment range of the observation noise is limited so that the observation noise will not be too large, thereby reducing the situation that causes instability in the physical parameter estimation method and improving the stability of physical parameter estimation in mobile devices.

[0240] In one embodiment of the present invention, the preset error threshold is determined in the following manner: the preset error threshold is determined from a preset chi-square distribution table according to preset degrees of freedom and a preset significance level value.

[0241] In this way, it is possible to accurately determine whether there are significant differences between multiple sensor readings, that is, to determine whether there is a significant difference in the error between the first observation value and the second observation value. The observation noise of each sensor can be adjusted more accurately to improve the accuracy of the first predicted value of the target physical parameter.

[0242] In one embodiment of the present invention, the observation noise of the sensor is represented by an observation noise matrix, wherein the observation noise matrix is ​​a square matrix, the order of the observation noise matrix is ​​the same as the number of sensors arranged on the mobile device, and each diagonal element in the observation noise matrix corresponds to the observation noise of each sensor;

[0243] The observation noise adjustment module is specifically used to: increase the values ​​of the diagonal elements corresponding to the first sensor in the observation noise matrix, and increase the values ​​of the diagonal elements corresponding to the second sensor in the observation noise matrix; reduce the values ​​of the diagonal elements corresponding to the first sensor in the observation noise matrix, and reduce the values ​​of the diagonal elements corresponding to the second sensor in the observation noise matrix.

[0244] In this way, by using the observation noise matrix to represent the noise of each sensor, the observation noise corresponding to each sensor can be adjusted more conveniently by adjusting the diagonal elements in the observation noise matrix, thereby improving the flexibility of adjusting the observation noise.

[0245] In one embodiment of the present invention, the observation noise adjustment module is specifically used to determine a gap characterization value between the first observation value and the second observation value based on the difference; determine a normalization basis based on the observation noise of the target sensor; and obtain a normalized error based on the gap characterization value and the normalization basis.

[0246] In this way, the normalized error between the first observation value and the second observation value can be calculated more accurately to more accurately determine whether there is a significant difference between the first observation value of the first sensor and the second observation value of the second sensor, thereby more accurately adjusting the observation noise of the sensor and improving the accuracy of the target physical parameter estimation.

[0247] In one embodiment of the present invention, the device further comprises:

[0248] A parameter prediction module, configured to predict a target physical parameter of the mobile device based on data collected by a sensor disposed on the mobile device, and obtain a second predicted value of the target physical parameter;

[0249] A covariance determination module, used for determining a priori state covariance based on a posterior state covariance and a process noise covariance;

[0250] A gain coefficient determination module, configured to determine, for each sensor, a gain coefficient of the sensor based on the adjusted observation noise of the sensor and the prior state covariance, wherein the gain coefficient of the sensor represents a degree of change in noise uncertainty of an observation value obtained based on data collected by the sensor;

[0251] A covariance updating module, configured to determine an updated a posteriori state covariance based on the a priori state covariance and the determined gain coefficients of each sensor;

[0252] The observation noise adjustment module is specifically used to estimate the first predicted value of the target physical parameter based on the determined gain coefficients of each sensor and the measurement residuals between each observation value and the second predicted value.

[0253] In summary, combining the influence of the uncertainty of the second predicted value and the uncertainty of the observed value to reduce the uncertainty of the second predicted value and the uncertainty of the observed value, comprehensively considering the second predicted value and the observed value, and re-determining the posterior state covariance can reduce the uncertainty of state estimation, improve the accuracy of the subsequent calculation of the gain coefficient that characterizes the degree of change of the noise uncertainty of the observed value, and thereby improve the accuracy of the first predicted value of the target physical parameter.

[0254] In one embodiment of the present invention, the target physical parameter is: heading angle;

[0255] The parameter prediction module is specifically used to: predict the heading angle change according to the roll angle, pitch angle, angular velocity of the roll angle and angular velocity of the pitch angle collected by a sensor loaded on the mobile device; and obtain a second predicted value of the heading angle based on the predicted heading angle change.

[0256] In this way, the heading angle prediction can be performed without directly using the heading angle and the angular velocity of the heading angle. A different method than using sensor data to obtain the heading angle can be used to predict the heading angle, thereby reducing the impact of the sensor data on the second predicted value. This can improve the accuracy of the second predicted value obtained, and further improve the accuracy of the first predicted value of the heading angle obtained.

[0257] The embodiment of the present invention further provides an electronic device, such as Figure 6 As shown, it includes a processor 601 , a communication interface 602 , a memory 603 and a communication bus 604 , wherein the processor 601 , the communication interface 602 , and the memory 603 communicate with each other via the communication bus 604 .

[0258] Memory 603, used for storing computer programs;

[0259] The processor 601 is used to implement any of the above-mentioned sensor-based physical parameter estimation methods when executing the program stored in the memory 603.

[0260] The communication bus mentioned in the above electronic device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0261] The communication interface is used for communication between the above electronic device and other devices.

[0262] The memory may include a random access memory (RAM) or a non-volatile memory (NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0263] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0264] In another embodiment of the present invention, a computer-readable storage medium is provided, in which a computer program is stored. When the computer program is executed by a processor, any of the above-mentioned sensor-based physical parameter estimation methods is implemented.

[0265] In another embodiment of the present invention, a computer program product including instructions is provided, which, when executed on a computer, enables the computer to execute any of the above-mentioned sensor-based physical parameter estimation methods.

[0266] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive Solid State Disk (SSD)), etc.

[0267] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0268] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, electronic device, storage medium, and computer program product embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0269] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. A method for estimating physical parameters based on sensors, characterized in that: The method comprises: Acquire a first observation value of a first sensor and a second observation value of a second sensor provided on the mobile device; wherein the first observation value is: an observation value of a target physical parameter obtained based on data collected by the first sensor, and the second observation value is: an observation value of the target physical parameter obtained based on data collected by the second sensor; calculating a difference between the first observation and the second observation; Obtaining a normalized error according to the difference and an observation noise of a target sensor, wherein the target sensor is: the first sensor or the second sensor; According to the normalized error, adjusting the first observation noise and the second observation noise; A first predicted value of a target physical parameter is estimated based on the first observation noise and the second observation noise.

2. The method according to claim 1, characterized in that The adjusting the first observation noise and the second observation noise according to the normalized error includes: If the normalized error is greater than a preset error threshold, adding the first observation noise and the second observation noise to obtain a first adjusted noise of the first sensor and a second adjusted noise of the second sensor; If the normalized error is less than or equal to a preset error threshold, the first observation noise and the second observation noise are reduced to obtain a third adjusted noise of the first sensor and a fourth adjusted noise of the second sensor.

3. The method according to claim 2, characterized in that The adding the first observation noise and the second observation noise to obtain a first adjusted noise of the first sensor and a second adjusted noise of the second sensor includes: Increasing the first observed noise by a preset noise growth step to obtain an increased first observed noise; and obtaining a first adjusted noise based on the increased first observed noise; The second observed noise is increased by a preset noise growth step to obtain an increased second observed noise; and based on the increased second observed noise, a second adjusted noise is obtained.

4. The method according to claim 3, characterized in that The step of obtaining a first adjusted noise based on the increased first observed noise comprises: If the increased first observed noise is greater than or equal to the preset noise upper limit threshold, the preset noise upper limit threshold is used as the first adjusted noise; if the increased first observed noise is less than the preset noise upper limit threshold, the increased first observed noise is used as the first adjusted noise.

5. The method according to claim 2, characterized in that: The preset error threshold is determined in the following manner: the preset error threshold is determined from a preset chi-square distribution table according to preset degrees of freedom and preset significance level values.

6. The method according to claim 2, characterized in that The observation noise of the sensor is represented by an observation noise matrix, wherein the observation noise matrix is ​​a square matrix, the order of the observation noise matrix is ​​the same as the number of sensors arranged on the mobile device, and each diagonal element in the observation noise matrix corresponds to the observation noise of each sensor; The adding of the first observation noise and the second observation noise comprises: Increasing the values ​​of the diagonal elements in the observation noise matrix corresponding to the first sensor, and increasing the values ​​of the diagonal elements in the observation noise matrix corresponding to the second sensor; The reducing the first observation noise and the second observation noise comprises: The values ​​of the diagonal elements corresponding to the first sensor in the observation noise matrix are reduced, and the values ​​of the diagonal elements corresponding to the second sensor in the observation noise matrix are reduced.

7. The method according to claim 1, characterized in that The obtaining a normalized error according to the difference and the observation noise of the target sensor includes: Determine, according to the difference, a difference representation value between the first observation value and the second observation value; Determine a normalization basis according to the observation noise of the target sensor; A normalized error is obtained based on the gap characterization value and a normalized basis.

8. The method according to any one of claims 1 to 7, characterized in that The method further comprises: Predicting a target physical parameter of the mobile device based on data collected by a sensor disposed on the mobile device to obtain a second predicted value of the target physical parameter; Determine the prior state covariance based on the posterior state covariance and the process noise covariance; After adjusting the first observation noise and the second observation noise according to the normalized error, the method further includes: For each sensor, determine a gain coefficient of the sensor based on the adjusted observation noise of the sensor and the prior state covariance, wherein the gain coefficient of the sensor represents a degree of change in noise uncertainty of an observation value obtained based on data collected by the sensor; Determining an updated a posteriori state covariance based on the a priori state covariance and the determined gain coefficients of each sensor; The estimating a first predicted value of a target physical parameter according to the first observation noise and the second observation noise includes: A first predicted value of the target physical parameter is estimated based on the determined gain coefficients of the sensors and the measurement residuals between the observed values ​​and the second predicted value.

9. The method according to claim 8, characterized in that The target physical parameters are: heading angle; The step of predicting a target physical parameter of the mobile device based on data collected by a sensor disposed on the mobile device to obtain a second predicted value of the target physical parameter includes: Predict the heading angle change according to the roll angle, the pitch angle, the angular velocity of the roll angle and the angular velocity of the pitch angle collected by the sensor mounted on the mobile device; Based on the predicted heading angle change, a second predicted value of the heading angle is obtained.

10. A sensor-based physical parameter estimation device, characterized in that: The device comprises: An observation value acquisition module, used to acquire a first observation value of a first sensor and a second observation value of a second sensor provided on the mobile device; wherein the first observation value is: an observation value of a target physical parameter obtained based on data collected by the first sensor, and the second observation value is: an observation value of the target physical parameter obtained based on data collected by the second sensor; A difference calculation module, used for calculating the difference between the first observation value and the second observation value; a normalized error acquisition module, configured to acquire a normalized error according to the difference and an observation noise of a target sensor, wherein the target sensor is: the first sensor or the second sensor; An observation noise adjustment module, used for adjusting the first observation noise and the second observation noise according to the normalized error; A numerical estimation module is used to estimate a first predicted value of a target physical parameter based on the first observation noise and the second observation noise.

Images