Device and method for reducing noise of external measuring sensors of a moving object

Abstract

An apparatus and method for reducing noise of an external measurement sensor of a moving object, the apparatus for reducing noise includes a storage unit storing measurement values of an external measurement sensor of a moving object, the external measurement sensor including an ultrasonic sensor measuring a distance value between the moving object and an object outside the moving object; and an arithmetic unit generating a plurality of functions derivable from the measurement values stored in the storage unit, deriving a final function having the most measurement values in a first range among the plurality of functions, deriving an estimated value as an expected value of the measurement values through the final function, and selecting a candidate final value closest to the estimated value in a second range as a final value among actual measurement values at a time point corresponding to the estimated value.

Description

[0001] Cross-reference to related applications

[0002] This application claims priority to Korean Patent Application No. 10-2021-0170126, filed on December 1, 2021, the entire contents of which are incorporated herein by reference for all purposes. Technical Field

[0003] The present invention relates to an apparatus and method for reducing noise in external measurement sensors of moving objects, which additionally utilizes past data to reduce noise in current data measured in external measurement data of moving objects, and simultaneously outputs current data robust to abnormal signals by using a function and a range of the function obtained from past data measured at a immediately preceding time, and by using these, collision warning or braking devices of moving objects can be effectively controlled. Background Technology

[0004] External measurement sensors (such as ultrasonic sensors) for moving objects (vehicles, etc.) are used for collision warning or braking by detecting objects in the vehicle system. Traditional ultrasonic sensors detect the relative distance to an object by emitting ultrasonic signals (emitted waves) and determining whether the intensity of the signal reflected from the object (received waves) exceeds a certain threshold. The conventional ultrasonic sensor system then outputs the nearest distance value where the intensity of the received signal is greater than or equal to a threshold. If the output distance value is less than a certain distance, the collision warning system or emergency braking system is activated. Therefore, in real-world vehicle environments with various noises having frequencies similar to the emitted wave, there is a risk of outputting incorrect distance values ​​if noise signals exceeding a threshold are received simultaneously with the received wave. Therefore, it is necessary to archive common noise by removing various types of noise to achieve robust control over anomalous signals.

[0005] Specifically, the least mean square method has been disclosed as a method for predicting current data based on past data measured by an ultrasonic sensor. However, if anomalous signals from the past data are included, there is a fatal problem that the current data may be incorrectly predicted. Therefore, a technique is needed to solve this problem.

[0006] The matters described as background art are merely for enhancing the understanding of the background of the present invention and should not be construed as an admission that they correspond to prior art known to those skilled in the art. Summary of the Invention

[0007] The present invention is proposed to solve the above-mentioned problems, and the object of the present invention is to provide an apparatus and method for reducing noise of an external measurement sensor of a moving object, which additionally utilizes past data to reduce the noise of the current data measured in the external data of the moving object, and at the same time can output robust current data to abnormal signals by using a function and the range of the function obtained from past data measured from a immediately preceding time, and by using these, collision warning or braking devices of the moving object can be effectively controlled.

[0008] The device for reducing noise of an external measuring sensor of a moving object according to the present invention includes a storage unit and a processing unit. The storage unit stores measurement values ​​of the external measuring sensor of the moving object, the external measuring sensor including an ultrasonic sensor that measures the distance between the moving object and an object outside the moving object. The processing unit generates a plurality of functions that can be derived from the measurement values ​​stored in the storage unit, derives a final function with the most measurement values ​​in a first range from the plurality of functions, derives an estimate as an expected value of the measurement values ​​from the final function, and selects a candidate final value that is closest to the estimate as the final value from the actual measurement values ​​at the time point corresponding to the estimate in a second range.

[0009] The storage unit can store measurements taken by the external measuring sensors of the moving object at regular time intervals.

[0010] The computational unit can select as many measurement values ​​as a first number of elements from the stored measurement values, and can generate the plurality of functions that can be derived from the selected measurement values.

[0011] The computational unit can select two measurement values ​​from the stored measurement values ​​and can generate multiple linear functions that can be derived from the selected measurement values.

[0012] If the number of functions generated is greater than or equal to the first threshold number, the operation unit will no longer generate the multiple functions.

[0013] If one or more of the selected measurements are zero (0), the arithmetic unit may not generate the plurality of functions from the measurements.

[0014] If the absolute value of the slope of the function derived from any two of the selected measurements is greater than or equal to a first threshold, then the computational unit may not generate the plurality of functions from the measurements.

[0015] The computational unit can exclude functions whose number of measured values ​​is less than a first threshold from the final function, and can derive the final function with the most measured values ​​from functions whose number of measured values ​​is greater than or equal to the first threshold.

[0016] If there is no function among the plurality of functions in which the number of measured values ​​within the first range is greater than or equal to the second threshold, then the computation unit may not derive the estimated value.

[0017] If multiple final functions exist, the computational unit may derive a function having the maximum sum of values ​​obtained by multiplying the measured values ​​within the first range by a first weight, wherein the first weight increases as the measured values ​​are closer to the current time point.

[0018] When the measured value is closest to the current time point, the first weight can be the number of measured values ​​within the first range, and when the measured value is farthest from the current time point, the first weight can be one (1).

[0019] If multiple final functions exist, the computational unit can derive the function having the least squares sum of the residuals between the measured value and the corresponding function value as the final function.

[0020] If multiple final functions exist, the computation unit can derive the function having the maximum sum of values ​​obtained by multiplying the measured values ​​within the first range by a first weight, the first weight increasing as the measured values ​​are closer to the current time point, and if the sums are equal, the computation unit can derive the function having the least squares sum of the residuals between the measured values ​​and the corresponding function values ​​as the final function.

[0021] If at the time point corresponding to the estimated value, there exists an approximate final value that is an actual measured value whose residual with respect to the candidate final value is greater than or equal to a third threshold, then the computation unit may select the approximate final value as the final value.

[0022] If there exists an approximate final value whose residual with the candidate final value is greater than or equal to the third threshold, and simultaneously there exists a second approximate final value that is a measured value whose residual with the second candidate final value is greater than or equal to the third threshold, and among the actual measured values ​​at the time point immediately preceding the time point corresponding to the estimated value, the second candidate final value is the measured value within the second range that is closest to the estimated value at the time point immediately preceding the time point corresponding to the estimated value, then the calculation unit may select the approximate final value as the final value.

[0023] The noise reduction device may further include a noise flag unit that calculates the ratio of outliers among the measurements taken before a predetermined time point corresponding to the estimated value. If the ratio of outliers is greater than or equal to a first ratio, the calculation unit may select only the candidate final value as the final value. If the ratio of outliers is less than the first ratio, and if there is an approximate final value at the time point corresponding to the estimated value whose residual with the candidate final value is greater than or equal to a third threshold, the calculation unit may select the approximate final value as the final value. If there is no approximate final value, the calculation unit may select the candidate final value as the final value.

[0024] If there exists an approximate final value whose residual with the candidate final value is greater than or equal to the third threshold, and simultaneously there exists a second approximate final value that is a measurement whose residual with the second candidate final value is greater than or equal to the third threshold, then among the actual measurements at the time point immediately preceding the time point corresponding to the estimated value, the second candidate final value is the measurement closest to the estimated value at the time point immediately preceding the time point corresponding to the estimated value within the second range. If the absolute value of the slope of the straight line connecting the approximate final value and the second approximate final value at each time point is less than or equal to the second slope, then the calculation unit can select the approximate final value as the final value; otherwise, the calculation unit can select the candidate final value as the final value.

[0025] A method for reducing noise using an external measuring sensor of a moving object according to the present invention includes the following steps: storing measurement values ​​of the external measuring sensor of the moving object, the external measuring sensor including an ultrasonic sensor that measures the distance between the moving object and an object outside the moving object, via a storage unit; generating a plurality of functions that can be derived from the measurement values ​​stored in the storage unit via a processing unit; deriving a final function with the most measurement values ​​in a first range via the processing unit; deriving an estimated value as an expected value of the measurement values ​​via the final function via the processing unit; and selecting a candidate final value that is closest to the estimated value in a second range from actual measurement values ​​at a time point corresponding to the estimated value via the processing unit as the final value.

[0026] According to the device and method for reducing noise of the external measurement sensor of the moving object, past data is additionally used to reduce the noise of the current data measured in the external data of the moving object, and at the same time, the current data robust to abnormal signals can be output by using the function and range of the function obtained from the past data measured from a certain time immediately preceding, and the collision warning or braking device of the moving object can be effectively controlled by using these. Attached Figure Description

[0027] Figure 1 This is a block diagram of a device for reducing noise in an external measuring sensor of a moving object according to an embodiment of the present invention.

[0028] Figures 2 to 3 This is a diagram illustrating a device for reducing noise using an external measurement sensor of a moving object according to an embodiment of the present invention.

[0029] Figure 4 This is a flowchart of a method for reducing noise using an external measurement sensor of a moving object according to an embodiment of the present invention. Detailed Implementation

[0030] The descriptions of specific structures, functions, or features of embodiments of the invention disclosed in this specification or application are merely illustrative for the purpose of describing embodiments of the invention, and embodiments of the invention can be implemented in various different forms and should not be construed as limited to these embodiments. Hereinafter, embodiments of the invention will be described in detail with reference to the accompanying drawings.

[0031] Figure 1 This is a block diagram of a device for reducing noise of an external measuring sensor of a moving object according to an embodiment of the present invention; Figures 2 to 3 This is a diagram illustrating a noise reduction device employing an external measurement sensor of a moving object according to an embodiment of the present invention; and Figure 4 This is a flowchart of a method for reducing noise using an external measurement sensor of a moving object according to an embodiment of the present invention.

[0032] Figure 1 This is a block diagram of a noise reduction device for an external measuring sensor of a moving object according to an embodiment of the present invention. Here, the moving object refers to a vehicle and applications including tools for transportation (including translation, rotation, vibration, and combinations thereof) and tools for flight. The noise reduction device for the moving object is a control device that removes abnormal signals and detects normal signals from physical signals measured or sensed (hereinafter referred to as measurements, etc.) by an internal measuring device or an external measuring sensor of the moving object. In the case of the noise reduction device (N) for the external measuring sensor of the moving object according to an embodiment of the present invention, the external measuring sensor (S) is mounted outside the moving object to measure information about physical quantities or sense signals from the moving object and the natural world outside the moving object. A storage unit (M) stores the measurement values ​​measured or sensed by the external measuring sensor (S) of the moving object. An arithmetic unit (C) executes logic to remove abnormal signals from the measurement values ​​stored in the storage unit (M) and detect normal signals.

[0033] The normal signals detected by the processing unit (C) refer to information about external risks to the moving object. Specifically, the processing unit (C) continuously feeds back whether the final value corresponds to the setpoint, range, or error value of each device to be controlled, in order to ultimately control each device. Specifically, if the device is a drive device, braking device, or warning device for the moving object, it can be controlled to prepare for the risk of collision, etc. As an example, a collision warning system (W) or automatic emergency braking device (B) for the moving object receives the normal signals detected by the processing unit (C) and determines whether there is a risk of collision or emergency braking through predetermined processing. Next, the collision warning system (W) controls whether to activate warning devices, such as warning sounds, based on the risk of front-to-rear collisions or collisions with ground lines, stop lines, and curbs during driving. The automatic emergency braking device (B) controls whether to activate the braking system based on the risk of collision or emergency braking, such as when the driver of the moving object is drowsy and negligent in driving, or unable to control the driving (e.g., sudden acceleration of the moving object).

[0034] An embodiment of the present invention is a system for removing noise and detecting normal signals from measurements taken by an external measuring sensor of a moving object. Specifically, the signal measured by the external measuring sensor (S) can be the intensity measured in the x, y, and z directions of sound waves, including ultrasound, or light including infrared, visible light, laser, and radio waves, as a signal generated by the moving object itself or by internal and external devices installed in the moving object. Ultrasound refers to sound waves with respect to the shorter distance between the moving object and an obstacle outside the moving object, which can be measured at low cost using time-of-flight (TOF). Time-of-flight refers to the time until the ultrasound waves continuously or periodically generated by an ultrasonic generator installed in the moving object are reflected by the obstacle and received by the ultrasonic receiver of the moving object, and the measurement result is converted into a distance value. d is the time-of-flight (TOF), c is the velocity of ultrasound, and t is the transmission / reception time. Therefore, in the process of measuring time-of-flight, there is a high probability that the ultrasound generated by the ultrasonic generator of the moving object or generated in the external environment at short distances contains noise.

[0035] Specifically, embodiments of the present invention remove outlier signals (i.e., noise mixed with the current time measurement) and use past measurements prior to the current time for this purpose. Conventional systems with this principle remove noise by comparing an estimate predicted by the trend of past measurements with the current time measurement. In this case, conventional systems derive a function representing the trend of past measurements, and specifically utilize the least mean square (LMS) method as a means of deriving such a function.

[0036] The least mean square method is a method that derives a function whose squared residual between its measured value and function value is minimized. However, since the least mean square method is calculated by summing the least mean squares, i.e., the squared residual between the fault signal and the current signal, it may yield incorrect approximations if past measurements contain noise. In particular, since noise is calculated as the squared residual in all cases, regardless of whether the noise is discrete, sparse, or present within a certain period, given that there is a certain ratio or higher of noise compared to the normal signal, the resulting predictions have low accuracy despite the fast measurements. Collision warning systems (W) and automatic emergency braking systems (B) controlled in this way are prone to malfunction or over-operation, which may interfere with driving or cause risks such as collisions with other or other moving objects. Therefore, it is desirable to remove noise from past measurements. Accordingly, embodiments of the present invention derive a noise-robust final value by means of an estimate evaluated by removing noise from past measurements. In this way, control of the internal and external devices of a moving object can be performed quickly and accurately.

[0037] Figure 1 This is a block diagram of an apparatus for reducing noise of an external measurement sensor of a moving object according to an embodiment of the present invention, more specifically including a storage unit (M) and a processing unit (C). The storage unit (M) stores the measurement values ​​of the external measurement sensor (S) of the moving object; the processing unit (C) generates a plurality of functions that can be derived from the measurement values ​​stored in the storage unit (M), derives a final function with the most measurement values ​​in a first range from the plurality of functions, derives an estimate of the expected value of the measurement values ​​through the final function, and selects the candidate final value that is closest to the estimate in a second range from the actual measurement values ​​at the time point corresponding to the estimate as the final value.

[0038] In this context, the measurement by the external measuring sensor (S) of the moving object includes both the measurement and sensing of physical signals, such as determination. Measurement involves taking the signal itself as a measured value, while sensing involves taking the signal as electrical information (e.g., voltage), accompanied by some processing, such as converting this information into a measured value. Furthermore, the external measuring sensor (S) of the moving object itself, or the storage unit (M) of the noise reduction device, can be designed to measure or store only physical signals exceeding a certain threshold, rather than measuring or storing all physical signals. Moreover, the measured value can be a point, a vector, or a matrix using points or vectors, and depending on the measurement or storage method, the measured value can be continuous or discrete. The measurement or storage method can be configured differently depending on the characteristics of the corresponding moving object and the external measuring sensor, the type and nature of the measured value, the amount and reliability of normal signals and noise, and the speed and accuracy of the device to be controlled.

[0039] Furthermore, the functions derived from the arithmetic unit (C) are binary relations, where each element of one set (domain) matches a unique element (range) of another set (common domain). Here, functions include polynomials, rational functions, irrational functions, algebraic functions including power functions, periodic functions including trigonometric functions, discrete functions capable of having non-differentiable partitions, exponential functions, their analytic functions (such as Fourier series), and their composite and inverse functions. In addition, the domain of a function includes sets of real or complex numbers, the function itself, vector spaces, matrix spaces, etc. Furthermore, the type of function derived from the arithmetic unit (C) can be set differently depending on the characteristics of the corresponding moving object and external measuring sensors, the type and nature of the measured values, the quantity and reliability of normal signals and noise, and the speed and accuracy of the device to be controlled.

[0040] Importantly, unlike the least mean square method, embodiments of the present invention derive a final function for removing noise from current measurements by selecting the model with the most normal signals in past measurements. Furthermore, according to one embodiment, the trend of past measurements is tracked by including measurements within a first range for which the standard is set as the function value. Accordingly, in one embodiment, the derived final function accurately reflects the trend of normal signals while removing noise from past measurements (assuming there are more normal signals than noise). Through robust control based on the estimates and final values ​​predicted in this way, a collision warning system (W) or automatic emergency braking system (B) for moving objects can essentially prevent collisions.

[0041] Furthermore, according to embodiments of the present invention, even when noise is sporadically present in past measurements, or when noise is present in past measurements over a certain period of time, the final function can be derived by detecting normal signals. To overcome the shortcomings of the least mean square method, conventional systems may disclose techniques for identifying and removing noise from segments that are significantly different from normal signals when a segment contains noise. However, assuming that past measurements are discrete rather than continuous, conventional systems cannot determine noise because they cannot measure accumulated noise over a certain period. Furthermore, if this determination is applied to the assumed environment, normal signals in predetermined segments containing sporadic noise can be identified as noise. However, according to embodiments of the present invention, sporadic noise contained in past measurements can also be removed by selecting a model that includes the most normal signals. Accordingly, regardless of the type, shape, distribution, etc., of the noise, embodiments of the present invention can remove noise and thus provide more robust noise reduction capabilities than conventional systems.

[0042] Here, assuming the difference between the measured value and the corresponding function value (i.e., the residual) follows a normal distribution for the reliability of the control, the first range can be reasonably determined within a set of ranges having confidence intervals of 95% to 99%. That is, the first range can be set to any one of 1.96 to 2.58 times the square root of the value obtained by dividing the sum of the squares of the residuals by the number of measured values ​​(i.e., the standard deviation). Alternatively, the first range can be set to up to three times the standard deviation to include 99.9% of the normal signal for higher reliability.

[0043] More specifically, Figure 1 According to an embodiment of the present invention, the storage unit (M) of the noise reduction device (N) of the external measuring sensor of the moving object can store the measurement values ​​measured by the external measuring sensor (S) of the moving object at regular time intervals. That is, the storage unit (M) can store the measurement values ​​discretely rather than continuously.

[0044] One problem to be solved by embodiments of the present invention is the removal of noise generated by ultrasonic waves generated at short distances and those generated in the external environment when measuring time of flight (TOF) using an ultrasonic generator and receiver on a moving object. However, in this case, the processing speed of the arithmetic unit (S) may be very slow due to the excessive number of measurements when the storage unit (M) continuously stores all signals regardless of whether the signal is noise or normal. Furthermore, while the final function and final value derived from a large number of consecutive measurements may be ideal in terms of robustness and reliability, they may be less ideal in terms of derivation speed and the resulting control environment.

[0045] For example, assuming an obstacle suddenly and rapidly appears near a moving object, the collision warning system (W) and automatic emergency braking system (B) must be controlled more quickly to prevent the risk of collision and to execute emergency braking. Furthermore, when exposed to ambient noise for a sustained period, the collision warning system (W) and automatic emergency braking system (B) can perform more precise control operations using the final values ​​processed by measurements and operations at regular time intervals (rather than continuous measurements and operations). Accordingly, the storage unit (M) according to an embodiment of the invention reduces the computational load on the arithmetic unit (C) by storing measurements taken at regular time intervals, thereby enabling more stable and faster execution of the obtained control.

[0046] Figure 2 This is a diagram illustrating a noise reduction device using an external measurement sensor for a moving object according to an embodiment of the present invention. In the noise reduction device for an external measurement sensor for a moving object according to an embodiment of the present invention, the arithmetic unit can select a number of measurement values ​​from stored measurement values, as many as a first number of elements, and generate multiple functions that can be derived from the selected measurement values. In other words, the arithmetic unit can generate corresponding functions by combining a first number of elements from the multiple measurement values.

[0047] As previously mentioned, the type and nature of the function can be set differently depending on the characteristics of the moving object and the external measuring sensor, the type and nature of the measured values, the amount and reliability of normal signals and noise, and the speed and accuracy of the device to be controlled. However, considering that one objective of the embodiments of the present invention is to prevent the risk of collision with the moving object, it is necessary to control the device controlled by the final value of the noise reduction device more quickly. For this purpose, it is necessary to reduce the computational load of the arithmetic unit, and before the number of functions derived by the arithmetic unit, it is necessary to set the concept of a first number (hereinafter referred to as r) of the elements selected from all the measured values ​​(hereinafter referred to as n) from which such functions are derived.

[0048] Here, the total number of functions is nCr (C: the number of cases in which the first number of elements is selected from all the measurements regardless of order). However, according to Pascal's triangle and the binomial theorem, the sum of the combinations of selecting the first number of elements from zero (0) to n as the first number of elements is 2n. That is, if the first number of elements is not determined, the computational load of the arithmetic unit will increase exponentially with the number of all measurements. Accordingly, embodiments of the present invention reduce the computational load of the arithmetic unit to the hardware level, and correspondingly enable more stable and faster execution of control.

[0049] Furthermore, in the noise reduction device for the external measurement sensor of a moving object according to an embodiment of the present invention, the arithmetic unit can select two measurement values ​​from the stored measurement values ​​and generate multiple linear functions that can be derived from the selected measurement values. That is, by combining multiple measurement values, the arithmetic unit can generate corresponding functions, particularly functions of straight lines connecting two measurement values.

[0050] Figure 2 The linear function of the application of an external measuring sensor for a moving object for noise reduction according to an embodiment of the present invention is shown. Figure 2 In the diagram, the vertical axis indicates the same measurement value as TOF, and the horizontal axis indicates the measurement time. Specifically, it shows the case where an ultrasonic sensor, as one type of external measuring sensor that is a moving object, measures TOF at regular time intervals (e.g., 1 / 60 of a second) n times, and illustrates the measurement value when a total of six (6) measurements are performed. Here, it is assumed that... Figure 2 The two endpoints of the straight line (i.e., the measurement immediately preceding the previous point and the measurement at the furthest point) are selected based on the current time point. In an embodiment of the invention, the function connecting the two points can be plotted as a linear function. Here, to determine whether the remaining four unselected measurements are normal signals or noise signals, the range within the first range of the function can be set as a boundary line, such as a dashed line. After determining the points within the boundary line as normal signals and the other points as noise, the number of normal signals and the corresponding function are stored, and this process is repeated continuously. The maximum number of repetitions is 6C2, i.e., 15 times, and the corresponding number of linear functions will also be 15. Therefore, according to an embodiment of the invention, a linear function is derived, which is the final function with the most measurements.

[0051] like Figure 2 As shown, by setting the first number of elements r to two (2) and generating a linear function accordingly, the computational load of the computation unit can be further reduced by using the minimum number of functions (nC2) and a simple function (y=ax+b). In addition, the combination of the two points selected in the above embodiment and the linear function can reflect the arithmetic mean, that is, the average expected value of the remaining unselected points, thereby including the measured values ​​within the first range.

[0052] Furthermore, the linear function reflects the motion state over a finite time interval measured by sensors such as ultrasonic sensors, when the distance between the moving object and the obstacle is measured. The measurement time (TOF) between a sensor-equipped moving object and an obstacle can be so short as to be insufficient for purposes such as collision warning or emergency braking. Therefore, it is assumed that the relative velocity between the moving object and the obstacle will be constant over such a finite time interval; that is, constant-rate motion will be prevalent. Thus, based on general internal and external environments, noise contained in past measurements can be effectively removed using the arithmetic mean (which is the best representative value calculated based on the above assumptions) and its extended value.

[0053] on the other hand, Figure 2 This diagram illustrates a noise reduction device employing an external measurement sensor of a moving object according to an embodiment of the invention. If the number of times a function is generated is greater than or equal to a first threshold number, the processing unit may not generate more functions. Furthermore, if among a plurality of functions there exists a function in which the number of measured values ​​within a first range is less than the first threshold, the processing unit of the noise reduction device according to an embodiment of the invention excludes that function from the final function and derives the final function with the most measured values ​​from the functions in which the number of measured values ​​is greater than or equal to the first threshold. That is, the range of the final function is limited by restricting the number of functions to up to the first threshold number, or by using only functions in which the number of normal signals is greater than or equal to the first threshold.

[0054] As mentioned above, if we choose the same number of measures as the first number of elements, then the number of functions is... That is, the equation of n with degree (nr). For example, suppose r is two (2), then the total number of functions is The number of operations performed by the arithmetic unit can increase significantly depending on the number of measured values. Accordingly, in one embodiment of the invention, by limiting the number of functions to a first threshold number and deriving the final function within a finite number of functions, the device can be smoothly executed to handle urgent control needs.

[0055] in addition, Figure 2 This is a diagram illustrating a noise reduction device using an external measurement sensor of a moving object according to an embodiment of the invention, and showing that if at least one zero (0) is included in the selected measurement value, the arithmetic unit may not generate multiple functions from these measurement values. That is, if no past measurement values ​​exist, the arithmetic unit may not generate multiple functions from the measured measurement values.

[0056] In the noise reduction apparatus according to an embodiment of the present invention, the computational load generated in deriving the final function having the most measured values ​​within a first range is important in terms of computational speed. In particular, if it is determined that there is not a large amount of noise in the surrounding environment, faster control is required for a collision warning system (W) or an automatic emergency braking system (B) for moving objects. Therefore, an increase in computational load is undesirable in signal processing for obstacles around moving objects, where noise reduction is the objective. Therefore, the noise reduction apparatus according to an embodiment of the present invention reduces computational load by not generating multiple functions having measured values ​​of zero (0) due to ambient noise, and performs precise control through rapid signal processing.

[0057] Furthermore, if the absolute value of the slope of the function derived from any two of the selected measurements is greater than or equal to a first threshold, the computation unit may not generate multiple functions from that measurement. That is, if the measurement value is TOF, the computation unit may not generate multiple functions from measurements whose relative rates are too large.

[0058] The noise reduction device for an external measurement sensor of a moving object according to an embodiment of the present invention can process measurement values ​​related to obstacles outside the moving object. Specifically, the noise reduction device according to an embodiment of the present invention can process measurement values ​​measured by time-of-flight (TOF) of relative distance, relative velocity, or relative acceleration of obstacles around the moving object. Assuming the obstacle is close to the moving object (i.e., in the case of negative relative velocity), a constant relative velocity means that the obstacle is approaching at a constant speed, which is measured by the slope of the function. However, even considering the maximum output or maximum velocity of the moving object, if the absolute value of the corresponding slope is unreasonably large, the measurement value can be identified as noise. Therefore, the noise reduction device for an external measurement sensor of a moving object according to an embodiment of the present invention does not generate multiple functions when the absolute value of the slope of the function is greater than or equal to a first threshold, so that measurement values ​​with a large amount of noise are not calculated, thereby improving the calculation speed and enabling fast control.

[0059] As another example, if there exists a function among multiple functions whose number of measured values ​​within a first range is less than a first threshold, the arithmetic unit of the noise reduction apparatus according to an embodiment of the invention excludes that function from the final function, and the arithmetic unit can derive the function with the most measured values ​​among functions whose number of measured values ​​is greater than or equal to the first threshold. Generally, when selecting r elements from at least one of N functions (hereinafter referred to as N) as the first threshold number, the number of selections at a level where p, as the probability of extracting only a normal signal at least once, is 99.9% or higher, is preferred for the reliability of the control. Assuming independent trials, the probability of extracting a normal signal when extracting a measured value is the ratio of the number of normal signals to all measured values ​​(hereinafter referred to as a). Therefore, the probability of extracting only noise at least once is 1-a. r Furthermore, in independent trials, when iterating with N functions, the probability p of extracting only the normal signal at least once is 1-(1-a). r ) N Accordingly, such an N can be derived as greater than or equal to Let p be a natural number, and assume that p is greater than or equal to 0.99.

[0060] Furthermore, storing the measurements within a first range and all derived functions is inefficient in terms of computational complexity and speed for deriving the final function. Therefore, the number of measurements within the first range needs to be filtered within a reliable range. Specifically, if the selected measurements are close to each other, the function derived from the selected measurements is likely to deviate from the reliable number of measurements. This is true regardless of whether the selected measurements contain noise. Accordingly, in embodiments of the invention, the range of the final function is limited to functions in which the number of normal signals derived from such measurements is greater than or equal to a first threshold, thereby reducing the computational complexity to match the level of the hardware to be controlled, and consequently enabling more stable and faster control execution.

[0061] Furthermore, if there is no function among the multiple functions whose number of measured values ​​within the first range is greater than or equal to the second threshold, the computational unit according to an embodiment of the present invention may not derive an estimated value. In other words, the minimum ratio of the normal signal in the number of measured values ​​of the final function (hereinafter referred to as n for convenience) can be set as (second threshold / n).

[0062] like Figure 2As shown, the noise reduction device according to an embodiment of the invention derives an estimated value based on a final function with the most measured values, compares this estimated value with the current measured values, and selects the closest value as the final value for noise reduction. Therefore, if 50% or more of the measured values ​​stored in the storage unit are noise, the final function and the derived estimated value are unreliable even if many measured values ​​are included. In particular, because the medium is air, the measured values ​​of ultrasonic sensors are susceptible to temperature fluctuations or wind. Therefore, in environments with rapid temperature and wind flow, most of the measured values ​​may not be normal signals. Even in this case, selecting the final function and final value to control a collision warning system or automatic emergency braking system for a moving object is extremely dangerous. Accordingly, according to an embodiment of the invention, if the number of measured values ​​is less than a first threshold, the reliability and accuracy of the noise reduction device are improved by excluding these measured values ​​from the final function, thereby ensuring precise control.

[0063] On the other hand, in another embodiment of the invention, if multiple final functions exist, then Figure 2 The computational unit can derive a final function having a maximum sum of values ​​obtained by multiplying measured values ​​within a first range by a first weight, which increases as the measured values ​​are closer to the current time point. Furthermore, when the measured values ​​are closest to the current time point, the first weight can be the number of measured values ​​within the first range, and when the measured values ​​are farthest from the current time point, the first weight can be -1 (1). That is, the final function having a larger number of measured values ​​close to the current time point can be given priority.

[0064] like Figure 2As shown, the computing unit of the noise reduction device performs calculations on measurements stored at regular time intervals immediately preceding the current time point. Specifically, in another embodiment, the time-of-flight (TOF) of the ultrasonic waves from the moving object's ultrasonic sensor can be measured as a measurement value, and in this case, only two measurements are selected from the measurements to derive a final function and a final value using multiple linear functions. However, it is assumed that the linear function (especially in the case of y = ax + b) represents a constant rate of motion. That is, it is not assumed that the relative velocity between the moving object and the obstacle immediately preceding the current time point has accelerated or decelerated. Therefore, even if the relative velocity increases or decreases, the linear function selects the measurement value closest to a straight line, i.e., the estimated value and the final value derived under the assumption that the relative velocity is constant. However, there is a risk of loss of capability for vehicle dynamic control (VDC), traction control system (TCS), limited-slip differential (LSD), collision warning system, automatic emergency braking system, anti-braking system (ABS), etc., used to prevent malfunctions in the driving force and direction control of the moving object or collisions with obstacles. This is because even if the moving object becomes dangerous due to its sudden acceleration, its approach to an obstacle, or its sudden deceleration or change of direction due to its sudden braking (i.e., even if it suddenly approaches TOF), the final value under the assumption of constant speed motion is derived, so control may be executed a step too late.

[0065] Therefore, in another embodiment of the invention, a first weight, acting as a deceleration / acceleration factor, is multiplied to derive a function with the maximum sum as the final function. This other embodiment thus solves the control problem arising from the preference for deriving measurements with a shape similar to itself, such as polynomial functions like linear functions. That is, in another embodiment, it is considered that when further acceleration or deceleration occurs at the current time point, the TOF can decrease sharply, exceeding the estimated value as the expected value of the corresponding function immediately preceding the current time point. As a result, even when exposed to the risk of collision between a moving object and an obstacle, another embodiment of the invention can control the collision avoidance device in advance by selecting a precise final value.

[0066] In particular, in another embodiment of the invention, the first term of the first weight can be the number of measurements within a first range, and the last term of the first weight can be -(1). In this case, the first weight between them can be an arithmetic sequence with a common difference of -(1), or a geometric sequence with a common ratio of (1 / n) and the square root of n. The geometric sequence gives more weight to the measurements immediately preceding the current time point than the arithmetic sequence. In particular, the weights can be assigned differently depending on the sensitivity of the moving object to collisions with obstacles. However, in another embodiment of the invention, the predictability of the weights and the resulting final function is increased by setting the reference point to n and -(1), and a control system optimized for the moving object can be provided by flexibly setting the reference point according to the acceleration characteristics of the moving object and obstacles.

[0067] on the other hand, Figure 2 This diagram illustrates a noise reduction device using an external measurement sensor of a moving object according to an embodiment of the present invention. If multiple final functions exist, the computation unit can derive a final function having the least squares sum of the residuals between the measured values ​​and the corresponding function values. Furthermore, if multiple final functions exist, the computation unit can derive a function having the maximum sum of values ​​obtained by multiplying measured values ​​within a first range by a first weight, which increases as the measured values ​​approach the current time point. If the sums are the same, the computation unit can derive a final function having the least squares sum of the residuals between the measured values ​​and the corresponding function values. That is, the former embodiment determines the priority of the final function by combining it with an existing least mean square method, while the latter embodiment determines the priority of the final function by using another embodiment of the present invention as the priority and combining the conventional least mean square method with the latter priority.

[0068] In embodiments of the present invention, since the number of measured values ​​within the first range is the largest, a final function can be derived individually. However, since the number of measured values ​​is the same, multiple final functions can be derived. In the event of competition among multiple final functions to determine which function to derive, the computational unit can randomly determine the final function, or it can determine it through priority. However, the priority control method in embodiments of the present invention, by providing space for further noise removal, allows for faster and more accurate operation of collision warning systems for moving objects, etc.

[0069] Furthermore, in embodiments according to the invention, the function with the least squares sum of the residuals between the measured values ​​and the corresponding function values ​​can be derived as the optimal final function. As in the embodiments mentioned above, for fast control, the computational unit can limit the number of functions that can be derived from the measured values ​​to less than a first threshold number. This means that a final function with the most measured values ​​within a first range can be derived, but this may not be the case outside the reliability interval. In this case, even among multiple functions, it is necessary to derive the optimal final function. In this case, even if multiple final functions exist, they will be in a state where noise is almost completely removed. Therefore, in the competition among final functions, the Least Mean Square (LSM) method is used to select one of the final functions that accurately reflects the trend of past measured values. In particular, the linear least mean square method is operated algebraically (matrix) by synthesizing normal equations, and because it is easily computed using existing programming tools (such as MATLAB), it will perform the fastest and most accurate control. That is, according to an embodiment of the invention, instead of controlling to randomly select one of a plurality of final functions, the collision warning system (W) for moving objects can be operated more quickly and accurately by controlling to derive the function having the least squares sum of the residuals between the measured values ​​and the corresponding function values ​​as the optimal final function.

[0070] As another example, one embodiment of the invention can derive a final function having a function that is the maximum sum of values ​​obtained by multiplying measured values ​​within a first range by a first weight, which increases as the measured values ​​are closer to the current time point. If the sum is the same, a function having the least squares sum of the residuals between the measured values ​​and the corresponding function values ​​can be derived as the final function. When the measured values ​​are the time-of-flight (TOF) of ultrasound, the embodiment preferably derives a final function for predicting accelerated motion between a moving object and an obstacle, and if multiple functions exist, a least mean square method is used to perform filtering. Accordingly, embodiments of the invention prioritize the risk of collision between the moving object and the obstacle and place a final function that more accurately reflects the trend of past measured values ​​in a lower priority, thus allowing the controlled device to be operated according to its function and purpose.

[0071] On the other hand, in another embodiment of the invention, which converts the prioritization method described above into a weighted method, the computational unit can derive a function as the final function from functions whose number of measurements is greater than or equal to a first threshold. This function is obtained by multiplying the number of measurements within a first range by at least one parameter derived from those measurements. This parameter can be obtained by dividing the sum of values ​​obtained by multiplying the measurements within the first range by a first weight that increases as the measurements approach the current time point by the number of measurements within the first range, or by dividing the sum of the squares of the residuals between the measurements within the first range and the corresponding function values ​​by the number of measurements within the first range. That is, according to another embodiment of the invention, the noise reduction device for an external measurement sensor of a moving object derives a final function by a final function evaluation value obtained by multiplying all average indices, which are obtained by dividing the aforementioned evaluation indices (the cumulative sum of values ​​multiplied by the first weight, the cumulative sum of least squares values) by the number of measurements within the first range. Therefore, embodiments of the present invention derive a final function that more accurately reflects the trends of past measurements and the risk of collisions between moving objects and obstacles, thereby enabling the optimized operation of the device to be controlled according to its function and purpose.

[0072] Figure 3 This is a diagram illustrating a noise reduction device employing an external measurement sensor of a moving object according to an embodiment of the present invention. Figure 3 As shown, the noise reduction apparatus according to an embodiment of the present invention selects a candidate final value that is closest to the estimated value within a second range. However, if there exists an approximate final value among the actual measurements at the time point corresponding to the estimated value, and the residual between the measured value and the candidate final value is greater than or equal to a third threshold, the computation unit may select the approximate final value as the final value.

[0073] If the device to be controlled in the embodiments of the present invention is related to preventing collisions or decelerating the speed of a moving object, it is necessary to prioritize the measured value rather than the estimated value, even if the measured value is noise, in order to prevent risks in advance. Specifically, when the measured value is the time-of-flight (TOF) of ultrasound, it is necessary to detect obstacles that suddenly appear to approach a moving object in order to prevent the risk of collision in advance; this will show a measured value with a much smaller trend than past measured values. However, it is undesirable to identify noise from devices that continuously generate ultrasound without moving (such as vending machines and parking lot gates) as such obstacles. Accordingly, embodiments of the present invention prioritize only the measured values ​​whose residuals are closest to the estimated value within a second range and are greater than or equal to a third threshold, thereby detecting obstacles dynamically approaching a moving object in advance and effectively controlling the collision prevention device.

[0074] Here, the third threshold can be set to a value 20 times or greater than the diffraction limit of the wave source of the external measuring sensor of the moving object. Specifically, since the external ultrasonic sensor of the moving object generally has a frequency band of 40 kHz to 50 kHz and a wavelength of 6.5 mm to 8.5 mm, the third threshold should be at least twice the corresponding wavelength to distinguish noise from normal signals while taking resolution into account. Considering the range of differentiation and the noise ratio (within 10%), the third threshold can be derived at a level that can be used industrially in embodiments of the invention. Clearly, the third threshold can be further set according to the specifications of the external measuring sensor of the moving object, the type and wavelength of the signal, and the resolution improvement based on signal localization.

[0075] Figure 3 It shows Figure 2 The concept in the expansion of the chart, and Figure 3 The graph illustrates the measurement closest to the estimated value within a second range according to an embodiment of the invention. Here, the measurement value on the vertical axis represents the TOF of the external ultrasonic sensor of the moving object, and the horizontal axis represents the TOF measurement time. In this embodiment, the final function is derived from multiple linear functions derived by selecting two of the seven measurements immediately preceding the current time point, and the estimated value (○) at the current time point and the n final values ​​(●) immediately preceding the current time point are shown. Figure 3 In one embodiment, among the measurements at the time point corresponding to the estimated value (○), the closest measurement (◆ within the dashed line) among the measurements close to the second range (▲ and ◆ within the dashed line) is selected as the final value.

[0076] Meanwhile, in another embodiment of the invention, when approximate values ​​exist continuously, only the approximate final value can be derived as the final value. This approximate value is a measurement whose residual with the candidate final value is greater than or equal to a third threshold. As a result, obstacles closest to the moving object can be quickly identified when the object suddenly pops up (the time of flight (TOF) is smaller than the estimated value by more than the third threshold), and devices for responding to risks such as collisions can be quickly controlled.

[0077] also, Figure 3This is a diagram illustrating a noise reduction device using an external measurement sensor of a moving object according to an embodiment of the present invention. If there exists an approximate final value whose residual with respect to a candidate final value is greater than or equal to a third threshold, and there exists a second approximate value as a measured value whose residual with respect to a second candidate final value is greater than or equal to the third threshold, then among the actual measured values ​​at a time point immediately preceding the time point corresponding to the estimated value, the second candidate is the measured value closest to the estimated value at a time point immediately preceding the time point corresponding to the estimated value within a second range, and the arithmetic unit can select the approximate final value as the final value. In other words, according to an embodiment of the present invention, it is possible to check whether an object suddenly approaches or whether noise suddenly occurs by continuously acquired measured values.

[0078] A key technology of the noise reduction device according to embodiments of the present invention is the extraction of accurate measurement values ​​from noise and the extraction of highly reliable measurement values. As described above, the measurement value may be the Time-of-Flight (TOF) measured between a moving object and an external obstacle using an ultrasonic sensor. However, outputting an approximate final value simply because an approximate final value exists in the TOF measured at the current time point may mean that an accurate measurement value has been extracted from the noise, but it cannot guarantee that the approximate final value has high reliability. This is because it is impossible to distinguish whether it is a suddenly appearing object or a suddenly occurring noise. Therefore, the noise reduction device according to embodiments of the present invention prevents unnecessary emergency braking or collision warnings caused by sudden noise by selecting the approximate final value only when a second approximate final value exists. In summary, embodiments of the present invention only detect suddenly appearing objects and filter other noises through candidate final values, thereby providing more reliable measurement values.

[0079] Figure 3 This is a diagram illustrating a noise reduction device employing an external measurement sensor of a moving object according to an embodiment of the present invention. The noise reduction device further includes a noise flag unit that calculates the ratio of outliers among actual measurements taken within a predetermined time period before deriving an estimate. If the ratio of outliers calculated by the noise flag unit is greater than or equal to a first ratio, the calculation unit selects a candidate final value as the final value. If the ratio of outliers is less than the first ratio, the calculation unit selects the approximate final value as the final value only if an approximate final value exists; otherwise, the calculation unit selects a candidate final value as the final value. That is, in a noisy environment, candidate final values ​​have priority.

[0080] The noise reduction apparatus according to an embodiment of the present invention outputs robust current data to anomalous signals by utilizing a function and a range of functions obtained from past data measured at a time immediately preceding the measurement. Therefore, logic is required that generates the final function only in noisy environments where the external moving object sensor is present, and selects the candidate final value closest to the estimated value as the final value. That is, before executing control according to an embodiment of the present invention, it is necessary to perform a control step to determine whether the measured values ​​received within a predetermined time period contain an excessive number of outliers. Therefore, the noise reduction apparatus for an external measuring device of a moving object according to an embodiment of the present invention can perform precise control by selecting the candidate final value as the final value only if the ratio of outliers is greater than or equal to a first ratio.

[0081] Here, outliers are measurements that are greater than or equal to a first reference value (e.g., measurements exceeding three times the standard deviation of the mean) obtained by dividing the z-value (the value obtained by subtracting the average of measurements over a certain time period) by the standard deviation, or measurements that are greater than or equal to or less than the first reference value based on the upper and lower quartiles. Outliers can thus be distinguished from normal values. Furthermore, outliers can be values ​​filtered by frequency, such as low-frequency or high-frequency filters. Additionally, outliers can be distinguished from normal values ​​by the difference between the waveform or intensity detected by an external motion sensor and the waveform or intensity of normal values, or by convolution, Fourier transform, etc., which are integral values ​​based on time changes.

[0082] at the same time, Figure 4 This is a flowchart of a method for reducing noise using an external measurement sensor of a moving object according to an embodiment of the present invention. The method for reducing noise using an external measurement sensor of a moving object according to the present invention includes the following steps: storing measurement values ​​of the external measurement sensor of the moving object via a storage unit (S100); generating multiple functions that can be derived from the stored measurement values ​​via a processing unit (S200); deriving a final function with the most measurement values ​​within a first range from the multiple functions via the processing unit (S300, S400); deriving an estimated value as a desired value of the measurement values ​​via the final function via the processing unit (S500); and selecting, via the processing unit, an approximate final value that is closest to the estimated value within a second range from the actual measurement values ​​at the time point corresponding to the estimated value as the final value (S600).

[0083] Importantly, the method for noise reduction using an external measurement sensor according to an embodiment of the invention derives a final function for removing noise from the current measurement value by selecting a model in which the number of normal signals included in past measurements is the largest, unlike the least mean square method. Furthermore, according to the embodiment, the trend of past measurements is tracked by setting the included criteria as measurements within a first range of function values. Accordingly, the final function derived by the embodiment accurately shows the trend of normal signals while removing noise from past measurements (provided the number of normal signals is greater than the number of noise). Thus, embodiments of the invention can robustly control a collision warning system or automatic emergency braking system against noise based on the predicted accurate estimates and final values. In this way, the collision prevention effect of each device can be substantially achieved.

[0084] The method for reducing noise using an external measurement sensor is a control method for removing abnormal signals from physical signals measured by the external measurement sensor and for detecting normal signals. In the case of the method for reducing noise using an external measurement sensor for a moving object, as an embodiment of the present invention, the measurement value measured by the external measurement sensor in the first step is stored (S100). In this case, the measurement value may be the time-of-flight (TOF) of an ultrasonic wave or the like, which can be detected when it is greater than or equal to a threshold of a certain intensity. Furthermore, in this step, measurement results measured at regular time intervals may be stored.

[0085] In the next step, logic for removing abnormal signals from the stored measurements is executed, and normal signals are detected. Thus, firstly, multiple functions derived from the stored measurements are generated (S200). In this step, functions can be generated from measurements selected by a first number of elements in the measurements. The first number of elements can be two (2), and the function can be a linear function. As a subsequent step, the number of measurements within a first range in the measurements, etc., based on the corresponding measurement time, is measured along with the function value and stored along with the corresponding function (S300). The second step described above can be repeated as many times as the first threshold. That is, functions derived from measurements using iterative logic as many times as the first threshold, and the number of measurements within the first range, can be stored continuously. Alternatively, functions where the number of measurements is less than the first threshold may not be stored.

[0086] In subsequent steps, the function with the largest number of stored measurements, i.e., the function with the most measurements, is derived as the final function (S400). If multiple final functions exist, the function with the largest sum of values ​​obtained by multiplying the measurements within a first range by a first weight, which increases as the measurements get closer to the current time point, can be derived as the final function when the measurements are closest to the current time point, and when the measurements are furthest from the current time point, the first weight can be the number of measurements within the first range, and when the measurements are furthest from the current time point, the first weight can be one (1). Alternatively, the function with the smallest sum of squares of the residuals between the measurements and the corresponding function values ​​can be derived as the final function. As another example, the function with the largest sum of values ​​obtained by multiplying by the first weight is derived as the final function, and if the sum is the same, the function with the smallest sum of squares of the residuals between the measurements and the corresponding function values ​​is derived as the final function.

[0087] In a method for reducing noise using an external measurement sensor of a moving object according to an embodiment of the present invention, a first weight is applied in deriving the final function. When the measured value is the time of flight (TOF) of the ultrasonic wave, the final function predicting the acceleration motion between the moving object and the obstacle is first derived. Therefore, in embodiments of the present invention, by prioritizing the risk of collision between the moving object and the obstacle, optimized operation of the device (vehicle dynamic control (VDC), traction control system (TCS), limited-slip differential (LSD), collision warning system, automatic emergency braking system, anti-braking system (ABS), etc.) can be performed to prevent the risk of failure or collision of the driving force and directional control of the moving object, depending on the function and purpose of the device.

[0088] Furthermore, the method for noise reduction of the external measurement sensor of the moving object according to embodiments of the present invention prioritizes determining the final function, rather than the conventional method of randomly determining the final function. Accordingly, embodiments of the present invention provide further space for noise removal, thereby enabling faster and more accurate operation of collision warning systems for moving objects, etc.

[0089] As a next step, an estimate of the expected value of the measured value can be derived using the final function (S500). However, in this step, if there is no function among the stored functions whose number of measured values ​​is greater than or equal to the second threshold, the estimate may not be derived. As a final step, the value among the actual measured values ​​at the time point corresponding to the estimate that is closest to the estimate within a second range can be selected as a candidate final value (S600). However, in this step, if there are consecutive approximate final values ​​among the actual measured values ​​at the time point corresponding to the estimate whose residuals to the candidate final values ​​are greater than or equal to the third threshold, the approximate final value can be selected as the final value.

[0090] The present invention relates to an apparatus and method for reducing noise in an external measurement sensor of a moving object, and more specifically, to an apparatus and method for reducing noise in an external measurement sensor of a moving object, wherein past data is additionally used in measuring external data of the moving object to reduce noise in the measured current data, and a function and a range of the function are used from past data measured at a immediately preceding time.

[0091] External measurement sensors (such as ultrasonic sensors) for moving objects (vehicles, etc.) are used for collision warnings or braking by detecting objects in the vehicle system. Traditional ultrasonic sensors detect relative distance to objects by emitting ultrasonic signals (emitted waves) and determining whether the intensity of the signal reflected from the object (received waves) exceeds a certain threshold. Therefore, conventional noise filtering techniques have emerged to robustly control anomalous signals by removing various types of noise contained in the received waves. As part of this, the least mean square method is a method for deriving a final function to predict current data based on past data measured by ultrasonic sensors. However, if anomalous signals are present in the past data, the function is calculated by summing the squares of the differences between the anomalous signals and the current signal, which may lead to incorrect predictions. Specifically, since the noise is calculated as the square of the residual in all cases—whether the noise is discrete, sparse, or present over a time period—given a certain ratio or higher to the normal signal, this results in predictions with low accuracy despite the fast measurement.

[0092] Correspondingly, if the collision warning system (W) and automatic emergency braking system (B) of the moving object to be controlled malfunction or over-operate, this may interfere with the driving of the moving object or cause risks such as collisions with other moving objects or collisions between other moving objects. Therefore, it is desirable to remove noise from past measurements. Accordingly, the means and method for noise reduction of the external measuring sensors of the moving object according to embodiments of the present invention derives a noise-robust final value from the estimated value evaluated by removing noise from past measurements. Thus, the internal and external devices of the moving object can be controlled quickly and accurately.

[0093] Although the invention has been shown and described with reference to specific embodiments, it will be apparent to those skilled in the art that various modifications and alterations may be made to the invention without departing from the spirit of the invention as set forth in the appended claims.

[0094] Explanation of reference numerals in the attached figures

[0095] B: Automatic emergency braking device

[0096] N: Devices used to reduce noise

[0097] C: Operation unit

[0098] S: External measurement sensor

[0099] M: Storage unit

[0100] W: Collision warning system.

Claims

1. A noise reduction device for an external measuring sensor of a moving object, the external measuring sensor including an ultrasonic sensor, the noise reduction device comprising: A storage unit stores the distance values ​​between the moving object and an object outside the moving object, as measured by the external measuring sensor. and The computation unit generates multiple functions that can be derived from the measured values ​​stored in the storage unit, derives a final function with the most measured values ​​in a first range from the multiple functions, derives an estimated value as the expected value of the measured values ​​through the final function, and selects the candidate final value that is closest to the estimated value in a second range from the actual measured values ​​at the time point corresponding to the estimated value as the final value. The computational unit selects two measurement values ​​from the stored measurement values ​​and generates one of a plurality of linear functions that can be derived from the selected measurement values; and The computational unit repeatedly generates the linear function to derive the final function.

2. The device for reducing noise according to claim 1, wherein, The storage unit stores the measurement values ​​taken by the external measurement sensor at regular time intervals.

3. The device for reducing noise according to claim 1, wherein, The arithmetic unit selects a first number of measurement values ​​from the stored measurement values ​​and generates the plurality of functions that can be derived from the selected measurement values.

4. The noise reduction device according to claim 1, wherein, When the number of functions generated is greater than or equal to the first threshold, the operation unit stops generating the multiple functions.

5. The device for reducing noise according to claim 1, wherein, If one or more of the selected measurements are zero, the computation unit does not generate the plurality of functions from the measurements.

6. The device for reducing noise according to claim 1, wherein, If the absolute value of the slope of the function derived from any two of the selected measurements is greater than or equal to a first threshold, then the computation unit does not generate the plurality of functions from the measurements.

7. The device for reducing noise according to claim 1, wherein, The computation unit excludes functions whose number of measured values ​​is less than a first threshold from the final function, and derives the final function with the most measured values ​​from functions whose number of measured values ​​is greater than or equal to the first threshold.

8. The device for reducing noise according to claim 1, wherein, If there is no function in which the number of measured values ​​within the first range is greater than or equal to the threshold, then the computation unit does not derive the estimated value.

9. The device for reducing noise according to claim 1, wherein, If multiple final functions exist, the computational unit will derive a function as the final function that is the maximum sum of values ​​obtained by multiplying the measured values ​​within the first range by a first weight, wherein the first weight increases as the measured values ​​are closer to the current time point.

10. The device for reducing noise according to claim 9, wherein, When the measured value is closest to the current time point, the first weight is the number of measured values ​​within the first range, and when the measured value is farthest from the current time point, the first weight is one.

11. The device for reducing noise according to claim 1, wherein, If multiple final functions exist, the computational unit derives the function having the least squares sum of the residuals between the measured value and the corresponding function value as the final function.

12. The device for reducing noise according to claim 1, wherein, If multiple final functions exist, the operation unit derives the function having the maximum sum of values ​​obtained by multiplying the measured values ​​within the first range by a first weight, wherein the first weight increases as the measured values ​​are closer to the current time point, and if the sums are equal, the operation unit derives the function having the least squares sum of the residuals between the measured values ​​and the corresponding function values ​​as the final function.

13. The device for reducing noise according to claim 1, wherein, If at the time point corresponding to the estimated value, there exists an approximate final value that is an actual measurement value whose residual with respect to the candidate final value is greater than or equal to a threshold, then the computation unit selects the approximate final value as the final value.

14. The noise reduction apparatus according to claim 13, wherein, If the residual between the approximate final value and the candidate final value is greater than or equal to the threshold, and there exists a second approximate final value that is a measurement value whose residual with the second candidate final value is greater than or equal to the threshold, then among the actual measurements at the time point immediately preceding the time point corresponding to the estimated value, the second candidate final value is the measurement value within the second range that is closest to the estimated value at the time point immediately preceding the time point corresponding to the estimated value, then the calculation unit selects the approximate final value as the final value.

15. The noise reduction apparatus of claim 1, further comprising a noise flag unit, the noise flag unit calculating the ratio of outliers among measurements taken before a predetermined time prior to the time point corresponding to the estimated value. in, If the ratio of outliers is greater than or equal to a first ratio, the computation unit selects only the candidate final value as the final value; and if the ratio of outliers is less than the first ratio, if there is an approximate final value at the time point corresponding to the estimated value whose residual with the candidate final value is greater than or equal to a threshold, the computation unit selects the approximate final value as the final value, and if there is no approximate final value, the computation unit selects the candidate final value as the final value.

16. The noise reduction apparatus according to claim 15, wherein, If there exists an approximate final value whose residual with the candidate final value is greater than or equal to a third threshold, and simultaneously there exists a second approximate final value that is a measurement whose residual with the second candidate final value is greater than or equal to the third threshold, then among the actual measurements at the time point immediately preceding the time point corresponding to the estimated value, the second candidate final value is the measurement closest to the estimated value at the time point immediately preceding the time point corresponding to the estimated value within the second range. If the absolute value of the slope of the straight line connecting the approximate final value and the second approximate final value at each time point is less than or equal to a second slope, then the calculation unit selects the approximate final value as the final value; otherwise, the calculation unit selects the candidate final value as the final value.

17. A method for reducing noise of an external measuring sensor of a moving object, the external measuring sensor comprising an ultrasonic sensor, the method comprising: The storage unit stores the measured distance values ​​between the moving object and an object outside the moving object, as measured by the external measuring sensor. The processing unit generates multiple linear functions that can be derived from the measured values ​​stored in the storage unit; The computational unit derives, from the plurality of linear functions, a final function having the most measured values ​​within a first range; The calculation unit derives an estimated value as the expected value of the measured value through the final function. as well as The calculation unit selects the candidate final value that is closest to the estimated value from the actual measured values ​​at the time point corresponding to the estimated value within a second range as the final value. Generating multiple linear functions involves selecting two measurements from the stored measurements and generating one of a plurality of linear functions that can be derived from the selected measurements; and Repeat the generation of the linear function to derive the final function.

Images