JP2680900B2 - Gyro offset removal method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial applications]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a problem of removing output offset, which is a problem when trying to detect a turning angular velocity of a vehicle using a turning angular velocity detection sensor such as an optical fiber gyro, a vibration gyro, or a mechanical gyro.

[Prior art]

FIG. 1 shows the configuration of a system that detects the turning angle of a vehicle using a gyro and obtains the direction in which the vehicle is traveling. There are a wheel speed sensor that determines the rotational angular velocity of the wheel, and a gyro that detects the rotational angular velocity of the vehicle body of the automobile. The gyro may be an optical fiber gyro, a vibration gyro, or a mechanical gyro. It may be one-dimensional or two-dimensional, but it detects the angular velocity at which the vehicle body rotates about an axis fixed to a certain vehicle body. Since the data corresponding to the angular velocity is output, the orientation of the vehicle body can be obtained by integrating the data. The traveling distance is obtained by accumulating pulses generated every time the wheel speed sensor travels a certain distance. When the differential traveling distance and the azimuth at each time are known, the position of the vehicle in this coordinate system will be calculated when an orthogonal coordinate system is assumed with a certain reference point as the origin. This is the output of this system, which gives the position in a coordinate system and the orientation of the car. The output of this system is collated with the map data recorded in the memory device, for example, the navigation system that seeks and displays the position of the user on the map, or the position of the user can be notified to the central management system by radio or the like. It is used for various location systems. In reality, the output of the gyro is a voltage output proportional to the rotational angular velocity. The voltage output is converted into a digital value by an AD converter and input to the computer system at fixed time intervals. In the computer system, it is integrated to obtain the bearing value.

[Problems to be solved by the invention]

The problem here is the offset included in the output value of the gyro and its drift. If the rotational angular velocity is 0, the output value of the gyro should be 0. But not really. The deviation of the gyro output value from 0 when the rotational angular velocity is 0 is called an offset. Offsets are not invariant but time-varying. Fluctuations over time are called drifts. It is necessary to remove the offset from the output value of the gyro. The value obtained by removing the offset is the true value indicating the rotational angular velocity. If the offset is not removed from the output value of the gyro, it will increase as a cumulative error in the computer, and the calculated heading will deviate from the original correct heading. Since the azimuth is obtained by integration, if the offset exists, its influence will be accumulated. The exact offset must be removed. In order to obtain the offset value, the gyro output when the vehicle is stopped due to a traffic signal is averaged to obtain the offset value. Then, while traveling, the offset value is subtracted from the gyro output to obtain the true value. When the vehicle is stopped, the angular velocity of the vehicle body should be zero. The output of the gyro at this time should include only the offset. However, as shown in FIG. 2, the output of the gyro is not constant and fluctuates slightly even when the vehicle is stopped. It may fluctuate around a certain value, but it may fluctuate around a value outside this range. The minute fluctuation has a small amplitude and a small cycle, but it can be said that this is short cycle noise. On the contrary, it may fluctuate so as to deviate from a constant value, and this is called long-period noise because its amplitude and period are relatively large. In this way, the output of the gyro is not constant and has a waveform including noise whenever the vehicle is stopped. However, it is certain that the gyro output while stopped will give an offset. It is reasonable to use the average value of the gyro output when the vehicle is stopped as the offset value. There are always various ways to use the average value. For example, it may be possible to average a certain period when the vehicle is stopped for a certain short time. In this case, it is difficult to remove the influence of long period noise. Perhaps the most accurate would be to time average the gyro output over the entire stoppage time. The problem then is that the time t ₁
And to know exactly the time t _{2 when the} vehicle started to travel. The traveling speed of an automobile is usually obtained by providing a mechanism for converting the rotation of tires into pulses (wheel speed sensor). The determination as to whether the vehicle is running or stopped is made based on the presence or absence of the pulse. That is, if there is a pulse from the wheel speed sensor, it is determined that the vehicle is running, and if there is no pulse, it is determined that the vehicle is stopped. However, strictly speaking, it is not correct to judge whether the vehicle is running or stopped based on the presence / absence of a pulse from the wheel speed sensor. When the vehicle is going to stop and the vehicle is going to run again, there is always a time in which the wheel speed pulse is not output although the vehicle is moving, depending on the resolution of the wheel speed pulse. FIG. 2 shows the gyro output (a) and the wheel speed pulse (b) when the vehicle slows down and stops and then starts traveling again. The wheel speed pulse does not occur before the stopped time t ₁ (t ₃ ). The time between the time t ₃ when the last wheel speed pulse is generated and the time t ₁ when the vehicle is stopped is a period in which the vehicle actually travels but appears to be stopped by the wheel speed sensor. On the contrary, at the moment of departure, a wheel speed pulse is generated after the start of running t ₂ (t ₄ ). Departure time t ₃ and pulse generation time
The time period from t ₄ is also a period in which the vehicle actually travels but seems to be stopped by the wheel speed sensor. It is judged from the wheel speed pulse that the vehicle is stopped, and t ₃ ~
When the average time between t _4,違Tsuta value is obtained from the time average of between t ₁ ~t ₂ is a time of actual stop. The desired offset value Δ is gyro output as Ω (t) However, if the wheel speed pulse is used to determine whether the vehicle is stopped, Will be obtained. Since Δ and Δ have different integration ranges, different values are given. Period is short, so average value It seems that the effect on the can be ignored, but it is not. As shown in FIG. 2 (a), the gyro output is significantly different between when the vehicle is running and when the vehicle is stopped. Even if the speed is low, the amplitude of fluctuation of the gyro output is large and the period is long. The gyro output becomes extremely small while the vehicle is stopped. Therefore, when the gyro output Ω (t) is integrated at t ₃ to t ₄ , the influence of the period of and is superior. Since this is divided by (t ₄ −t ₃ ), t ₁ ~ t ₂
The value will be completely different from the average value Δ during the period. When the gyro output during running enters the integration, such a large error is caused, and it is considered that this should be eliminated. Therefore, for example, it is conceivable that the offset is calculated by averaging during a short period while the vehicle is stopped. However, this makes it more susceptible to long-period noise,
After all there is an error. It is desirable to obtain the offset by averaging the gyro outputs while the vehicle is stopped and for a sufficiently long time. It is an object of the present invention to provide a method in which a gyro output can be integrated and an average value can be obtained while the vehicle is stopped and as long as possible.

[Means for Solving the Problems]

In order to achieve the purpose of the previous term, the gyro offset removal method of the present invention is to determine the traveling distance by a wheel speed sensor that detects the rotational angular velocity of the wheel, and integrate the gyro output to determine the angular velocity of the moving body to determine the direction of the moving body. In a mobile unit-equipped system that calculates the gyro output, the offset value is subtracted from the gyro output while the mobile unit is stopped, and the gyro output is corrected by temporarily storing the gyro output sampled data. It consists of a buffer memory, a block that accumulates the data that overflows from the buffer memory, a counter block that counts the number of accumulated data, and an arithmetic unit that divides the accumulated data by the number of data to obtain an average value. After a certain period of time, the data begins to be stored in the buffer memory and A) The data overflowing from the memory is integrated by the integration block, the number of data integrated at the same time is counted by the counter block, and then the integrated data is divided by the number of data at the time when the running is started. The average value is obtained, the contents of the buffer memory are discarded, and the average value is used as the offset value. FIG. 3 shows a schematic structure of the gyro offset removing method of the present invention. The top row of waveforms is the wheel speed pulse. The dots on the second row indicate the sampling points of the gyro data. Data is output at each sampling time such as D ₁ , D ₂ , ... And is input to the buffer memory 1. The buffer memory 1 is used for temporarily storing the gyro output data. However, it is not a normal memory configuration, but when the memory is full, it is configured so that the data inserted first will overflow. In other words, it is a first-in first-out memory. The purpose of this is to delay the averaging operation of the gyro data output and exclude the data in the period as shown in FIG. 2 from the averaging operation. The data overflowing from the buffer memory is integrated by the integration block 2. In addition, the number of pieces of data accumulated at the same time is also counted by the counter block 3. By dividing the integrated value of the data by the number of data n, the average value of the overflowed data Ask for. There are two things to note here. One is that "the data starts to be stored in the buffer memory after a certain period of time after the mileage sensor indicates the stop". This is because the period shown in FIG. 2 is excluded from the averaging operation. The other is to use the buffer memory 1 and average the overflowed data. The data stored in the buffer memory 1 at the time of averaging is not included in the averaging operation. This is because the period shown in FIG. 2 is excluded from the averaging operation.

【Example】

When the car stops, the wheel speed pulse is interrupted (t _{3 in} Fig. 2). However, even if the wheel speed pulse is interrupted, after that,
Since the car is moving up to time t ₁ , anticipate it and wait for _a certain time Ta. This is referred to as the front waiting time T _a here. The time when waiting for T _{a is} set as t ₅ . t ₃ + T _a = t ₅ (3). t ₅ should be after t ₁ . In other words, the previous waiting time T _a is set in advance as the maximum value of the time (t ₁ −t ₃ ) from when the wheel speed pulse is interrupted (t ₃ ) to when the vehicle stops (t ₁ ) or a value greater than the maximum value. Make a decision. T _a ≧ max {t ₁ −t ₃ } (4). max {...} means the maximum value of the group {...}. From the wheel speed pulse is Todan Ete (t _3), slide into entering the Jiyairodeta to the buffer memory 1 after T _a time. That Jiyairodeta of time t ₃ ~t ₅ is not input from the beginning to the buffer memory 1. Data D ₁ , D ₂ , ... Are input from t ₅ . This is the gyro data output after a complete stop. Let the capacity of the buffer memory be m. The data is sequentially stored in the buffer memory 1. The data does not overflow from the buffer memory until the mth data is stored. However, when the mth data is input, the memory cell becomes full, so when more data is input, the first data D ₁
The memory overflows in order. The number of overflowed data is n. The counter block 3 counts this number n. Overflowing data D ₁ ,
The integrated block 2 calculates the integrated value of. Let S be the output of the counter block and Q be the output of the integration block. S = n (5) It is. As it is, it is difficult to understand the correspondence between the times t and n. Therefore, the relationship with the gyro output Ω (t) at time t will be described. Let τ be the sampling interval. T _a = τh (7) T _b = τm (8) As described above, T _a is the time from when the wheel speed pulse disappears to when data input to the buffer memory starts. T _b is a delay time caused by the number m of memory cells in the buffer memory. T _a and T _b are periods,
Is a delay time that is provided to remove from the average operation. Expressions (5) and (6) are associated with an arbitrary time t after t ₅ as follows. S = (t−t ₅ −T _b ) / τ (9) The counting of the number of data n does not start immediately at t ₅ , but after the time T _b required for the data to pass through the buffer memory. Therefore, the time until the time t is (t−t ₅ −T _b ). Since one data is output for each τ, the number of data n is given by (9) as a value obtained by dividing this by τ. Overflowing time of the data is from (t ₅ + T _b), because it integrates so far time t, the integral range of Q is (t ₅ +
T _b ) to τ. However, since it takes T _b to pass through the buffer memory, the data output at time t is actually the value of t−T _b . So the integrand is Ω
(T−T _b ). However, D _k and Ω (t) are the same gyro data. Since the way of giving the independent variable and the suffix is different, the notification is changed. Since D _k is a value sampled at every τ from time t ₅ , there is a relationship of t = t ₅ + (n−1) τ (11) between t and n, and at this time, D _n = Ω (t) ( 12) However, this is not given to the output of the buffer memory at time t. The value of (12) appears at the buffer memory output at time t + T _b . It is delayed by T _b . If Q / S is calculated, the average value will be given. However, Q / S is not calculated at any time. Instead, the average value of the integrated data is obtained at the time when the wheel speed pulse is input again (t _{4 in} FIG. 2). This is the offset value. Offset value Therefore, the offset value will be calculated from the apparent gyro output the next time the vehicle starts running. What is subtracted is the true gyro output. From this, the angular velocity of the car is calculated. Obtaining Q / S at time t ₄ and performing averaging means
Calculate the value that can be calculated by equations (5) and (6) up to t = t _{4, and} then Q / S
Is to calculate. As for n, it starts counting with a delay of T _b from time t ₅ and stops at t ₄ , It is. This of from time t ₄ to T _b only time earlier t ₆ time, it can be considered as a number of pulses. Time t _6, which was newly defined must be actually earlier than the time t ₂ when the automobile was starting. That is, T _b must be longer than the maximum value from the time the vehicle starts (t ₂ ) to the time when the wheel speed pulse appears (t ₄ ). (T ₄ −
Although t ₂ ) changes depending on the starting position and the starting position, the maximum value of t ₂ ) is a determinable value. T _b ≧ max {t ₄ −t ₂ } (16) Now, the value integrated by the integration block is the sum of D ₁ to D _{n. From} equation (14), n is the sample number at time t _6. I know there is something. That is, the data is the one between the t ₅ ~t ₆ are integrated. Since the number of samples from t _{5 to} t ₆ is n and the integrated value of the data during this period is Q, it means that Q / S gives the average value of the gyro output from t ₅ to t ₆ . The same thing can be explained by using the equations (9) and (10). At t = t ₄ (time when the wheel speed pulse occurs),
Since it calculates Q / S, the result is Becomes However, in the conversion of the expression from (17) to (18), t−T _b is replaced with t. From (18), Q / S is t
It is seen apparent that ₅ ~t ₆ is a time average of data output. In this way, the time after the time the car stopped
From t ₅ , the average value of the gyro output until time t ₆ immediately before the vehicle starts moving is obtained. The gyro output data when the car is moving is not included in the averaging operation. However, in the case of the present invention, the offset value is automatically calculated even when the actual stop time is short. In this case, offset correction may be performed with a value that does not include the influence of long-period noise. Considering such a possibility, when the averaging time is short, the value is not adopted and is discarded. Alternatively, if the previous offset calculation value and weighted averaging are performed, it is conceivable to perform some processing together.

【The invention's effect】

The advantages of the present invention are as follows. Since the number of data of the gyro output to be sampled as the offset value while the vehicle is stopped is not limited, the data can be sampled over almost the entire stop time and averaged. There is no effect of long period noise. More accurate average value can be obtained. The buffer memory is used to average the gyro data strings that have passed through the buffer memory and overflowed. The data in the buffer memory when the wheel speed pulse is output is not used as the data for obtaining the average value and is discarded. As a result, the gyro output data in the period from when the vehicle starts to move until when the wheel speed pulse comes in can be excluded from the calculation of the offset value.
Therefore, the size m of the buffer memory needs to be a little larger than the data for the time from when the vehicle is stopped to when the wheel speed pulse is generated. When you try to stop, the wheel speed pulse stops
Gyro data is not stored in the buffer memory for T _a time. Therefore, the gyro output from the time when the wheel speed pulse is not output to the time when the vehicle is stopped is not included in the average value calculation. An accurate offset value can be obtained regardless of the size of the stop time. It is effective to use it by incorporating it into a navigation system or location system that includes offset drift.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 shows the direction of an automobile from a wheel speed sensor and a gyro,
The schematic block diagram of the system which calculates | requires a position. FIG. 2 is a waveform diagram of a gyro output and wheel speed pulse when the automobile runs, stops, and runs. FIG. 3 is a diagram for explaining the average calculation according to the present invention when the automobile runs, stops, and runs. 1 ... Buffer memory 2 ... Integration block 3 ... Counter block t ₁ ...... The time when the car stopped t ₂ ...... The time when the car started t ₃ ...... The wheel speed pulse stopped immediately before the stop Time t ₄ ...... Time at which the first wheel speed pulse is generated immediately after starting t ₅ ...... Time when integration of gyro data is started in the present invention t ₆ ...... In the present invention, the integrated value of gyro data is divided by the number of data Time to calculate the average value

Claims (1)

(57) [Claims] 1. A moving body mounting system for calculating a traveling distance by a wheel speed sensor for detecting a rotational angular velocity of a wheel, and integrating a gyro output for obtaining an angular velocity of the moving body to calculate an azimuth of the moving body. A method for correcting the gyro output by obtaining the offset value from the gyro output and then subtracting the offset value, a buffer memory for temporarily storing the data sampled from the gyro output,
It consists of a block that accumulates the data that overflows from the buffer memory, a counter block that counts the number of accumulated data, and an arithmetic unit that divides the accumulated data by the number of data to obtain an average value. Starting to store the data in the buffer memory after a fixed time, the data overflowing from the buffer memory is integrated by the integration block, and the number of the simultaneously integrated data is counted by the counter block, and at the time when the running is started, A gyro offset removing method, wherein an average value is obtained by dividing the integrated data by the number of data, the contents of the buffer memory are discarded, and the average value is used as an offset value.