Patent Application
20110277532
1. Field of the Invention
The present invention relates to a method for calibrating an acceleration sensor.
2. Field of the Invention
Methods of this kind are generally known. For example, printed German patent application document DE 10 2007 002 835 A1 describes a method for calibrating a yaw rate sensor system, in which calibration is carried out using inclination data from an acceleration sensor system. The acceleration sensor system also sends a calibration-initiating zero signal to the yaw-rate sensor system as soon as a resting state is detected, so that in each instance, the calibration of the yaw-rate sensor system is only carried out in the resting states.
However, this method is not suitable for calibrating acceleration sensors in portable devices by measuring gravitational acceleration in a resting state, i.e., a “1 g” state of the portable device, since in this case, in addition to the detection of the resting state (detection of the absence of accelerations, which are superimposed on the gravitational acceleration), the resting state must also be distinguished from “free-fall,” since in “free fall,” it is not possible to measure gravitational acceleration using only an acceleration sensor.
The method and the electronic device according to the present invention have the advantage over the related art that the presence of a resting state is detectable in a comparatively simple, energy-saving and efficient manner and therefore, a calibration of the acceleration sensor is only carried out in the presence of such a resting state. In the spirit of the present invention, a resting state (also denoted here as a “1 g” state) is, in particular, a state in which, in essence, only gravitational force (“1 g” acceleration, “g” standing for the acceleration due to gravity) is acting as an acceleration force and no “free fall” is present. For example, the portable device in which the acceleration sensor to be calibrated is situated lies stationary on a support surface such as a granite table plate. This has the advantage that the calibration of the acceleration sensor is carried out exclusively in the resting state and, in particular, not during the presence of other dominating acceleration forces, since a precise measurement of acceleration due to gravity is only possible in this case. In this context, the acceleration sensor is adjusted or calibrated, in particular, in light of the measurement of the known acceleration due to gravity. Therefore, both a calibration of an already pre-calibrated acceleration sensor and a calibration of a still uncalibrated acceleration sensor are rendered possible in an advantageous manner. This allows, for example, (post- and/or re-) calibration of the acceleration sensor during the use of a device in which the acceleration sensor is integrated. This is especially advantageous, in particular, in consumer products equipped with acceleration sensors, such as cellular phones, digital cameras, laptops, notebooks, PDA's (personal digital assistant), hand-held GPS devices, game consoles, input devices for game consoles/computers (mouse, joystick, game controller) and the like, since the method of the present invention may allow a complicated and cost-intensive (pre-) calibration of the acceleration sensor during manufacture or immediately after manufacture of the device to be dispensed with. In addition, the acceleration sensor is, in particular, a three-channel acceleration sensor, i.e., sensitive along the three spatial directions X, Y, Z. Alternatively, the acceleration sensor includes three individual, single-channel acceleration sensor units, which are oriented such that accelerations along each of the three spatial directions are detected by one of the acceleration sensor units. The acceleration sensor preferably includes a micromechanical acceleration sensor. The micromechanical acceleration sensor preferably includes a seismic weight, which is movably suspended with respect to a substrate, and whose deflection with respect to the substrate as a result, of inertial forces generated by external acceleration forces is measured capacitively (for example, using a finger-electrode and/or capacitor-plate set-up). The calibration of the acceleration sensor is used, in particular, for determining and possibly compensating for manufacturing-specific offsets and sensor sensitivities for each of the three spatial directions.
According to a preferred, specific embodiment, it is provided that in a first partial step of the sixth method step, for each of the three spatial directions, an average value is determined as a function of the respective acceleration values, the average values preferably being ascertained using a least squares method; and that in a second partial step of the six method step, the calibration of the acceleration sensor is carried out on the basis of the average values. Consequently, a pre-calibration (i.e., a rough adjustment) of the acceleration sensor is advantageously carried out. In this manner, it is ensured that subsequent, in particular, iterative method steps for post-calibration (i.e., for fine adjustment) of the acceleration sensor converge, and that consequently, the accuracy of the calibration is markedly increased. In addition, the computational expenditure, i.e., the number of required iteration steps, and thus, also the time required for these subsequent method steps, are reduced. On the basis of the ascertained average values, current offset values and sensory sensitivities are preferably ascertained for each spatial direction, and therefore, pre-calibration of the acceleration values is carried out.
According to a preferred embodiment, it is provided that in a third partial step of the sixth method step, a calibration of the acceleration sensor be carried out using an iterative approximation method, the iterative approximation method preferably being implemented as a function of the acceleration values and/or as a function of further acceleration values. Consequently, the accuracy of the calibration is successively increased in an advantageous manner, using the iterative approximation method. In this context, the iterative approximation method preferably includes a Kalman filter, a Newtonian approximation method and/or a method of least squares. In this context, for each spatial direction, the offset values and the sensor sensitivities are successively brought closer to the actual offset values and sensor sensitivities of the acceleration sensor, using the iterative approximation method.
According to a preferred embodiment, it is provided that in a fourth partial step of the sixth method step, a checking method be implemented, in which a sum of squares is calculated as a function of at least one further acceleration value for each of the three spatial directions, and in which the sum of squares is compared to a third threshold value. Consequently, the quality of the calibration is advantageously determined, which means that as a function of the comparison of the sum of squares and the third threshold value, it is decided if a further iteration step is necessary for improving the calibration or if the calibration method may be terminated at this point (when the calibration quality is sufficiently high).
According to a preferred embodiment, it is provided that the third and fourth partial steps be sequentially repeated until, in the fourth partial step, the sum of squares is less than the third threshold value; and/or, that in the fourth partial step, the sum of squares is compared to a fourth threshold value, the method preferably being restarted at the first method step when the sum of squares is greater than the fourth threshold value. The comparison of the sum of squares to the fourth threshold value advantageously gives the option of completely restarting the method of the present invention (this means that all of the offset values and sensor sensitivities ascertained up to this point are discarded), when a quality of the calibration is detected as overly poor. Consequently, the pre-calibration, i.e., the rough adjustment, is also carried out once more.
According to a preferred embodiment, it is provided that in the second method step, for each of the spatial directions, both a maximum value and a minimum value in the respective acceleration values be determined; for each of the three spatial directions, the comparison value being determined, in each case, as a difference of the respective minimum value and the respective maximum value of the corresponding spatial direction. In this manner, the comparison value may be ascertained in a particularly simple manner and is thereby a reliable criterion for determining the dynamics along the specific spatial direction. Such functionality is, in particular, implementable in hardware and/or software in a comparatively simple, space-saving, and energy-efficient manner. It is particularly preferably provided that the acceleration values for each spatial direction, i.e., in particular, an X, Y and Z spatial direction, be saved in a separate FIFO memory (first in, first out memory), that is, a plurality of X-movement values are stored in an X-FIFO memory, a plurality of Y-movement values are stored in a Y-FIFO memory and a plurality of Z-movement values are stored in a Z-FIFO memory. Subsequently, a maximum and a minimum value are determined in each FIFO memory, and after that, the specific differential value of these maximum and minimum values is determined for each of the three spatial directions X, Y and Z, so that an X comparison value, a Y comparison value and a Z comparison value are generated. These three comparison values are then individually compared to a common, first threshold value or, as an alternative, to a separate, respective, first threshold value for each spatial direction. If the first threshold value of each comparison value, or each first threshold value of the respective comparison value, is not reached, it is ensured that the acceleration sensor is not subjected to any external acceleration forces large enough to hereby render impossible the measurement of the gravitational acceleration. Consequently, the calibration of the acceleration sensor is carried out if it is ensured, in the fifth method step, that no “free fall” is present.
According to a preferred embodiment, it is provided that in the fourth method step, the cumulative value be constituted of a sum of, in each instance, an acceleration value for each of the three spatial directions; and/or that in the fourth method step, for each of the three spatial directions, an average acceleration value be calculated from the respective acceleration values, the cumulative value being calculated as a sum of the specific, average acceleration values for each of the three spatial directions. In the fourth method step, the cumulative value is calculated, in particular, as a sum of absolute values as a function of at least one acceleration value for each spatial direction. Only reduced acceleration forces act in the state of free fall, which means that using the third method step alone, the resting state cannot be distinguished from “free fall.” Now, with the aid of the fifth method step, the resting state is advantageously distinguishable from the state of “free fall,” since in the case of free fall, the sum of absolute values of the specific acceleration values is less than in the resting state, which means that a differentiation is rendered possible by a suitable choice of the second threshold value. Calculating the cumulative value as a function of average acceleration values now advantageously increases the reliability and the accuracy in distinguishing the resting state from “free fall” in a simple manner.
A preferred embodiment provides that for each spatial direction, acceleration values ascertained in the first method step and/or further acceleration values be compared to previously ascertained acceleration values; the newly ascertained acceleration values and/or further acceleration values, as well as acceleration values already ascertained beforehand, each being preferably rounded, and the newly ascertained acceleration values and/or further acceleration values then being discarded, in particular, when the rounded, newly ascertained acceleration values and/or further acceleration values essentially correspond to the rounded acceleration values ascertained beforehand, and/or when a particular number of rounded, previously ascertained acceleration values are present, which correspond to the rounded, newly ascertained acceleration values and/or further acceleration values. Therefore, it is advantageously determined if the newly ascertained acceleration values are present in a “cluster,” which is already occupied by acceleration values already ascertained beforehand. In other words, it is checked if there are already similar acceleration values among the acceleration values ascertained beforehand. This prevents acceleration values that are only the same or similar from being used for calibrating the acceleration sensor again, since such a calibration would not produce the desired calibration quality. The newly ascertained acceleration values are preferably rounded to produce a particular accuracy, for example, by dividing the acceleration values by second powers and comparing them to the already previously ascertained acceleration values rounded in the same manner. In particular, the above-mentioned rounding on the basis of 2nd powers may be implemented in hardware in a comparatively simple manner. Subsequently, the newly ascertained acceleration values are only then further used, when they are in an unoccupied “cluster” and/or only a particular number of acceleration values already ascertained beforehand are in the corresponding “cluster.” In an analogous manner, newly ascertained, further acceleration values are likewise rounded and compared to acceleration values already ascertained beforehand and rounded and/or to further acceleration values. This also prevents only the same or similar, further acceleration values from being used to calibrate the acceleration sensor again, since such a calibration would likewise not produce the desired calibration quality.
An electronic device having an acceleration sensor and an evaluation unit constitutes further subject matter of the present invention; the acceleration sensor being configured to ascertain acceleration values along three spatial directions; the evaluation unit being configured to generate, in each instance, a comparison value for each of the three spatial directions from the acceleration values, as well as to compare each of the comparison values to a first threshold value; the evaluation unit further being configured to generate a cumulative value as a function of at least one acceleration value for each of the three spatial directions, as well as to compare the cumulative value to a second threshold value; the evaluation unit being configured to calibrate the acceleration sensor, when, for each of the three spatial directions, the comparison value is less than the threshold value, as well as when the cumulative value is greater than the second threshold value. Thus, in an advantageous manner, the calibration of the acceleration sensor is only then carried out, when the electronic device is in the resting state. In addition, the electronic device is specially configured to implement the method according to the present invention. The electronic device includes, in particular, a cellular phone, digital camera, laptop, notebook, PDA (personal digital assistant), hand-held GPS device, game console, input device for game consoles/computers (mouse, joystick, game controller) and/or the like. The evaluation unit includes, in particular, a sensor logic circuit, which is preferably implemented in an ASIC and/or microcontroller and preferably has at least three FIFO memories.
In the spirit of the present invention, a resting state (also referred to as a “1 g” state) is, in particular, a state in which, in essence, only the gravitational force (“1 g” acceleration, where “g” stands for the gravitational acceleration) acts as an acceleration force on system 1 and, in addition, no free fall is present. The calibration method is realized, in particular, using software and/or using additional logic circuitry implemented in hardware. Accordingly, evaluation unit 32 preferably includes an ASIC (made of, e.g., a digital part and an analog part), a microcontroller and/or a computer chip. The exact method of functioning of the calibration method is described in detail in the following, in particular, in view of
A schematic flow chart of a method according to a first specific embodiment of the present invention is illustrated in
In second method step 11, for each of the three spatial directions X, Y, Z, a comparison value 2X, 2Y, 2Z is ascertained from acceleration values 1X, 1Y, 1Z; in third method step 12, each of comparison values 2X, 2Y, 2Z being compared to a first threshold value 20. The first threshold value 20 not being reached by specific comparison value 2X, 2Y, 2Z is a measure that no or only negligible dynamics are occurring along corresponding spatial direction X, Y, Z. A cumulative value 21 is further calculated in fourth method step 13, the addends of cumulative value 21 being formed as a function of at least one acceleration value 1X, 1Y, 1Z for each of the three spatial directions X, Y, Z. Cumulative value 21 is therefore preferably proportional to the vector sum of specific acceleration values 1X, 1Y, 1Z. In fifth method step 14, cumulative value 21 is subsequently compared to a second threshold value 22, in order to distinguish the resting state from the state of “free fall.” Second threshold value 22 is preferably greater than 0.2 g, particularly preferably greater than 0.5 g and especially preferably greater than 0.6 g, where g is the acceleration due to gravity.
Provided that, in third method step 12, for each of the three spatial directions X, Y, Z, comparison value 2X, 2Y, 2Z is less than threshold value 20, and at the same time, in fifth method step 14, cumulative value 21 is greater than second threshold value 22, a calibration of acceleration sensor 30 is carried out in a subsequent, sixth method step 15. In this context, sixth method step 15 includes a first partial step 40; for each of the three spatial directions X, Y, Z, an average value 3X, 3Y, 3Z preferably being calculated from acceleration values 1X, 1Y, 1Z, using the method of least squares. In a second partial step 41 of sixth method step 15, a pre-calibration, i.e., a rough adjustment, of acceleration sensor 30 is subsequently carried out on the basis of average values 3X, 3Y, 3Z; at this point, the sensor offset and the sensor sensitivity being determined for each of the three spatial directions X, Y, Z. After that, in a second partial step 42 of sixth method step 15, an iterative approximation method for post-calibrating, i.e., finely adjusting, acceleration sensor 30 is carried out. The iterative approximation method preferably includes a Kalman filter and is carried out on the basis of previously ascertained acceleration values 1X, 1Y, 1Z and/or preferably on the basis of further acceleration values 1′X, 1′Y, 1′Z.
Further acceleration values 1′X, 1′Y, 1′Z are ascertained in an intermediate step 43 of sixth method step 15; analogously to above-mentioned acceleration values 1X, 1Y, 1Z, further acceleration values 1′X, 1′Y, 1′Z each preferably also having to successfully pass through the “clustering” step and the “resting-state detection” step beforehand. Now, on the basis of these further acceleration values 1′X, 1′Y, 1′Z, a checking method is executed in a fourth partial step 44 of sixth method step 15; in the checking method, in a first sub-step 44′, a sum of squares 23 being calculated as a function of at least a further acceleration value 1′X, 1′Y, 1′Z, and in a second sub-step 44″, this sum of squares 23 being compared to a third threshold value 24. For further understanding: in this checking step, it is particularly checked if, in the resting state of acceleration sensor 30, the vector sum of a further acceleration value 1′X, 1′Y, 1′Z for each of the three spatial directions X, Y, Z, respectively, is close to the 1 g vector.
If sum of squares 23 exceeds third threshold value 24, then the quality of calibration is not yet sufficient, and third partial step 42 is executed once more to improve the quality of calibration. As an option, in the checking method, sum of squares 23 is even compared to a fourth threshold value 25 as well, which is greater than third threshold value 24. If fourth threshold value 25 is also exceeded by sum of squares 23, it is assumed that the previously executed sensor calibration is exceedingly poor, which means that the method of the present invention is completely restarted (otherwise, there is the risk of the iterative approximation method not converging). In this instance, the sensor offsets and sensor sensitivities already ascertained are discarded.
If sum of squares 23 is less than fourth threshold value 24, then, in a seventh method step 16, the method goes into a holding loop. After a run-off of a particular predefined period of time, which is preferably automatically set as a function of the difference of sum of squares 23 and fourth threshold value 24, intermediate step 43, as well as the checking method, are repeated in order to ensure permanent monitoring of the calibration quality of acceleration sensor 30 during use. Sum of squares 23 is calculated, for example, as (1X)2+(1Y)2+(1Z)2. As an alternative, it is conceivable to use the sum of (1−sqrt[(1X)2+(1Y)2+(1Z)2])2 for determining the calibration quality.
In a subsequent auxiliary step 50, it is checked, with the aid of a logical AND gate, if all three comparison values 2X, 2Y, 2Z are each less than first threshold value 20. Concurrently to this, for each FIFO memory 33X, 33Y, 33Z, an average acceleration value 6X, 6Y, 6Z is initially calculated from respective acceleration values 1X, 1Y, 1Z for each of the three spatial directions X, Y, Z, and in a fourth method step 13, cumulative value 21 is then calculated from the three average acceleration values 6X, 6Y, 6Z (|6X|+|6Y|+|6Z|). Alternatively, it is also conceivable for cumulative value 21 to be calculated, in each instance, from a single acceleration value 1X, 1Y, 1Z per spatial direction X, Y, Z, i.e., without calculating an average value (|1*X|+|1*Y|+|1*Z|) (in this case, for example, the last acceleration value 1X, 1Y, 1Z in respective FIFO memory 33X, 33Y, 33Z is always used). In fifth method step 14, it is now checked if second threshold value 22 is exceeded by cumulative value 21.
In a subsequent, further auxiliary step 51, it is checked, with the aid of a further logical AND gate, if both second threshold value 22 is exceeded by cumulative value 21 and all three comparison values 2X, 2Y, 2Z are each less than first threshold value 20. Only when these two conditions are satisfied, are acceleration values 1X, 1Y, 1Z written into a data cluster 52 and, in a sixth method step 15, a calibration of acceleration sensor 30 carried out (provided enough acceleration values 1X, 1Y, 1Z are present). Due to the two mathematical conditions, it is ensured that, on one hand, the resting state is distinguished from free fall (fifth method step 14) and that, on the other hand, the maximum allowed dynamics of acceleration sensor 30 in the resting state are limited for individual spatial directions X, Y, Z (third method step 12). In a further, first “clustering step” 10′, rounded, newly ascertained, further acceleration values 1′X, 1′Y, 1′Z are then compared to these acceleration values 1X, 1Y, 1Z currently written into data cluster 52 and checked to determine if they are in one and the same “cluster.”
| Number | Date | Country | Kind |
|---|---|---|---|
| 10 2010 028 827.6 | May 2010 | DE | national |
