DE102022204010A1 - System and method for determining a temperature compensation offset value in a gyroscope of a device - Google Patents

Abstract

An apparatus (100) for determining a temperature compensated offset value in a gyroscope (102) is provided. The device (100) includes a control unit (110) and a detection module (104) to detect a temperature of the gyroscope (102). The control unit (110) includes a calibration module (104) configured to acquire and process data from the acquisition module and gyroscope (102). The calibration module (106) is further configured to calculate a gyro offset value using a variance-limited averaging method from the captured gyro sensor data. The device (100) further includes a TCO estimation module (108) configured to estimate a temperature compensation (TCO) offset factor using a first Kalman filter. The apparatus (100) further comprises a bias estimator (112) configured to calculate a temperature compensated gyro offset value by combining an estimated predicted TCO offset and gyro offset value using a second Kalman filter.

Description

field of invention

The present invention relates to a system and a method for determining a compensation offset value in a gyroscope of an apparatus and a control unit therefor.

Background of the Invention

A gyroscope sensor is used as the main sensor to determine orientation. At present, most of the devices use a high-precision gyroscope with a higher cost to achieve the higher orientation precision. The gyroscope signal has an offset error due to its imperfections. The device orientation is calculated based on the algorithm using the integration of the gyroscope signal. Because the signal has the offset error, the calculated orientation also has an offset. As the algorithm uses integration, this error continues to increase. To avoid this deviation, an occasional gyroscope calibration is performed. A typical gyroscope calibration mechanism requires a steady state device without any vibration. Typically this can be achieved by turning off all typical sources of vibration such as motors of devices. Still, this doesn't give a better user experience as it stops in between during its normal operation.

Current products use industrial grade sensors for vacuum cleaner robots due to their better performance. This idea helps to improve the performance of the orientation result by compensating for the preload drift due to a temperature change, which allows the industrial grade sensors to be replaced by MEMS (microelectromechanical system) based sensors. This significantly reduces the cost of the product. This invention provides a software solution of a temperature compensation mechanism in use with minimal calibration pause during its normal operation. The proposed method performs most of the calibrations while devices are charging and uses this learning during normal operations. This gives a better user experience because the frequency of calibration is lower.

the state of the art, US9581466B2 US 5,000,000 discloses systems, methods, devices, and computer-readable media for scheduling execution of a task such as e.g. B. a latency-insensitive non-real-time background task on a computing device. In one implementation, the technique includes detecting a first state of a device, the first state of the device being associated with a first power level and a first task, the first power level being based at least in part on a power consumption of a first task, determining that the first Power level associated with the first state is above a threshold and in response to determining that the first power level associated with the first state is above the threshold and scheduling execution of a second task at the device, wherein the second task is associated with automatically collecting calibration data using at least one sensor.

character list

• 1 stellt ein Blockdiagramm einer Vorrichtung, die mit einem Gyroskop verbunden ist, zum Bestimmen eines Temperaturkompensationsversatzwerts eines Gyroskops gemäß einer Ausführungsform der Erfindung dar;
• 2 stellt einen Ablaufplan eines Verfahrens zum Bestimmen eines Temperaturkompensationsversatzwerts im Gyroskop gemäß der vorliegenden Erfindung dar; und
• 3 stellt einen Graphen für Gyroskopversatzwerte dar, die eine Hysteresebeziehung aufzeigen, wenn Temperaturwerte variieren.

Various modes of the invention are disclosed in detail in the description and illustrated in the accompanying drawings:

• 1 12 illustrates a block diagram of an apparatus, coupled to a gyroscope, for determining a temperature compensation offset value of a gyroscope, according to an embodiment of the invention;
• 2 Figure 12 illustrates a flow chart of a method for determining a temperature compensation offset value in the gyroscope in accordance with the present invention; and
• 3 Figure 12 presents a graph of gyroscope offset values showing a hysteresis relationship as temperature values vary.

Detailed description of the embodiments

1 10 illustrates an apparatus 100 for determining a temperature-compensated offset value of a gyroscope 102. The gyroscope 102 is coupled to the apparatus 100 according to an embodiment of the invention. In one embodiment, device 100 may be one of a group of devices be selected to include a vacuum cleaner, a lawnmower, a robotic arm, and the like. The device 100 further includes a control unit 110 and a detection module 104 to detect a temperature of the gyroscope 102 . Each component is described in more detail below.

The control unit 110 further includes a calibration module 106, a TCO estimation module 108 and a preload estimator 112.

wobei $$g_{nor{m}_k}=\sqrt{g_x^2+g_y^2+g_z^2}$$

• g_thre = Gyronormschwellenwert für keine Bewegung
• T_Th = Minimale Zeitgrenze ohne Bewegung

The calibration module 106 is configured to acquire and process data from the acquisition module 104 and gyroscope 102 . The calibration module 106 is further configured to calculate a gyro offset value from the captured gyro sensor data using a variance-limited averaging method. The calibration module 106 is triggered upon detection of a steady state of the device 100 . Here, the steady state is the state when the device 100 is not in motion. In an exemplary embodiment, the state of device 100 may be identified by monitoring the 3-axis gyroscope norm, as described by Equation 3, when the gyroscope norm is below a predefined threshold (e.g., 1 dps) for a continuous predefined period of time (e.g., 0 .5 seconds). $$nOMOtiOnfla{G}_k=\mathrm{1,}falls{G}_{nO\mathrm{right}m}<G_{tH\mathrm{right}e}f\ddot{\mathrm{and}}\mathrm{right}allevO\mathrm{right}He\mathrm{right}iGenZeitscH\mathrm{right}ittek\mathrm{i.e}e\mathrm{right}Da\mathrm{and}e\mathrm{right}{T}_{tH}$$

whereby $$G_{nO\mathrm{right}{m}_k}=\sqrt{G_x^2+G_y^2+G_{\mathrm{e.g}}^2}$$

• g _thre = gyronorm threshold for no movement
• T _Th = Minimum time limit without motion

$$Vorbelastungsvertrauen,R=P_1*t_b+P_2*t_b^2$$

Further, the calibration module 106 is configured to calculate the bias by averaging the gyro samples that have a variance less than a threshold (shown in Equation 1). In one embodiment, the calibration module is configured to calculate a confidence level by monitoring the steady state time of the device 100 as described in Equation 2 . $$VO\mathrm{right}belast\mathrm{and}nG=\frac{1}{N}{\displaystyle {\sum }_{i=1}^N{w}_{avG}},w_{avG}\le V_{tH\mathrm{right}e}$$

$$VO\mathrm{right}belast\mathrm{and}nGsve\mathrm{right}t\mathrm{right}a\mathrm{and}en,R=P_1*t_b+P_2*t_b^2$$

• W_var = Varianz von Gyroskopsensordatenabtastwerten von 0,5 Sekunden
• V_thre = Schwellenwert für die Varianzbegrenzung
• N = 0,5 * f_s * t_b
• f_s = Abtastintervall
• t_b = Vorbelastungsdauer in Sekunden
• p₁, p₂ = Koeffizient des Vorbelastungsvertrauens, abgeleitet durch Simulationsexperiment

applies here

• W _var = variance of 0.5 second gyroscope sensor data samples
• V _thre = variance limit threshold
• N = 0.5 * _fs * _tb
• f _s = sampling interval
• t _b = preload duration in seconds
• p ₁ , p ₂ = coefficient of preload confidence derived by simulation experiment

The TCO estimation module 108 is configured to estimate the temperature compensation offset (TCO) factor using a first Kalman filter. The TCO estimation module 108 is triggered upon detection of a temperature change and a high preload confidence of the device 100 .

In one embodiment, the TCO estimation module 108 is further configured to maintain a table of gyroscope offset values for various temperature values. In this embodiment, if there is a temperature change and it is not listed in the table, the TCO estimation module 108 updates the table and estimates the new TCO factor. Basis models behind the Kalman filter are defined by Equation 4. The model equation is as follows $$\begin{array}{cc}MO\mathrm{i.e}ell& \Rightarrow b_{T2}=b_{T1}+tcO*\mathrm{\Delta }T\\ Z\mathrm{and}stan\mathrm{i.e}& \Rightarrow x_k={\left[\begin{array}{cccccc}{b}_x& tc{O}_x& b_y& tc{O}_y& b_{\mathrm{e.g}}& tc{O}_{\mathrm{e.g}}\end{array}\right]}^T\\ MessvektO\mathrm{right}& \Rightarrow {\mathrm{e.g}}_k={\left[\begin{array}{ccc}{b}_x& b_y& b_{\mathrm{e.g}}\end{array}\right]}^T\\ MessmO\mathrm{i.e}ell& \Rightarrow H=\left[\begin{array}{cccccc}1& \Delta T& 0& 0& 0& 0\\ 0& 0& 1& \mathrm{\Delta }T& 0& 0\\ 0& 0& 0& 0& 1& \mathrm{\Delta }T\end{array}\right]\\ Z\mathrm{and}stan\mathrm{i.e}s\ddot{\mathrm{and}}be\mathrm{right}GanGsmO\mathrm{i.e}ell& \Rightarrow A=I\\ Ste\mathrm{and}e\mathrm{right}einGabe& \Rightarrow {\mathrm{and}}_k=\mathrm{0,}keineSte\mathrm{and}e\mathrm{right}einGabe\\ Ste\mathrm{and}e\mathrm{right}mO\mathrm{i.e}ell& \Rightarrow B=1\end{array}\}$$

• b = Vorbelastung
• tco = von der Temperatur abhängiger Koeffizient
• ΔT = Temperaturänderung
• I = Einheitsmatrix

here

• b = preload
• tco = temperature dependent coefficient
• ΔT = temperature change
• I = identity matrix

In one embodiment, the relationship between temperature and gyro bias is piecewise linear and this is achieved based on the simulation experiment using 180 sensor samples. During initialization, the TCO estimation module 108 runs for all points in the offset temperature table. In addition, the TCO estimation module 108 reruns for any new entry addition to the table. This can also be optimized by storing and recalling the temperature table over the power cycle.

The bias estimator (112) is configured to calculate a temperature compensated gyro offset value by combining an estimated predicted TCO offset and gyro offset value using a second Kalman filter.

In general, Kalman filtering methods are well known in the art. The principle of the working and implementation of the Kalman filtering method according to this disclosure is that during the iterative Kalman filtering method the measured value of the parameter is received again in each iteration. In operation of the Kalman filter, an initial covariance matrix and the initial state of the device 100 are received from off-line measurement experiments performed in the laboratory setup during the manufacture of the detection module 104 . Using temperature change values and the measured preload, the Kalman filter estimates the TCO of the gyroscope 102. The basic model to be fitted in the Kalman filter uses two different sets of steps, the first is a prediction step and the second is a correction step. In an exemplary embodiment, sets of equations for both the prediction step and the correction step are applied in every kth state. The prediction step is performed when a change in temperature is observed. To predict when the temperature change is received, the Kalman filter forward projects the state according to the equation below: $$P_k^{-}=A{P}_{k-1}^{-}{A}^T+Q$$

Also in this embodiment, the Kalman filter in this prediction step projects the error covariance ahead of time according to the following equation: $$K_k=P_k^{-}{H}^T{\left(H{P}_k^{-}{H}^{T^{-}}+R\right)}^{-1}$$

The Kalman filter correction step is a measurement update step, in this step the Kalman filter calculates the Kalman gain as explained below: $${}^{{\mathrm{e.g}}_k}\widehat{x}{}_k^{-}={\widehat{x}}_k^{-}+K_k\left({\mathrm{e.g}}_k-H{\widehat{x}}_k^{-}\right)$$

Further, using the Kalman gain, the Kalman filter estimates the measurement and updates it accordingly according to the following equation: $${}^{\text{'}}P{}_k=\left(1-K_{k}H\right)P_k^{-}$$

Furthermore, in the correction step, the Kalman filter calculates the error covariance according to the equation: $$\widetilde{x_k}=A\widetilde{x_{k-1}}+B{\mathrm{and}}_k$$

• x = die vorherige Schätzung, die sozusagen die grobe Schätzung vor der Messaktualisierungskorrektur bedeutet
• P_k = die vorherige Fehlerkovarianz
• X_k = die Schätzung von x zur Zeit k
• K_k = Kalman-Verstärkung

applies here

• x = the previous estimate, which means, so to speak, the rough estimate before the measurement update correction
• P _k = the previous error covariance
• X _k = the estimate of x at time k
• K _k = Kalman gain

If there is a temperature change, the TCO estimation module 108 estimates the gyro bias based on the TCO. A lower confidence bias from the offset estimator is also used here to estimate a better gyro bias.

The basic model used for the Kalman filter prediction and Kalman filter correction steps can be detailed as the following equation: $$\begin{array}{cc}MO\mathrm{i.e}ell& \Rightarrow b_{T2}=b_{T1}+tcO*\mathrm{\Delta }T\\ Z\mathrm{and}stan\mathrm{i.e}& \Rightarrow x_k={\left[b_x.b_y,b_{\mathrm{e.g}}\right]}^T\\ Ste\mathrm{and}e\mathrm{right}inGabe& \Rightarrow {\mathrm{and}}_k=\Delta T\\ Ste\mathrm{and}e\mathrm{right}mO\mathrm{i.e}ell& \begin{array}{l}\Rightarrow B=tcO\Rightarrow vOn\mathrm{i.e}e\mathrm{right}TCO-\\ ScH\ddot{a}t\mathrm{e.g}ein\mathrm{right}icHt\mathrm{and}nG\end{array}\\ MessvektO\mathrm{right}& \Rightarrow {\mathrm{e.g}}_k={\left[\begin{array}{ccc}{b}_x& b_y& b_{\mathrm{e.g}}\end{array}\right]}^T\\ Z\mathrm{and}stan\mathrm{i.e}s\ddot{\mathrm{and}}be\mathrm{right}GanGsmO\mathrm{i.e}ell& \Rightarrow A=I\\ MessmO\mathrm{i.e}ell& \Rightarrow H=I\end{array}\}$$

• b = Vorbelastung
• tco = von der Temperatur abhängiger Koeffizient
• ΔT = Temperaturänderung
• I = Einheitsmatrix

applies here

• b = preload
• tco = temperature dependent coefficient
• ΔT = temperature change
• I = identity matrix

The first and second Kalman filters are configured to select an initial value of the state and the covariance matrix thereof by performing sensor characterization. The Kalman filter uses previous values in measurement update equations (correction step).

In one embodiment, offset values for gyroscope 102 follow a piecewise linear relationship with temperature. In general, piecewise linear estimation is a technique for generating a function that adapts a nonlinear objective function by adding additional binary variables, continuous variables, and constraints to reformulate the original problem. Further, in this embodiment, the gyroscope offset values exhibit a hysteresis relationship as temperature values increase (ie, rise) and decrease (ie, sink) (as in 3 shown). In such situations of varying gyroscope temperature values, Kalman filters are used to estimate the TCO and gyro offset values.

In operation, the preload of the gyroscope 102 exhibits a hysteresis relationship with respect to gyroscope temperature variation. In this embodiment, the control unit 110 maintains two separate ones Tables for the gyroscope bias corresponding to each monitored temperature value. Here temperature values are provided by the acquisition module 104 . These two tables are maintained for the upward direction and downward direction of variation in temperature values, respectively. In the calibrator, the direction of temperature movement is monitored and respective table values are updated/used in the first Kalman filter and second Kalman filter.

2 12 illustrates a flowchart of a method 200 for determining a compensation offset value in gyroscope 102 of device 100 in accordance with the present invention. In step 201, data from gyroscope 102 and acquisition module 104 is acquired and pre-processed. The device 100 is stopped for a first predefined amount of time when the device 100 is powered on, and a first signal is received by the gyroscope 102 during the steady state of the device. In step 202, the gyroscope offset value is calculated from the acquired data in step 201 using a variance-limited averaging method. in step 203 the temperature compensation offset factor (TCO factor) is estimated using the first Kalman filter. In step 204, the temperature compensated gyro offset value is calculated by combining the estimated predicted TCO offset in step 203 and the gyro offset value in step 202 using a second Kalman filter. The device 100 continuously monitors the process and the method starts from 201 and runs in a loop.

It should be understood that the embodiments set forth in the above specification are illustrative only and do not limit the scope of this invention. Many such embodiments and other modifications and changes in the embodiment described in the specification are contemplated. The scope of the invention is only limited by the scope of the claims.

QUOTES INCLUDED IN DESCRIPTION

This list of documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent Literature Cited

• US 9581466 B2 [0004]

Claims (9)

Device (100) for determining a temperature-compensated offset value in a gyroscope (102) of the device (100), the device (100) having a control unit (110) and a detection module (104) for detecting a temperature of the gyroscope (102), wherein the control unit (110) comprises: - a calibration module (106) configured to acquire and process data from the acquisition module (104) and gyroscope (102), the calibration module (106) being further configured to do so to calculate a gyroscope offset value using a variance-limited averaging method from the acquired gyroscope sensor data; characterized in that - the TCO estimation module (108) is configured to estimate a temperature compensation (TCO) offset factor using a first Kalman filter; - the bias estimator (112) is configured to calculate a temperature compensated gyro offset value by combining the estimated predicted TCO offset and gyro offset values using a second Kalman filter. Device (100) according to claim 1 wherein the device (100) is selected from a group of devices comprising a vacuum cleaner, a lawnmower, a robotic arm and the like. Device (100) according to claim 1 , wherein the calibration module (106) is triggered upon detection of a steady state of the device (100). Device (100) according to claim 1 wherein the calibration module (106) is further configured to calculate a bias by averaging samples from the gyroscope (102) that have a variance of less than a threshold. Device (100) according to claim 1 , where the gyroscope offset values exhibit a hysteresis relationship as temperature values increase or decrease. Device (100) according to claim 1 , wherein the first and second Kalman filters are configured to select an initial value of the state and a covariance matrix thereof by performing sensor characterization for both Kalman filters. Device (100) according to claim 1 wherein the TCO estimation module (108) is triggered upon detection of a change in temperature and a high preload confidence of the device (100). Device (100) according to claim 1 wherein the TCO estimation module (108) is further configured to maintain and update a table of gyroscope offset values for different temperature values. A computer-implemented method (200) for determining a temperature compensation offset value in a gyroscope (102) of a device (100), comprising the steps of: acquiring and preprocessing (201) data from the acquisition module (104) and gyroscope (102); calculating (202) a gyroscope offset value using a variance-limited averaging method from the acquired data in step (201); estimating (203) a temperature compensation (TCO) offset factor using a first Kalman filter; and calculating (204) a temperature compensated gyro offset value by combining an estimated predicted TCO offset and gyro offset value using a second Kalman filter.

Images