KR101318397B1 - Method and apparatus for tremor signal estimation for real time application and robot surgery system - Google Patents

Abstract

Disclosed are a method and apparatus for estimating tremor for real time application and a robotic surgical system. Method for estimating the vibration from the accelerometer for real-time application according to the present invention, comprising measuring the vibration using the accelerometer, obtaining the measurement value information for the vibration; And applying the measured value information to a Bandlimitied Multiple Fourier Linear Combiner (BMFLC) -Kalman algorithm or a Bandlimitied Multiple Fourier Linear Combiner (BMFLC) -Recursive Least Squares (RLS) algorithm to estimate the shake signal mathematically. It is characterized by.

Description

METHOD AND APPARATUS FOR TREMOR SIGNAL ESTIMATION FOR REAL TIME APPLICATION AND ROBOT SURGERY SYSTEM}

The present invention relates to a method and apparatus for estimating a shaking signal for real time application and a robotic surgical system using the same. More specifically, the present invention relates to a method and apparatus for estimating a shake signal for real-time application by applying a shake signal measured from an accelerometer to a single step algorithm for shake estimation, and a robotic surgery system using the same.

The surgical procedure requires a high degree of expertise and accuracy, which can lead to deterioration of concentration, such as shaking hands, because the operation must be performed for a long time. In particular, when performing small lesions such as microsurgery, effectively overcoming these problems is the most important factor in the outcome of the surgery.

Accordingly, accurate surgical techniques using surgical robots have been developed. Accurate filtering techniques of tremor (particularly physiological tremors) are very important in the surgical field using robots. To this end, it is required to develop a robust algorithm of a single step for applying the measured vibration using an accelerometer in real time.

이다. 그러나, 생리적 떨림은 수술시 부정확한 문제를 초래할 수 있어, 수술용 로봇을 이용한 수술시에 발생하는 생리적 떨림의 보상은 매우 중요하다. 예를 들면, 로봇에 고정된 도구를 이용한 망막 미세수술 시 외과적인 생리적 떨림을 제거할 수 있는 수단이 개발되어 왔다. 일반적인 생리적 떨림은 각각의 중심축에서 6~15HZ 밴드대역에서 100um의 크기의 진폭을 가진다. 이러한 생리적 떨림은 기술적으로 다른 문제를 나타낸다. 왜냐하면 로봇을 이용한 수술시에 생리적 떨림은 실시간 적용되며 고주파 대역이기 때문이다. 이런 수술용 로봇을 이용한 수술시에 발생하는 생리적 떨림의 해결을 위해 오류 보상 제어 루프가 실시간으로 수행되어야 한다. 또한 시스템은 의식적/무의식적인 떨림을 하나의 샘플링 사이클 동안 구별하여야 가장 적합하도록 오류를 제거할수 있다.In general, the positional accuracy during surgery

to be. However, physiological tremors may cause inaccurate problems during surgery, so compensation of physiological tremors occurring during surgery using a surgical robot is very important. For example, a means has been developed to remove surgical physiological tremors in retinal microsurgery using robot-fixed tools. Typical physiological tremors have amplitudes of 100 μm in the 6-15 HZ band at each central axis. This physiological tremor presents a technically different problem. This is because the physiological vibration is applied in real time and the high frequency band is used in the robotic surgery. In order to solve the physiological vibration occurring during the operation using such a surgical robot, an error compensation control loop should be performed in real time. In addition, the system must distinguish between conscious and unconscious tremors during one sampling cycle in order to eliminate the error to best fit.

For example, physiological tremors are between 6 and 15 HZ, and surgical hand movements during microsurgery are nearly less than 0.5 to 1 HZ. Although linear filters perform successful image stabilization, the inherent time delay is a major drawback and requires a zero phase filter. The results show that a delay of 30ms also significantly reduces performance in robotic surgery.

To overcome this delay, weighted-frequency Fourier linear combiner (WFLC) algorithms and Bandlimitied multiple Fourier linear combiner (BMFLC) algorithms are used as adaptive algorithms. Can be.

WFLC is an adaptive algorithm for modeling quasi-periodic signals as sinusoids. It tracks the frequency, magnitude, and phase of quasi-periodic signals. This WFLC includes a frequency adaptation procedure in a Fourier Linear Combiner (FLC). The main drawbacks of these WFLCs are that they track signals at modulated frequencies, high frequency noise affects frequency adaptation, and force the use of pre-filtering (band-pass filters) that cause delays in the filtering process. There is this. Therefore, two-stage algorithms have been developed for the estimation of vibration using WFLC-KALMAN, which combines a band-pass filter and a Kalman filter with WFLC.

 Band-limited multi-linear Fourier couplers (BMFLCs) have also been developed as adaptive algorithms, tracking multiple dominant frequencies of the tremor for accurate tremor filtering. The adaptation process is achieved by using Least Mean Square (LMS) optimization, which is a least mean square algorithm, similar to WFLC. Such BMFLC can be separated into conscious / unconscious operation in a single step, and can provide a tremor signal in the displacement domain. For this reason, BMFLC is an ideal choice for shaking filtering when data is sensed in an accelerometer.

In the present invention, the conventional BMFLC algorithm uses a BMFLC-RLS and a Kalman filter including recursive least squares (RLS) instead of relying on the least mean square (LMS) to further improve through a modified process. By using two new methods of combined BMFLC-Kalman, we propose an accurate vibration estimation.

Accordingly, the present invention has been made to solve the above-mentioned problems of the prior art, and the object of the present invention is to provide a faster and more accurate estimation of tremor by combining RLS (recursive least squares) with a Bandlimitied multiple Fourier linear combiner (BMFLC) algorithm. have.

In addition, another object of the present invention is to combine the Kalman filter with the BMFLC algorithm to estimate the vibration more quickly and accurately.

In accordance with another aspect of the present invention, there is provided a method of estimating a shake signal, by measuring a shake using an accelerometer, and obtaining measured value information about the shake; and using the measured value information as a BMFLC (Bandlimitied Multiple Fourier). And applying the linear combiner) -Kalman algorithm or BMFLC (Bandlimitied Multiple Fourier Linear Combiner) -RLS (Recursive Least Squares) algorithm to estimate the shake signal mathematically. And BMFLC is represented by the following equation (1),

(One)

및 The BMFLC-Kalman algorithm is an updated covariance matrix for obtaining the adaptive weights a _rk and b _rk corresponding to the frequency W _r at the instantaneous value k in the equation.

And

을

식에 대입하는 알고리즘이며, 상기 BMFLC-RLS 알고리즘은 상기 수학식(1)에서 상기 a_rk, b_rk 를 구하기 위하여 갱신된 상관 행렬

및 계산된 칼만이득

을

식에 대입하는 알고리즘인것을 특징으로 할수 있다. 또한, 상기 알고리즘을 이용하여 의식적인 떨림신호와 무의식적인 떨림신호로 분류하는 단계를 더 포함하는 것을 특징으로 할수 있다. 그리고 상기 무의식적인 떨림신호는 생리적인 떨림을 포함하는 것을 특징으로 할수 있다. 또한, 상기 알고리즘은 프리필터(Pre-filter)를 포함하지 않는 것을 특징으로 할수 있다. 그리고 상기 알고리즘은 역행렬연산을 수행하지 않는 것을 특징으로 할수 있다.Calculated Kalman

of

The BMFLC-RLS algorithm is a correlation matrix updated to obtain a _rk and b _rk in Equation (1).

And calculated Kalman

of

It can be characterized by an algorithm that assigns to an expression. The method may further include classifying the signal into a conscious tremor signal and an unconscious tremor signal. The unconscious tremor signal may be characterized by including physiological tremor. In addition, the algorithm may be characterized as not including a pre-filter. The algorithm may be characterized as not performing inverse matrix operation.

A vibration signal estimating apparatus according to an embodiment of the present invention for achieving the above object is a measurement unit for obtaining the vibration signal measurement value information; and the measured value information BMFLC (Bandlimitied Multiple Fourier Linear Combiner) -Kalman algorithm or BMFLC ( A vibration signal estimator for estimating the measured value information by applying it to a Bandlimitied Multiple Fourier Linear Combiner (RLS) -RLS algorithm; . And BMFLC is represented by the following equation (1),

상기 BMFLC-Kalman 알고리즘은, 상기 수학식에서 순시값 k일때의 주파수 W_r에 부합하는 적응 가중치 a_rk, b_rk 를 구하기 위하여 갱신된 공분산 행렬

및 계산된 칼만이득

을

식에 대입하는 알고리즘이며, 상기 BMFLC-RLS 알고리즘은 상기 수학식(1)에서 상기 a_rk, b_rk 를 구하기 위하여 (One)

The BMFLC-Kalman algorithm is an updated covariance matrix for obtaining the adaptive weights a _rk and b _rk corresponding to the frequency W _r at the instantaneous value k in the equation.

And calculated Kalman

of

Algorithm to be substituted into the equation, the BMFLC-RLS algorithm is to calculate the a _rk , b _rk in the equation (1)

및 계산된 칼만이득

을

식에 대입하는 알고리즘인것을 특징으로 할수 있다. 또한 상기 떨림 신호추정부는, 상기 알고리즘을 적용하여 떨림 신호와 무의식적인 떨림 신호를 분류하는 떨림신호 처리부를 포함하는 것을 특징으로 할수 있다. 그리고 상기 무의식적인 떨림 신호는 생리적인 떨림을 포함하는 것을 특징으로 할수 있다. 또한 상기 진동 처리부는 프리필터(Pre-filter)를 포함하지 않는 것을 특징으로 할수 있다. 상기 알고리즘은 역행렬연산을 수행하지 않는 것을 특징으로 할수 있다. 그리고 상기 측정값 정보는 가속도계에서 측정되는 것을 특징으로 할수 있다.Updated correlation matrix

And calculated Kalman

of

It can be characterized by an algorithm that assigns to an expression. In addition, the shake signal estimation may include a shake signal processor for classifying a shake signal and an unconscious shake signal by applying the algorithm. The unconscious tremor signal may be characterized by including physiological tremor. In addition, the vibration processing unit may be characterized in that it does not include a pre-filter. The algorithm may be characterized as not performing inverse matrix operation. The measured value information may be measured by an accelerometer.

Robot operation system according to an embodiment of the present invention for achieving the above object is a measurement unit for obtaining the vibration signal measurement value information; The measured value information is applied to a Bandlimitied Multiple Fourier Linear Combiner (BMFLC) -Recursive Least Squares (RLS) algorithm or a Bandlimitied Multiple Fourier Linear Combiner (RLMLC) -Recursive Least Squares (RLS) algorithm. A shake signal estimating unit for estimating; A tremor signal processor for classifying a tremor signal and an unconscious tremor signal using the algorithm; A driving unit for driving a surgical robot; and a control unit for controlling to transmit only the conscious shaking signal to the driving unit. And BMFLC is represented by the following equation (1),

(One)

The BMFLC-Kalman algorithm calculates adaptive weights a _rk and b _rk corresponding to the frequency W _r at the instantaneous value k in the above equation.

및Updated Covariance Matrix

And

을Calculated Kalman

of

식에 대입하는 알고리즘이며, 상기 BMFLC-RLS 알고리즘은 상기 수학식(1)에서 상기 a_rk, b_rk 를 구하기 위하여 갱신된 상관 행렬

및

The BMFLC-RLS algorithm is a correlation matrix updated to obtain a _rk and b _rk in Equation (1).

And

을Calculated Kalman

of

식에 대입하는 알고리즘인것을 특징으로 할수 있다. 또한 상기 무의식적인 떨림신호는 생리적인 떨림을 포함하는 것을 특징으로 할수 있다.

It can be characterized by an algorithm that assigns to an expression. In addition, the unconscious tremor signal may be characterized by including physiological tremor.

As described above, according to various embodiments of the present invention, by combining RLS with the BMFLC algorithm, the shake signal can be estimated quickly and accurately, and the shake can be effectively filtered by classifying the conscious / unconscious motion into a single step.

In addition, according to various embodiments of the present invention, by combining a Kalman filter with the BMFLC algorithm, the shake signal may be estimated quickly and accurately, and the shake may be effectively filtered by classifying the conscious / unconscious motion into a single step.

1 is a view showing the configuration of a robot surgery system for real-time application according to an embodiment of the present invention.
2 is a diagram illustrating a configuration of a shake signal estimating apparatus according to an embodiment of the present invention.
3 is a diagram illustrating a configuration of a bandlimitied multiple fourier linear combiner (BMFLC) according to an embodiment of the present invention.
4 is a diagram for describing a shake signal processing process using a BMFLC-Kalman filter according to an embodiment of the present invention.
FIG. 5 is a diagram illustrating a shake signal measurement value (Raw Data) and conscious shaking signal data (Intended Motion) in the experiment.
FIG. 6 is a diagram illustrating an error value of a shake signal estimation according to each algorithm in an experiment.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings. FIG. 1 is a diagram illustrating a configuration of a robot surgery system for real-time application according to an embodiment of the present invention. Referring to FIG. 1, the robotic surgery system may include a driving unit 100, a measuring unit 200, a shaking estimating unit 300, and a shaking processing unit 400 and a control unit 500.

First, the driving unit 100 according to an embodiment of the present invention includes an apparatus for driving a posture of a robot (particularly, a surgical robot) that is a controlled object. For example, it may be a known actuator including one or more motors capable of rotational motion and one or more linear motion means capable of linear motion.

Next, the measurement unit 200 according to an embodiment of the present invention, when the operation is performed using a vibration, for example, a robot that is input from the outside, the movement of the hand to operate during the microsurgery may also be shaken Unintentional tremors of the hands during microsurgery (for example, tremors, pulse tremors, gravitational tremors, etc.) can all be shakes. The measuring unit includes a device for measuring such a shake signal. An example may be an accelerometer, but the present invention is not limited thereto.

Next, the vibration signal estimation unit 300 according to an embodiment of the present invention applies a preset algorithm from the vibration signal measurement value measured by the measurement unit 200 to estimate the measurement value information as an equation. . Such an algorithm can construct a BMFLC-RLS that includes recursive least squares (RLS) instead of relying on a conventional mean mean square (LMS) for the modification of the conventional BMFLC algorithm. In addition, an advanced algorithm can be constructed, which is a BMFLC-Kalman filter combining BMFLC and Kalmam filters for more accurate vibration estimation. This vibration signal estimation unit 300 will be more clearly described with reference to FIG.

Next, the tremor signal processor 400 according to an embodiment of the present invention is a conscious tremor signal (e.g., intentional movement for surgery at the time of microsurgery) from an equation estimated by the tremor signal estimation unit 300. And unconscious tremor signals (eg tremors, tremors caused by hydration, pulses, tremors caused by gravity). The algorithm according to the present invention, that is, BMFLC-RLS or BMFLC-Kalman filter combined with BMFLC and Kalman Filter for more accurate vibration estimation, can distinguish between conscious and unconscious shaking without pre-filtering. The shaking signal processing unit 400 will be more clearly described with reference to FIG. 2.

Next, the control unit 500 according to an embodiment of the present invention may perform a function of controlling the driving unit 100, the measuring unit 200, the shaking signal estimating unit 300, and the shaking signal processing unit 400. have. The controller 500 may control a control signal exchanged or flowing between the components and other components in a system not shown to allow each component to perform its own function. In addition, the controller 500 may serve to control only the conscious shaking among the signals distinguished by the shaking signal processing unit 400 from the robot surgery system to be transmitted to the driving unit 100 for driving the robot. In addition, the estimation value of the shake signal estimator provides only the information of the acceleration domain by using an accelerometer. In order to provide the signal to the driver 100, a change to the position domain is required. ) Changes the area accordingly. This change will be described in detail below if it can be carried out through integration. The controller 400 may be a central control module.

Tremor Of the estimator 300  Configuration

When the vibration signal estimator 300 according to the present invention is described in more detail, a band-limited multi in which sine and cosine are combined in series to estimate a vibration signal in a predefined band [W1-Wn] It can be configured as a bandlimitied multiple Fourier linear combiner (BMFLC). In more detail, it can be understood by Equation 1 and FIG. 3 below.

[Equation 1]

Here, Y _k means a signal estimated at the instantaneous value k. a _rk and b _rk represent an adaptive weight corresponding to the frequency W _r at the instantaneous value k. Here, r = 1 to n, because the vibration signal is estimated in the predefined band [W1-Wn].

3 is a diagram illustrating a configuration of BMFLC-Kalman or BMFLC-RLS.

Referring to Figure 3, the structure of the BMFLC algorithm can be clearly seen, where S _K is the reference signal, W _k Estimates the input signal amplitude and phase with an adaptive vector. Y _k Is an output signal, and e _k represents an error signal between the output signal and the reference signal. In this case, the error signal may be compensated by using any one or more of RLS and Kalman fliter.

Recursive Least Squares ( BMFLC - RLS Containing BMFLC

The following detailed description looks at an algorithm in which RLS is combined with BMFLC, which can replace the BMFLC-LMS algorithm, for further improved performance by the present invention.

 RLS (Recursive Least Squares) is an adaptive algorithm used by continuously modifying filter coefficients. It also minimizes the weighted least square error function associated with the input signal. In general, the RLS algorithm requires inverse matrix computation, which may apply Kalman gains to avoid direct inverse computation. The inverse matrix may be obtained by repeatedly performing a Kalman gain vector and updating the inverse of a correlation matrix. The RLS equation is represented by the following equation.

&Quot; (2) "

Calculated Kalman Gain K _k

&Quot; (3) "

Updated BMFLC Weights

&Quot; (4) "

Updated Correlation matrix P _k

는 망각 인자(forgetting factor)로 (전형적으로 0.9<=

<=1)범위를갖는다.

는 기준 입력 벡터이다.

Updated here

Is the forgetting factor (typically 0.9 <=

<= 1) has a range.

Is the reference input vector.

를 구하여 BMFLC 식에 a_rk, b_rk 값을 적용시켜서 떨림신호를 추정한다.
According to the above equations, the adaptive weights a _rk , b _rk corresponding to the frequency W _r with the instantaneous value k. Use RLS to get

Find the BMFLC expression in a _rk , b _rk Apply the value to estimate the shake signal.

Kalman Filter ( BMFLC - Kalman Containing BMFLC

와 같다.

는 상태천이(state transition)시에 상태에러(state error)를 의미한다. 상기의

로부터

의 상태공간 형식으로 다시 쓰여질수 있다. 여기서

는 측정에러(mesurement error)를 의미한다. 추정을 위해서는

와

는 비상관관계(uncorrelated), 제로평균(zero mean), 공분산이

인 Gaussian White Noise라고 가정한다. 만약 이러한 가정이 지켜지지 않더라도, 칼만 필터는 선형추정기 중에 가장 적은 에러를 제공한다.

여기서

는 기재값을 의미한다. BMFLC- Kalman는 하기 수학식과 같이 공식화 할 수 있다.

즉 기준입력의 시퀀스 이다.The Kalman filter is an important component for estimating the state in dynamic systems with iterative processing. The Kalman filter formula can be applied generally in state space. In order to model BMFLC in the state space format, the adaptation vector W _k may be considered as a state vector.

Same as

Denotes a state error during a state transition. The above

from

Can be rewritten in the form of statespace. here

Means a measurement error. For estimation

Wow

Is the uncorrelated, zero mean, and covariance

Assume Gaussian White Noise. If this assumption is not followed, the Kalman filter provides the least error among linear estimators.

here

Means a description value. BMFLC-Kalman can be formulated as follows.

That is, it is a sequence of standard inputs.

&Quot; (5) &quot;

Calculated Kalman Gain K _k

&Quot; (6) &quot;

Updated BMFLC Weights

[Equation 7]

Updated Covariance Matrix

를 구하여 BMFLC 식에 a_rk, b_rk 값을 적용시켜서 떨림신호를 추정한다.According to the above equations, the adaptive weights a _rk , b _rk corresponding to the frequency W _r with the instantaneous value k. Use kalman method to save

Find the BMFLC expression in a _rk , b _rk Apply the value to estimate the shake signal.

Thus, the BMFLC-RLS and BMFLC-Kalman algorithms of the present invention do not require an inverse because BMFLC has one output. As a result, the algorithm of the present invention has a feature that can be quickly calculated in real time.

Configuration of the shaking signal processor 400- distinction between conscious shaking and unintended shaking

Successful tremor compensation applied to surgical robots can include conscious tremors (eg, deliberate movements for surgery during microsurgery) and unintentional tremors (eg, tremors caused by hydrocephalus, pulse, or gravity) from tremor signals. Tremor).

,적응 가중벡터

로 변형시켜 상기 알고리즘에 적용함으로써 pre-filtering 과정 없이 의식적/의도하지 않은 떨림을 분리할수 있다.In the present invention, in order to solve this problem, the shaking signal processor 400 may apply a weighted bias suitable for separating conscious / unintentional shaking from the shaking signal. At this time, unlike the WFLC and WFLC-Kalman filters for removing conscious operation, it is possible to prevent the need for pre-filtering (eg, low pass filter). In detail, the shaking signal processor 400 may include a reference input vector.

Adaptive weighting vector

By applying the algorithm to the algorithm, it is possible to separate conscious / unintentional shaking without pre-filtering.

Configuration of the control unit 500

와 같이 추정된 떨림을 두번 적분하여 가속도계에서 위치계로 변경시켜서 구할수 있다. The controller 500 may control a control signal exchanged or flowing between the components and other components in a system not shown to allow each component to perform its own function. In addition, the controller 500 may serve to control only the conscious shaking among the signals distinguished by the shaking signal processing unit 400 from the robot surgery system to be transmitted to the driving unit 100 for driving the robot. In addition, the estimation value of the shake signal estimator provides only the information of the acceleration domain by using an accelerometer. In order to provide the signal to the driver 100, a change to the position domain is required. ) Changes the area accordingly. Specifically, the integration into the location area

It can be obtained by integrating the estimated vibration twice from accelerometer to positioner.

 As described above, the accelerometer-based robot can measure vibration in accordance with the accelerometer to remove vibration in real time during microsurgery. Elimination of vibration can be performed by piezoelectric actuators in the displacement domain.

 In addition, the WFLC algorithm can provide an algorithm that can obtain conscious motion and vibration motion through two-stage filtering. In this case, BMFLC can confirm the conscious motion without prefilter.

4 is a diagram for describing a shake signal processing process using a BMFLC-Kalman filter according to an embodiment of the present invention.

 Referring to FIG. 4, a measurement value of a shake signal measured from an accelerometer may be classified into a shake signal estimation, a conscious motion, and an unconscious motion through a BMFLC-Kalman filter. Integral analysis allows the movement of unconscious movement from the accelerometer to the displacement zone. It will be apparent that the Bandlimitied multiple Fourier linear combiner (BMFLC) -recursive least squares (RLS) is equally applicable. In addition, when the BMFLC algorithm is applied as shown in FIG. 4, unlike the WFLC, the vibration signal may be directly applied without a pre-filtering process. This is an important issue in the special delay problem. In the case of WFLC, high-frequency noise affects frequency adaptation, so pre-filtering is required, but in the case of BMFLC, this pre-filtering process is not necessary.

Work according to the present invention Experimental Example

Position recording in the present invention was performed by the Micro Motion Sensing System (M ₂ S ₂ ). M ₂ S ₂ is a pair of position detectors are located at right angles to each other. It may also include an infrared diode that can track the three-dimensional displacement of the tip of the microsurgical tool in real time, the location of the tip being calculated from the center point of the position sensitive detector (PSD). The M ₂ S ₂ finely measures the movement (vibration) of the surgical robot according to the movement of the test subjects.

 In the present invention, the tremor signal was measured by six healthy novice microsurgeons and six healthy skilled microsurgeons. Each subject underwent a surgical needle with thumb and tongs, each running in two different surgical settings. First, the task of simply teaching the laser light (static task) and secondly, the task of tracking the moving laser light (dynamic task).

FIG. 5 is a diagram illustrating a shake signal measurement value (Raw Data) and conscious shaking signal data (Intended Motion) in the experiment. Referring to FIG. 5, the BMFLC-Kalman filtering of the shake signal raw data may be performed. The recording of the tremor signal by the experiment is shown in FIG. At this time, the conscious movement can be checked, and the conscious movement can be classified in the source data without pre-filtering. As can be seen in FIG. 5, the unfiltered tremor signal has a higher frequency than the conscious tremor signal indicated by dotted lines due to the unconscious tremor signal.

FIG. 6 is a diagram illustrating an error value of a shake signal estimation according to each algorithm in an experiment. Referring to FIG. 6, the results of all algorithms (5 types in total) during the first operation are shown. BMFLC-RLS and BMFLC-Kalman shown in FIGS. 5-e and 5-f are better than other algorithms. It shows that the filtered vibration has a signal.

 In addition, the root mean square error (RMS) for accurate performance evaluation of the estimation algorithm can be understood by Equation (8).

[Equation 8]

Where m is the sampling number and S _k is the input shake signal, Y _k Is the vibration signal estimated during the instantaneous value k. The average RMSE for accurate vibration estimation for all five algorithms can be understood by Table 1 below.

Mean RMS Error (um) and Standard Deviation WFLC WFLC-Kalman BMFLC BMFLC-RLS BMFLC-Kalman 6 novice targets

0.5781.065

0.578 0.747

0.43 0.512

0.303 0.08

0.05 0.004

0.002 6 specialists 0.956

0.441 0.632

0.263 0.408

0.201 0.076

0.051 0.003

0.002

In Table 1, all five algorithms were performed, measured by six healthy novice microsurgeons and six healthy skilled microsurgeons. Of the average RMSE and standard deviation of the five algorithms measured, it is clear that BMFLC-Kalman can provide the signal with the least error.

As a result, the proposed single-step algorithm can make accurate vibration estimation using the sensing data from the accelerometer and separate the intended motion from the unintended motion. The BMFLC algorithm including the conventional LMS can be replaced by the algorithm including the Kalman filter, which shows more improved error.

   In the present invention, data records from six specialist microsurgeries and six novice surgeries were collected and comprehensively compared to interpret the performance of all algorithms. It is clear that the tested methods BMFLC-Kalman and BMFLC-RLS are better than conventional WFLC-Kalman and BMFLC-LMS.

0.002(um) 보다 향상되며 실시간 추정이 가능하였다. In particular, of the five algorithms, BMFLC-Kalman performed vibration estimation more accurately, which means an average RMS error of 0.003.

Better than 0.002 (um), real-time estimation was possible.

Meanwhile, program codes for performing the above driving method may be stored in various types of recording media. More specifically, it may be a random access memory (RAM), a flash memory, a ROM (Read Only Memory), an EPROM (Erasable Programmable ROM), an Electrically Erasable and Programmable ROM (EEPROM), a register, a hard disk, a removable disk, And may be stored in various types of recording media readable by a terminal, such as a memory, a CD-ROM, and the like.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, but, on the contrary, It will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the present invention.

100:
200: measuring unit
300: shaking signal estimator
400: shake signal processing unit
500:

Claims (16)

Measuring shaking using an accelerometer to obtain measurement value information on shaking; and
The measured value information is BMFLC (Bandlimitied Multiple Fourier Linear).
Estimating a tremor signal by applying a Kalman algorithm or a Bandlimitied Multiple Fourier Linear Combiner (BMFLC) -Recursive Least Squares (RLS) algorithm;
Vibration signal estimation method comprising a. The method of claim 1,
BMFLC is represented by the following equation (1),
(One)

The BMFLC-Kalman algorithm calculates adaptive weights a _rk and b _rk corresponding to the frequency W _r at the instantaneous value k in the above equation.
Updated Covariance Matrix

And
Calculated Kalman

of

Algorithm to assign to an expression,
The BMFLC-RLS algorithm is used to calculate the a _rk and b _rk in Equation (1).
Updated correlation matrix

And
Calculated Kalman

of

A tremor signal estimation method, characterized in that the algorithm is substituted into an equation. The method of claim 1,
And classifying the signal into a conscious tremor signal and an unconscious tremor signal using the algorithm. The method of claim 3,
The unconscious tremor signal includes a tremor signal estimation method. The method of claim 1,
The algorithm of claim 1, wherein the algorithm does not include a pre-filter. The method of claim 1,
The algorithm of claim 1, wherein the algorithm does not perform inverse matrix operation. Measurement unit for obtaining the shaking signal measurement value information; And
The measured value information is BMFLC (Bandlimitied Multiple Fourier Linear).
A shaker signal estimator for applying to a Kalman algorithm or a Bandlimitied Multiple Fourier Linear Combiner (BMFLC) -Recursive Least Squares (RLS) algorithm to estimate the measured value information mathematically;
Vibration signal estimation device comprising a. The method of claim 7, wherein
BMFLC is represented by the following equation (1),
(One)

The BMFLC-Kalman algorithm calculates adaptive weights a _rk and b _rk corresponding to the frequency W _r at the instantaneous value k in the above equation.
Updated Covariance Matrix

And
Calculated Kalman

of

Algorithm to assign to an expression,
The BMFLC-RLS algorithm is used to calculate the a _rk and b _rk in Equation (1).
Updated correlation matrix

And
Calculated Kalman

of

A vibration signal estimating device, characterized in that the algorithm is substituted into an equation. The method of claim 7, wherein
The shaking signal estimation unit,
And a shake signal processor for classifying the shake signal and the unconscious shake signal by applying the algorithm. 10. The method of claim 9,
The involuntary tremor signal is a tremor signal estimation device, characterized in that it comprises a physiological tremor. The shake signal estimating apparatus of claim 9, wherein the shake signal processor does not include a pre-filter. The method of claim 7, wherein
And the algorithm does not perform inverse matrix operation. The method of claim 7, wherein
The measurement signal information device, characterized in that measured in the accelerometer. A measuring unit obtaining vibration signal measurement value information;
The measured value information is BMFLC (Bandlimitied Multiple Fourier Linear).
A shaker signal estimator for applying to a Kalman algorithm or a Bandlimitied Multiple Fourier Linear Combiner (BMFLC) -Recursive Least Squares (RLS) algorithm to estimate the measured value information mathematically;
A tremor signal processor for classifying a tremor signal and an unconscious tremor signal using the algorithm;
A drive unit for driving a surgical robot; and
And a control unit for controlling to transmit only the conscious shaking signal to the driving unit. 15. The method of claim 14,
BMFLC is represented by the following equation (1),
(One)

The BMFLC-Kalman algorithm calculates adaptive weights a _rk and b _rk corresponding to the frequency W _r at the instantaneous value k in the above equation.
Updated Covariance Matrix

And
Calculated Kalman

of

Algorithm to assign to an expression,
The BMFLC-RLS algorithm is used to calculate the a _rk and b _rk in Equation (1).
Updated correlation matrix

And
Calculated Kalman

of

Robot surgery system, characterized in that the algorithm to substitute in the equation. 15. The method of claim 14,
The unconscious tremor signal includes a physiological tremor system.

Images