A six-dimensional force sensor offset compensation method and system

Abstract

The present invention discloses a six-dimensional force sensor offset compensation method and system. Firstly, the output values ​​of the six-dimensional force sensor in different postures in a no-load state are collected, and the actual posture and measured output value of the six-dimensional force sensor are obtained. Then, the initial offset value of each axis of the six-dimensional force sensor in the real-time posture is obtained by a quadratic interpolation method. The actual force value is calculated by the measured output value and the initial offset value. The problem of excessive deviation between the output value of the six-dimensional force sensor and the actual value can be solved, thereby improving the control accuracy of the admittance control. Moreover, a relatively high accuracy can be obtained without establishing a complex mathematical model, thereby improving the offset compensation efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of mechanical sensors, and in particular to a six-dimensional force sensor offset compensation method and system. Background Art

[0002] The robotic arm is a type of robot, which is characterized by high control accuracy, convenient and flexible operation. It is being used more and more widely in the manufacturing industry, such as welding, polishing, spraying, assembly, handling, testing, grinding, etc. As the application scenarios of the robotic arm become more and more complex, many application scenarios require tracking and controlling the operating force to achieve the flexibility, accuracy and safety protection of the intelligent operation of the robotic arm.

[0003] In the compliant control of the robot, especially the admittance control, the data accuracy of the six-dimensional force / torque sensor at the end directly affects the accuracy of the admittance control of the robot. If the offset value of the force / torque sensor is not eliminated, the additional force in addition to the actual force will continue to be input into the admittance control model, which will eventually lead to a deviation in the output displacement of the admittance model. Therefore, if you want to improve the accuracy of the admittance model, reducing the output offset value of the sensor is an important means. The zero point of the force / torque sensor will drift due to temperature, installation preload, and sensor posture. The gravity of the robot's end effector is a relatively more important influencing factor. These factors will have various effects on the force sensor at different temperatures and different postures. It is worth noting that the deadweight of the force / torque sensor cannot be regarded as a mass point. Directly subtracting its gravity from the coordinate change will still have a large error, and it will change with the robot's posture, and cannot directly reflect the actual contact force between the robot and the environment. This error can be removed by adding a static threshold in a few application scenarios, but this increases the initial input of the force in disguise, and is extremely unstable, and cannot be accurately sensed. Traditional models all establish mathematical models to find out the magnitude of the impact of end gravity on the signal, and then remove other offsets that affect the zero point offset as fixed values. Multi-dimensional force / torque sensors detect force and torque data in different directions by detecting the deformation of multiple strain gauges. Therefore, the sensor's own weight cannot be completely regarded as a mass point for force decomposition. The different distribution of its own gravity in different postures has a certain difference in the impact on the force and torque signal.

[0004] In the study of zero-point correction and gravity compensation of force / torque sensors, researchers have proposed different methods. The traditional method is to model the self-weight part of the force sensor, and decompose its own gravity to the axes of the sensor in all directions through the sensor's own posture, so as to eliminate the influence of the sensor's own gravity. This method does not consider the influence of the deviation value of factors other than gravity, and often there will be some errors, which can only be eliminated directly as fixed values, but this greatly reduces the accuracy of the force sensor. Later, a method of zero-point correction and gravity compensation was proposed, which comprehensively considered the installation error of the robot, the zero-point drift of the force / torque sensor, and the gravity compensation of the end effector, and solved the compensation parameters through three random postures to achieve zero-point correction and gravity compensation.

[0005] The disclosed technologies can achieve basic force sensor deviation adjustment, but these methods treat the force / torque sensor as a point mass and do not consider that the six-dimensional force / torque sensor is the result of the joint processing of multiple sensor signal data when detecting force and torque information. Under different postures, the impact of gravity on the sensor is somewhat different from the model calculation results, resulting in a certain error in the final result, which cannot truly reflect the contact force situation. Summary of the invention

[0006] In view of this, the present invention provides a six-axis force sensor offset compensation method and system, which can solve the problem of excessive deviation between the output value of the six-axis force sensor and the actual value, thereby improving the control accuracy of admittance control.

[0007] The specific technical solution adopted by the present invention is as follows:

[0008] A six-dimensional force sensor offset compensation method for admittance control of a robotic arm, comprising:

[0009] Obtain output values ​​of the six-dimensional force sensor in different postures in a no-load state, and obtain the actual posture and measured output value of the six-dimensional force sensor;

[0010] According to the output value and the actual posture, a quadratic interpolation method is used to obtain an initial offset value of the measured output value;

[0011] The actual force value of the six-dimensional force sensor is obtained by calculation according to the measured output value and the initial offset value.

[0012] Furthermore, the output values ​​of the six-dimensional force sensor in different postures under no-load state are obtained as follows:

[0013] According to the preset pitch angle β and roll angle γ collection step of the end of the manipulator, the output values ​​of the six-dimensional force sensor in different postures in the no-load state in the working surface are collected, wherein: Within the range, n points are uniformly collected according to the collection step size, and within the range of -π≤γ≤π, m points are uniformly collected according to the collection step size, and a total of mn points are collected, each point corresponds to a set of output values, wherein m and n are both positive integers.

[0014] Furthermore, the collecting of output values ​​of different postures of the six-dimensional force sensor in the no-load state in the working surface includes:

[0015] When the pitch angle β increases by one acquisition step and enters the next acquisition posture, the roll angle γ uses the end acquisition position of the current acquisition posture as the start acquisition position of the next acquisition posture;

[0016] Alternatively, when the roll angle γ increases by one acquisition step and enters the next acquisition posture, the pitch angle β uses the end acquisition position of the current acquisition posture as the start acquisition position of the next acquisition posture.

[0017] Furthermore, the actual posture of the six-dimensional force sensor is obtained as follows:

[0018] The actual posture of the six-axis force sensor is obtained by kinematic forward solution according to the angles of each motor of the robotic arm, wherein the actual posture includes the actual pitch angle β s and the actual roll angle γ w .

[0019] Furthermore, the method of obtaining the initial offset value of the measured output value by using a quadratic interpolation method includes:

[0020] Determine an approximate interval of the actual attitude, wherein the approximate interval includes the actual pitch angle β s The approximate interval β i ≤β s ≤β i+1 and the actual roll angle γ w The approximate interval γ j ≤γ w ≤γ j+1 ,in i≤n-1, j≤m-1, i, j, m, n are all positive integers;

[0021] According to the output value of the six-dimensional force sensor corresponding to the endpoint value of the approximate interval, the initial offset value of the measured output value is obtained by using the quadratic interpolation method.

[0022] Furthermore, the obtaining of the initial offset value of the measured output value further comprises:

[0023] First get the pitch angle as β s The initial offset value at , and then the pitch angle is β s And the roll angle is γw The initial offset value at ;

[0024] Or, first get the roll angle as γ w The initial offset value at time, and then the roll angle is γ w And the pitch angle is β s The initial offset value at .

[0025] Furthermore, the actual force value of the six-dimensional force sensor is calculated based on the measured output value and the initial offset value as follows:

[0026] The actual force value of the six-dimensional force sensor is obtained by subtracting the initial offset value from the measured output value.

[0027] A six-dimensional force sensor offset compensation system for admittance control of a robotic arm, comprising:

[0028] A data acquisition module, used to acquire output values ​​of the six-axis force sensor in different postures in an unloaded state, and to acquire the actual posture and measured output value of the six-axis force sensor;

[0029] An offset calculation module, used to obtain an initial offset value of the measured output value by using a quadratic interpolation method according to the output value and the actual posture;

[0030] The actual output calculation module is used to calculate the actual force value of the six-dimensional force sensor according to the measured output value and the initial offset value.

[0031] Furthermore, the data acquisition module obtains the output values ​​of the six-dimensional force sensor in different postures in the no-load state as follows:

[0032] According to the preset pitch angle β and roll angle γ collection step of the end of the manipulator, the output values ​​of the six-dimensional force sensor in different postures in the no-load state in the working surface are collected, wherein: Within the range, n points are uniformly collected according to the collection step size, and within the range of -π≤γ≤π, m points are uniformly collected according to the collection step size, and a total of mn points are collected, each point corresponds to a set of output values, wherein m and n are both positive integers.

[0033] Furthermore, the offset calculation module uses a quadratic interpolation method to obtain an initial offset value of the measured output value, including:

[0034] Determine an approximate interval of the actual attitude, wherein the approximate interval includes the actual pitch angle β s The approximate interval β i ≤β s ≤β i+1 and the actual roll angle γ w The approximate interval γ j ≤γw ≤γ j+1 ,in i≤n-1, j≤m-1, i, j, m, n are all positive integers;

[0035] According to the output value of the six-dimensional force sensor corresponding to the endpoint value of the approximate interval, the initial offset value of the measured output value is obtained by using the quadratic interpolation method.

[0036] Beneficial effects:

[0037] (1) A six-axis force sensor offset compensation method utilizes the mapping relationship between the zero offset value of each axis of the six-axis force sensor, i.e., the initial offset value, and the posture of the six-axis force sensor, and obtains the initial offset value of each axis of the six-axis force sensor under the real-time posture through a quadratic interpolation method. The actual force value is calculated through the measured output value and the initial offset value. This method can solve the problem of excessive deviation between the output value of the six-axis force sensor and the actual value, thereby improving the control accuracy of the admittance control. Moreover, a relatively high accuracy can be obtained without establishing a complex mathematical model, thereby improving the efficiency of offset compensation.

[0038] (2) When the output value of the six-dimensional force sensor is different in different postures, when entering the next collection posture, the end collection position of the previous collection posture is used as the starting collection position of the next collection posture, ensuring that the data of all points can be collected in one traversal, without taking repeated routes, avoiding unnecessary movements, ensuring that a set of data can be collected in each step, and the collection efficiency is higher.

[0039] (3) The initial offset value is obtained by using the quadratic interpolation method. First, the initial offset value of one attitude angle state is obtained, and then the initial offset value of two attitude angle states is obtained. This can improve the calculation efficiency and has the effects of high real-time performance and small estimation error. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is a flow chart of the six-dimensional force sensor offset compensation method of the present invention;

[0041] Figure 2 It is a schematic diagram of posture scanning of the six-dimensional force sensor of the present invention;

[0042] Figure 3 It is a structural block diagram of the six-dimensional force sensor offset compensation system of the present invention. DETAILED DESCRIPTION

[0043] The present invention provides a six-axis force sensor offset compensation method and system. First, the output values ​​of the six-axis force sensor in different postures in a no-load state are collected, and the actual posture and measured output value of the six-axis force sensor are obtained. Then, the initial offset value of each axis of the six-axis force sensor in the real-time posture is obtained by a quadratic interpolation method. The actual force value is calculated by the measured output value and the initial offset value. The problem of excessive deviation between the output value of the six-axis force sensor and the actual value can be solved, thereby improving the control accuracy of the admittance control. Relatively high accuracy can be obtained without establishing a complex mathematical model, thereby improving the offset compensation efficiency.

[0044] The method of the present invention does not rely on accurate gravity modeling, but the six-dimensional force synthesizes all the offsets of the sensor into a unique variable for processing. The size of the variable is directly related to the posture of the sensor, and a large amount of data is collected in the working surface to fit all the offset values ​​in the entire working surface. That is, the six-dimensional sensor, under no-load conditions, scans the output value of six signals (six dimensions, each sensor has three horizontal force signals and three torque signals) of some working points in the entire working surface and maps the relationship between its posture. When calculating the offset value in a posture, first calculate the value of the nearby mapping point, and then interpolate to obtain the approximate value of the offset in the posture. The sensor data minus the offset value can obtain the magnitude of the actual contact force.

[0045] The controller of the robot arm can detect the angles of each joint motor in real time. Through these angles and the forward kinematics model, the posture of the end of the robot arm can be detected in real time. The end posture is expressed by the Cartesian coordinate system, that is, the yaw angle (yaw), pitch angle (pitch) and roll angle (roll), which are represented by α, β, and γ respectively. The force and torque components (i.e., offset values) of gravity acting on the sensor in all directions are only related to the roll angle and pitch angle, and other influencing factors are not related to the posture. Therefore, under constant temperature conditions, the zero bias value of the entire sensor is only related to the posture of the sensor, that is, the zero bias value of each force or torque is a unique mapping of the roll angle and the pitch angle. The zero bias value of each corresponding force and torque can be found through the angles of these two angles.

[0046] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0047] A six-dimensional force sensor offset compensation method for admittance control of a robotic arm, comprising:

[0048] Step 1: Obtain output values ​​of the six-dimensional force sensor in different postures under no-load state, and obtain the actual posture and measured output value of the six-dimensional force sensor;

[0049] In a specific embodiment, the output values ​​of the six-dimensional force sensor in different postures under no-load condition are obtained as follows:

[0050] According to the preset pitch angle β and roll angle γ collection step of the end of the manipulator, the output values ​​of the six-dimensional force sensor in different postures in the no-load state on the working surface are collected, where: Within the range, n points are uniformly collected according to the collection step length, and within the range of -π≤γ≤π, m points are uniformly collected according to the collection step length, and a total of mn points are collected, each point corresponds to a set of the output values, wherein m and n are both positive integers.

[0051] In a specific embodiment, when the pitch angle β increases by one acquisition step and enters the next acquisition posture, the roll angle γ uses the end acquisition position of the current acquisition posture as the start acquisition position of the next acquisition posture;

[0052] Alternatively, when the roll angle γ increases by one acquisition step and enters the next acquisition posture, the pitch angle β uses the end acquisition position of the current acquisition posture as the start acquisition position of the next acquisition posture.

[0053] This collection method can ensure that the data of all points can be collected in one traversal, without taking repeated routes, avoiding unnecessary movements, and ensuring that a set of data can be collected for each step, which is more efficient.

[0054] In a specific embodiment, the actual posture of the six-dimensional force sensor is obtained as follows:

[0055] The actual posture of the six-dimensional force sensor is obtained through kinematic forward solution according to the angles of each motor of the robotic arm, where the actual posture includes the actual pitch angle β s and the actual roll angle γ w .

[0056] Step 2: According to the output value and the actual posture, the initial offset value of the measured output value is obtained by using the quadratic interpolation method;

[0057] In a specific embodiment, the quadratic interpolation method is used to obtain the initial offset value of the measured output value, including: first, determining the approximate interval of the actual attitude, wherein the approximate interval includes the actual pitch angle β s The approximate interval β i ≤β s ≤β i+1 and the actual roll angle γ w The approximate interval γ j ≤γ w ≤γ j+1 ,in i≤n-1, j≤m-1, i, j, m, n are all positive integers;

[0058] Afterwards, according to the output value of the six-dimensional force sensor corresponding to the endpoint value of the approximate interval, the quadratic interpolation method is used to obtain the initial offset value of the measured output value.

[0059] In a specific embodiment, obtaining the initial offset value of the measured output value further includes: first obtaining the pitch angle β s The initial offset value at , and then the pitch angle is β s And the roll angle is γ w Or, first obtain the roll angle γ w The initial offset value at time, and then the roll angle is γ w And the pitch angle is β s The initial offset value at the time of calculation is used to improve the calculation efficiency, and at the same time, it has the effects of high real-time performance and small estimation error.

[0060] Step 3: Calculate the actual force value of the six-dimensional force sensor based on the measured output value and the initial offset value.

[0061] The specific calculation method is to subtract the initial offset value from the measured output value to obtain the actual force value of the six-dimensional force sensor.

[0062] The present invention also provides a six-dimensional force sensor offset compensation system for the admittance control of a robotic arm. Figure 3 FIG. 1 is a structural block diagram of the six-dimensional force sensor offset compensation system of the present invention, as shown in FIG. Figure 3 As shown, the offset compensation system includes: a data acquisition module, which is used to obtain the output values ​​of the six-dimensional force sensor in different postures in a no-load state, and obtain the actual posture and measured output value of the six-dimensional force sensor; an offset calculation module, which is used to obtain the initial offset value of the measured output value by using a quadratic interpolation method according to the output value and the actual posture; and an actual output calculation module, which is used to calculate the actual force value of the six-dimensional force sensor according to the measured output value and the initial offset value.

[0063] In a specific embodiment, the data acquisition module acquires the output values ​​of the six-axis force sensor in different postures in the no-load state by: according to the preset acquisition step length of the pitch angle β and the roll angle γ of the end of the manipulator, the output values ​​of the six-axis force sensor in different postures in the no-load state in the working surface are acquired, wherein: Within the range, n points are uniformly collected according to the collection step length, and within the range of -π≤γ≤π, m points are uniformly collected according to the collection step length, and a total of mn points are collected, each point corresponds to a set of the output values, wherein m and n are both positive integers.

[0064] In a specific embodiment, the offset calculation module uses a quadratic interpolation method to obtain an initial offset value of the measured output value, including:

[0065] First, determine the approximate interval of the actual attitude, where the approximate interval includes the actual pitch angle β s The approximate interval β i ≤β s ≤β i+1 and the actual roll angle γ wThe approximate interval γ j ≤γ w ≤γ j+1 ,in i≤n-1, j≤m-1, i, j, m, n are all positive integers;

[0066] Afterwards, according to the output value of the six-dimensional force sensor corresponding to the endpoint value of the approximate interval, the quadratic interpolation method is used to obtain the initial offset value of the measured output value.

[0067] Other technical features of the above-mentioned six-axis force sensor offset compensation system that are not described in detail, such as the working mode and calculation method, are the same as those of the six-axis force sensor offset compensation method and will not be repeated here.

[0068] In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention is described below in conjunction with specific scenario embodiments.

[0069] Scenario Example 1

[0070] The flowchart of the six-dimensional force sensor offset compensation method provided by the present invention is as follows: Figure 1 As shown, firstly, the no-load data corresponding to the sampling posture points in the working surface are collected, that is, each β corresponds to n points are collected uniformly within the range, and m points are collected uniformly within the range of -π≤γ≤π corresponding to each γ. The information of the six forces corresponding to the mn points collected is stored. The denser the collected values, the longer the time required, but the more accurate the final value is. Stop once, from Start collecting when When γ starts from -π, every Collect once, collect 61 times in total, until the value of γ is π, that is, the coordinates of the collection point are The value of β becomes γ starts from π to -π and samples from the beginning to the end, and from the end to the beginning, saving the scanning time set, that is, the coordinates of the collection points are This avoids unnecessary movement of the γ angle, ensuring that a set of data can be collected for each step, which is more efficient. Finally, 31*61=1891 sets of data were collected, each of which contained three force data and three torque data.

[0071] Specifically, those skilled in the art should know that the size of the acquisition step can be set according to actual conditions, and the acquisition step of the pitch angle and the acquisition step of the roll angle can be the same or different, which is not limited here.

[0072] like Figure 2As shown in the figure, it is a schematic diagram of the posture scanning of the six-dimensional force sensor of the present invention. Before correcting the sensor data, the position and posture of the sensor are obtained by kinematics according to the angles of each motor of the robot arm, and the position and posture of the sensor are found to satisfy β i ≤β s ≤β i+1 , γ j ≤γ w ≤γ j+1 The position and posture of Among them, i≤n-1, j≤m-1, i, j, m, n are all positive integers.

[0073] Afterwards, secondary interpolation is performed to obtain the initial offset values ​​of the force and torque output by the three axes of the six-dimensional force sensor.

[0074] Taking the x-direction force Fx as an example, when β=β i ,γ=γ j When Fx=Fx i,j , when β=β i+1 ,γ=γ j+1 When Fx=Fx i+1,j+1 , the position and attitude of the sensor at this time is β s ,λ w .

[0075] First, use quadratic interpolation to approximate β = β s , and γ=γ j-1 ,γ=γ j and γ=γ j+1 The values ​​of Fx are s,j-1 、Fx s,j 、Fx s,j+1 , then: through

[0076]

[0077] in, a is just an algebraic symbol used in the calculation process to make the formula structure more concise. It is only used for calculations and has no actual meaning.

[0078] Similarly, we can obtain Fx' s,j , Fx' s,j+1 The values ​​are:

[0079]

[0080]

[0081] This allows the quadratic interpolation to approximate Fx' s,w :

[0082]

[0083] in, b is just an algebraic symbol used to make the formula structure more concise during the calculation process. It is only used for calculations and has no actual meaning.

[0084] Similarly, other zero bias values, namely initial offset values ​​Fy', can be obtained. s,w , Fz' s,w , Mx' s,w 、My' s,w , Mz' s,w , the actual measurement value of the six-dimensional force sensor is Fx s,w , Fy s,w , Fz s,w , Mx s,w 、My s,w , Mz s,w , then the value of the input admittance controller is:

[0085] F'x s,w =Fx s,w -F'x s,w

[0086] F'y s,w =Fy s,w -F'y s,w

[0087] F”z s,w =Fz s,w -F'z s,w

[0088] M"x s,w =Mx s,w -M'x s,w

[0089] M”y s,w =My s,w -M'y s,w

[0090] M”z s,w =Mz s,w -M'z s,w

[0091] like Figure 1 As shown, after obtaining the actual force value, that is, the actual force / torque, it also includes the conventional steps of admittance control such as inputting the admittance controller and solving the inverse kinematics to output the position, which will not be described in detail here.

[0092] In summary, the six-dimensional force sensor offset compensation method and system provided by the present invention uniformly processes the factors affecting the zero bias value of the sensor, including gravity, temperature, sensor structure, etc., and directly uses the initial value of each axis under no-load conditions under different postures of the scanning force / torque sensor as the zero bias value under a fixed posture. Under dynamic conditions, the mapping relationship between the zero bias value of each axis and the sensor posture is used to obtain the initial offset value of each axis of the sensor under the real-time posture through the method of secondary interpolation.

[0093] Among them, the pitch angle changes from small to large, and each time the pitch angle changes, the roll angle changes from small to large or from large to small to ensure that the data of all points can be collected in one traversal without taking repeated routes. The basic offset value is stored after being collected once, and there is no need to collect it again. All factors that affect the force sensor are considered uniformly, and relatively high accuracy can be obtained without establishing a complex mathematical model. After collecting the initial data, the quadratic interpolation method is used to estimate the offset value under the real-time posture, which has high real-time performance and small estimation error.

[0094] The above specific embodiments only describe the design principle of the present invention. The shapes and names of the components in the description may be different and are not limited. Therefore, those skilled in the art in the field of the present invention may modify or replace the technical solutions recorded in the above embodiments; and these modifications and replacements do not deviate from the creative purpose and technical solutions of the present invention and should all fall within the protection scope of the present invention.

Claims

1. A six-dimensional force sensor offset compensation method, characterized in that: Admittance control for robotic arms, including: The output values ​​of the six-dimensional force sensor in different postures under no-load state are obtained, and the actual posture and measured output value of the six-dimensional force sensor are obtained; the output values ​​of the six-dimensional force sensor in different postures under no-load state are obtained by: according to the preset pitch angle β and roll angle γ of the end of the manipulator, the output values ​​of the six-dimensional force sensor in different postures under no-load state in the working surface are collected, wherein, In the range of -π≤γ≤π, n points are uniformly collected according to the collection step length, and in the range of -π≤γ≤π, m points are uniformly collected according to the collection step length, and a total of mn points are collected, each point corresponds to a set of the output values, wherein m and n are both positive integers; when the pitch angle β increases by one of the collection step lengths and enters the next collection posture, the roll angle γ uses the end collection position of the current collection posture as the starting collection position of the next collection posture; or, when the roll angle γ increases by one of the collection step lengths and enters the next collection posture, the pitch angle β uses the end collection position of the current collection posture as the starting collection position of the next collection posture; According to the output value and the actual posture, a quadratic interpolation method is used to obtain an initial offset value of the measured output value; The actual force value of the six-dimensional force sensor is obtained by calculation according to the measured output value and the initial offset value.

2. The offset compensation method according to claim 1, characterized in that: The actual posture of the six-dimensional force sensor is obtained as follows: The actual posture of the six-axis force sensor is obtained by kinematic forward solution according to the angles of each motor of the robotic arm, wherein the actual posture includes the actual pitch angle β s and the actual roll angle γ w .

3. The offset compensation method according to claim 1, wherein: The method of using a quadratic interpolation method to obtain an initial offset value of the measured output value includes: Determine an approximate interval of the actual attitude, wherein the approximate interval includes the actual pitch angle β s The approximate interval β i ≤β s ≤β i+1 and the actual roll angle γ w The approximate interval γ j ≤γ w ≤γ j+1 ,in i≤n-1, j≤m-1, i, j, m, n are all positive integers; According to the output value of the six-dimensional force sensor corresponding to the endpoint value of the approximate interval, the initial offset value of the measured output value is obtained by using the quadratic interpolation method.

4. The offset compensation method according to claim 3, characterized in that: Obtaining the initial offset value of the measured output value further includes: First get the pitch angle as β s The initial offset value at , and then the pitch angle is β s And the roll angle is γ w The initial offset value at ; Or, first get the roll angle as γ w The initial offset value at time, and then the roll angle is γ w And the pitch angle is β s The initial offset value at .

5. The offset compensation method according to claim 1, wherein: According to the measured output value and the initial offset value, the actual force value of the six-dimensional force sensor is calculated as follows: The actual force value of the six-dimensional force sensor is obtained by subtracting the initial offset value from the measured output value.

6. A six-dimensional force sensor offset compensation system, characterized in that: Admittance control for robotic arms, including: The data acquisition module is used to obtain the output values ​​of the six-axis force sensor in different postures in the no-load state, and obtain the actual posture and measured output value of the six-axis force sensor; the output values ​​of the six-axis force sensor in different postures in the no-load state are obtained by: according to the preset pitch angle β and roll angle γ of the end of the manipulator, the output values ​​of the six-axis force sensor in different postures in the no-load state in the working surface are collected, wherein, In the range of -π≤γ≤π, n points are uniformly collected according to the collection step length, and in the range of -π≤γ≤π, m points are uniformly collected according to the collection step length, and a total of mn points are collected, each point corresponds to a set of the output values, wherein m and n are both positive integers; when the pitch angle β increases by one of the collection step lengths and enters the next collection posture, the roll angle γ uses the end collection position of the current collection posture as the starting collection position of the next collection posture; or, when the roll angle γ increases by one of the collection step lengths and enters the next collection posture, the pitch angle β uses the end collection position of the current collection posture as the starting collection position of the next collection posture; An offset calculation module, used to obtain an initial offset value of the measured output value by using a quadratic interpolation method according to the output value and the actual posture; The actual output calculation module is used to calculate the actual force value of the six-dimensional force sensor according to the measured output value and the initial offset value.

7. The offset compensation system according to claim 6, characterized in that: The offset calculation module uses a quadratic interpolation method to obtain an initial offset value of the measured output value, including: Determine an approximate interval of the actual attitude, wherein the approximate interval includes the actual pitch angle β s The approximate interval β i ≤β s ≤β i+1 and the actual roll angle γ w The approximate interval γ j ≤γ w ≤γ j+1 ,in i≤n-1, j≤m-1, i, j, m, n are all positive integers; According to the output value of the six-dimensional force sensor corresponding to the endpoint value of the approximate interval, the initial offset value of the measured output value is obtained by using the quadratic interpolation method.

Images