CN112802182A - Anthropomorphic touch object reconstruction method and system based on touch sensor - Google Patents
Anthropomorphic touch object reconstruction method and system based on touch sensor Download PDFInfo
- Publication number
- CN112802182A CN112802182A CN202110072638.6A CN202110072638A CN112802182A CN 112802182 A CN112802182 A CN 112802182A CN 202110072638 A CN202110072638 A CN 202110072638A CN 112802182 A CN112802182 A CN 112802182A
- Authority
- CN
- China
- Prior art keywords
- touch
- attribute data
- data
- touch sensor
- anthropomorphic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
- G01D21/00—Measuring or testing not otherwise provided for
- G01D21/02—Measuring two or more variables by means not covered by a single other subclass
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
- G06F16/285—Clustering or classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
Abstract
The invention relates to a method and a system for reconstructing a personified touch object based on cluster analysis, wherein the method comprises the following steps: the method comprises the steps of driving a touch sensor to continuously move on the surface of an object, contacting the touch sensor with the object to obtain elastic deformation, normal force and tangential force information, obtaining a three-dimensional coordinate record with a contact point according to the contact point and a moving track of the touch sensor, and constructing object attribute data; acquiring object attribute data of a plurality of sample objects, extracting the object attribute data by adopting a clustering algorithm, establishing a database, classifying the database, and establishing a corresponding feature library for similar objects; and acquiring an object to be detected, detecting a plurality of pieces of object attribute data of the object to be detected, and comparing the object attribute data with database data to obtain optimal object attribute data, thereby completing object reconstruction. Compared with the prior art, the method and the device improve the efficiency and accuracy of reconstructing the object to be detected, and enable the object to be reconstructed more comprehensively.
Description
Technical Field
The invention relates to the field of object reconstruction, in particular to a method and a system for reconstructing an anthropomorphic touch object based on a touch sensor.
Background
The existing object reconstruction method needs auxiliary identification through a vision or displacement sensor, and the identification mode is single; the material attribute recognition of the similar object is incomplete, and a proper and complete data system is not established;
for example, the invention with the publication number of CN112102474A discloses a novel method and a system for reconstructing a cylinder in three dimensions, the method comprises the following steps: acquiring dynamic scanning data of a line laser displacement sensor in a measuring mechanism on left and right journals of a pair of wheels on an axle; the scanning data is three-dimensional point cloud data, and the three-dimensional point cloud data is subjected to initial calculation to generate initial fitting parameters of the axle journal; and performing iterative optimization on the initial fitting parameters by using a particle swarm algorithm to obtain final fitting parameters.
The invention with publication number CN108090966A discloses a virtual object reconstruction method and system suitable for virtual scenes, wherein one method comprises the following steps: the background server acquires the current real geographic position L1 of the real object and complete object construction data; the background server determines a virtual area matched with the current real geographic position L1 from the virtual scene according to the current real geographic position L1 and the complete object construction data; and the background server reconstructs a virtual object corresponding to the real object in the virtual area according to the complete object construction data.
In the scheme, the laser displacement sensor is adopted or the current real geographical position of the object and the complete object construction data are directly obtained to reconstruct the object, so that the information of the object is not comprehensively obtained, and the object is easily subjected to wrong judgment; for example, when the apples in the garden are reconstructed, the shape and the position of the apples are obtained, if the apples are required to be picked, the maturity can be judged according to the shape of the apples, the apples are picked, and a large misjudgment risk exists.
Disclosure of Invention
The invention aims to overcome the defect that the prior art is easy to cause wrong judgment on an object, and provides a method and a system for reconstructing an anthropomorphic touch object based on a touch sensor.
The purpose of the invention can be realized by the following technical scheme:
a personification touch object reconstruction method based on cluster analysis comprises the following steps:
a step of anthropomorphic touch: driving a touch sensor to continuously move on the surface of an object, contacting the object through the touch sensor, acquiring elastic deformation, normal force and tangential force information, and acquiring a three-dimensional coordinate record with a contact point according to the contact point and a moving track of the touch sensor;
establishing an object attribute: acquiring object attribute data according to the elastic deformation, the normal force, the tangential force and the three-dimensional coordinate record of the contact point;
classifying the feature library: acquiring the object attribute data of a plurality of sample objects, extracting the object attribute data by adopting a clustering algorithm, establishing a data set, classifying the data set, and establishing a corresponding feature library for similar objects;
an object reconstruction step: and acquiring an object to be detected, sequentially executing the anthropomorphic touch step and the object attribute establishing step, constructing a plurality of pieces of object attribute data of the object to be detected, comparing the object attribute data with the data set data, acquiring the feature library corresponding to the object to be detected, acquiring optimal object attribute data from the plurality of pieces of object attribute data according to the feature library, and finishing object reconstruction.
Further, the object attribute establishing step specifically includes:
extracting hardness attribute data of the object contact surface according to the relation between the elastic deformation and the normal force;
extracting roughness attribute data of the object contact surface according to the information of the normal force and the tangential force;
and fitting the object appearance track information attribute data according to the three-dimensional coordinate record of the contact point.
Further, the normal force FnThe calculation expression of (a) is:
in the formula, q is the moment of a contact point measured by the touch sensor, and n is the normal vector of the contact surface of the touch sensor;
the tangential force FtThe calculation expression of (a) is:
Ft＝p-Fn
in the formula, p is a three-dimensional acting force measured by the touch sensor;
the object contact surface roughness attribute data comprises an object surface static friction factor mu, and the calculation expression of the object surface static friction factor mu is as follows:
further, according to the radius D of the stress surface of the fruit measured by the touch sensoriCalculating the hardness of the object, and acquiring the hardness attribute data of the contact surface of the object, wherein the calculation expression of the hardness P of the object is as follows:
further, the attribute data of the fitted object surface track information is specifically that if the object is in a nearly spherical shape, a spherical surface fitting mode is adopted to solve a spherical equation, and the object surface track information is obtained;
the expression of the spherical equation is:
x2+y2+z2+ax+by+cz+d＝0
the fitting solving process of the spherical equation comprises the following steps:
assuming that the measurement point matrix is A, the coefficient matrix is m, and the distance matrix is l, then
Am＝l
Wherein x is an x-direction coordinate, y is a y-direction coordinate, z is a z-direction coordinate,
is a coefficient;
fitting by using indirect adjustment and least square method to obtain coefficient matrix m ═ (A)TA)-1ATl, so as to solve the coefficient matrix m and obtain the object surface track information.
Further, the tactile sensor is a six-axis force-torque sensor.
Further, the clustering algorithm is a K-means clustering algorithm.
Further, the K-means clustering algorithm is used for describing the similarity degree between data by calculating the minimum square error of Euclidean distance between data, so that data classification is realized.
Further, in the step of anthropomorphic touch, the touch sensor is driven by the mechanical arm to continuously move on the surface of the object.
The invention also provides a cluster analysis-based anthropomorphic touch object reconstruction system, which comprises a processor, a touch sensor and a mechanical arm, wherein the processor is respectively connected with the touch sensor and the mechanical arm, the mechanical arm is also connected with the touch sensor, and the processor executes the steps of the method.
Compared with the prior art, the invention has the following advantages:
(1) an anthropomorphic touch method is designed by utilizing a touch sensor. According to the method, a large amount of touch data suitable for training and analysis in an actual scene can be acquired, wherein the touch data comprises contact point information, contact normal force, tangential force and the like, so that the subsequent database establishment is facilitated, more object attribute information can be extracted, and the attribute identification of the object is more complete;
analyzing objects of different materials by using a K-means clustering algorithm; on the basis of establishing a database, performing cluster analysis on related data respectively, and establishing feature library information corresponding to the touch data;
when the object reconstruction is actually applied, the characteristic library corresponding to the object to be detected is obtained by obtaining a plurality of pieces of object attribute data of the object and comparing the data in the data set, so that the object attribute data which is most matched with the characteristic library is selected for object reconstruction, the error influence of detection data in actual application is reduced, the accuracy and reliability of a reconstruction structure are improved, and the efficiency and accuracy of reconstructing the object to be detected can also be improved.
(2) According to the detected contact point information, the contact normal force and the tangential force, hardness attribute data of the contact surface of the object, roughness attribute data of the contact surface of the object and attribute data of the surface track information of the object are obtained through calculation, the attributes of the object are obtained more comprehensively, the object is reconstructed more comprehensively, and multi-dimensional object reconstruction is achieved.
(3) The method greatly expands the application range of the touch sensor in various fields; for example, in the agricultural field, a reliable theoretical basis and a reliable realization scheme are provided for the current realization of the follow-up industrial development of robot object picking and the like; in the industrial field, a new idea and a new method are provided for the detection of the current industrial manufacturing products.
Drawings
FIG. 1 is a schematic view of a structural part of a product in an embodiment of the present invention;
FIG. 2 is a schematic flow chart of a method for personifying touch in an embodiment of the present invention;
FIG. 3 is a schematic diagram of haptic information in an embodiment of the present invention;
FIG. 4 is an interface diagram of a touch sensor for touching an object surface at a time according to an embodiment of the present invention;
FIG. 5 is a diagram illustrating the fitting results for a spherical object according to an embodiment of the present invention;
FIG. 6 is a schematic diagram illustrating a principle of object recognition implemented by a K-means clustering algorithm according to an embodiment of the present invention;
in the figure, the device comprises a touch sensor 1, a mechanical arm 2, a mechanical arm 3 and an upper computer.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
Example 1
The embodiment provides a personification touch object reconstruction method based on cluster analysis, which comprises the following steps:
a step of anthropomorphic touch: the method comprises the steps of driving a touch sensor to continuously move on the surface of an object, contacting the touch sensor with the object to obtain elastic deformation, normal force and tangential force information, and obtaining a three-dimensional coordinate record with a contact point according to the contact point and a moving track of the touch sensor;
establishing an object attribute: acquiring object attribute data according to the elastic deformation, the normal force, the tangential force and the three-dimensional coordinate record of the contact point;
classifying the feature library: acquiring object attribute data of a plurality of sample objects, extracting the object attribute data by adopting a clustering algorithm, establishing a data set, classifying the data set, and establishing a corresponding feature library for the similar objects;
an object reconstruction step: the method comprises the steps of obtaining an object to be detected, sequentially executing a personification touch step and an object attribute establishing step, establishing a plurality of object attribute data of the object to be detected, comparing the object attribute data with data set data, obtaining a feature library corresponding to the object to be detected, obtaining optimal object attribute data from the plurality of object attribute data according to the feature library, and completing object reconstruction.
As a preferred embodiment, the object attribute establishing step specifically includes:
extracting hardness attribute data of the object contact surface according to the relation between the elastic deformation and the normal force;
extracting roughness attribute data of the object contact surface according to the information of the normal force and the tangential force;
and fitting the object appearance track information attribute data according to the three-dimensional coordinate record of the contact point.
Normal force FnThe calculation expression of (a) is:
in the formula, q is the moment of a contact point measured by the touch sensor, and n is the normal vector of the contact surface of the touch sensor;
tangential force FtThe calculation expression of (a) is:
Ft＝p-Fn
in the formula, p is a three-dimensional acting force measured by the touch sensor;
the object contact surface roughness attribute data includes an object surface static friction factor μ, which is calculated by the expression:
radius D of stress surface of fruit measured by touch sensoriCalculating the hardness of the object to obtain the hardness attribute data of the contact surface of the object, wherein the calculation expression of the hardness P of the object is as follows:
as a preferred embodiment, the attribute data for fitting the object surface trajectory information is specifically that, if the object is in a nearly spherical shape, a spherical surface fitting mode is adopted to solve a spherical equation to obtain the object surface trajectory information;
the expression of the spherical equation is:
x2+y2+z2+ax+by+cz+d＝0
the fitting solving process of the spherical equation is as follows:
assuming that the measuring point matrix is A, the coefficient matrix is m, and the distance matrix is l, then:
Am＝l
wherein x is an x-direction coordinate, y is a y-direction coordinate, z is a z-direction coordinate,
are coefficients.
Fitting by using indirect adjustment and least square method to obtain coefficient matrix m ═ (A)TA)-1ATl, so as to solve the coefficient matrix m and obtain the object surface track information.
In a preferred embodiment, the tactile sensor is a six-axis force-torque sensor.
As a preferred implementation mode, the clustering algorithm is a K-means clustering algorithm, and the K-means clustering algorithm is used for describing the similarity degree between data by calculating the minimum square error of Euclidean distance between data so as to realize data classification.
In a preferred embodiment, in the step of anthropomorphic touch, the mechanical arm drives the touch sensor to continuously move on the surface of the object.
The above preferred embodiments are combined to obtain an optimal embodiment, and a specific implementation process of the optimal embodiment is described below.
Summary of the invention
The personification touch object reconstruction method based on cluster analysis is used for touching an object with a certain shape and uniform texture by utilizing a touch sensor according to the characteristic that the object has particularity in the aspects of characteristics such as hardness, roughness and size, and synchronously acquiring related data. The touch sensor can solve the characteristic points at the contact position of the object according to the collected data of various forces and moments and the relevant characteristics of the various forces and moments, and perform cluster analysis on the collected data by using a neural network. In the touch process, the characteristic of the hardness of the contact surface of the object can be extracted from the relationship between the elastic deformation degree of the contact surface and the normal force when the contact surface is touched; the roughness of the contact surface of the extractable material is characterized by the normal force and the tangential force fed back from the contact surface; and recording the characteristics of the track information which can be fitted to the appearance of the object from the three-dimensional coordinates of the touch points. Different samples of the same kind of objects are tested on the basis of the anthropomorphic touch process designed by the touch sensor, related data are obtained, and the characteristics of the data are extracted, classified and recorded in a cluster analysis machine learning mode. Therefore, in the anthropomorphic touch process, the touch recording information acquired from the test sample is compared with the trained classification condition, so as to achieve the effect of identifying and reconstructing the object.
According to the scheme, based on the contact surface characteristics of the object, a clustering analysis and anthropomorphic touch method for data is designed, and the identification and reconstruction of the object types are realized.
Second, detailed description
2.1, product side
The processing of this embodiment includes selecting a touch sensor, extracting relevant data, building a database of data features, training an algorithm classifier, and the like. Therefore, a clustering analysis program based on a combination of a six-axis force-moment sensor, a touch sensor suitable for solving by a nonlinear least square method and a K-means clustering algorithm is selected in the scheme. And selecting a proper mechanical arm for driving the touch sensor to contact with the contact surface. After the touch sensor is contacted with the contact surface to be contacted, the touch sensor returns information to the upper computer, and the upper computer drives the touch sensor to continuously move on the contact surface through continuous closed-loop feedback control of feedback data to acquire complete touch information. After the data are collected and sorted into a database, classification is carried out on the basis of data with proper scale by utilizing a K-means clustering algorithm, and a characteristic library corresponding to the similar object is established. On the basis of establishing the feature library, untrained contact surface data are tested. And comparing the test data with the generated feature library data, identifying the related information of the test sample object, and finishing the object reconstruction function at the upper computer end according to the related information.
The personification touch process comprises the following steps: when the touch sensor contacts with an object along with the movement of the mechanical arm, the six-axis force-torque sensor sends a measured force-torque numerical value to the upper computer. In the upper computer, the data information is resolved through an inverse solution algorithm to obtain the contact point coordinates of the contact point on the shell, and key information such as normal force, tangential force and the like on the contact point is calculated. Meanwhile, the upper computer calculates the coordinate position of the contact point in the space according to the transmission information of the mechanical arm, and a touch track graph is formed according to the coordinate position.
The touch sensor comprises a triangular structure for supporting and a spherical structure for touching, and the sensor shell is formed by 3D printing of ABS resin materials. The above mentioned spherical model is 18mm in diameter and the printed top surface is the same size as the surface touching the spherical structure.
The tactile sensing section structure and the mechanical arm section structure are shown in fig. 1.
Touch sensors are important tools for gathering information. The six-axis force-torque sensor used here is ATI nano 17 force/torque sensor (Calibration SI-25-025resolution:1/160N for Fx Fy Fz 1/32Nmm for Mx My Mz Range FxFy-25N Fz-35N Mx Mz-250 Nmm).
The mechanical arm drives the touch sensor by using a Dobot magic mechanical arm, the number of axes is 4, the effective load is 500g, the maximum extension distance is 320mm, and the repeated positioning precision is 0.2 mm. During the touch process, the touch sensor needs to be controlled to continuously move on the surface of the object to be contacted through position feedback of the touch sensor and the mechanical arm.
2.2, technical side
As shown in fig. 2, the flow of the anthropomorphic touch method in this embodiment includes:
s101: the mechanical arm continuously moves until the touch sensor contacts the surface of the object;
s102: the upper computer or the processor collects data sent by the touch sensor;
s103: settling the current contact information by using a contact static friction model;
s104: transmitting the solved information to a control model of the mechanical arm;
s105: the upper computer or the processor transmits the data control signal to the mechanical arm;
s106: the robot arm receives the control signal, continuously moves on the surface of the object, and returns to step S102.
The functions of the components in this embodiment are as follows:
tactile sensor: the contact information data is transmitted to an upper computer or a processor.
Mechanical arm: control signals are obtained according to the mechanical arm control model, so that the touch sensor can continuously move on the surface of an object to acquire more information.
Host computer or controller: on one hand, the received tactile information is transmitted to a set mechanical arm control model for resolving at an upper computer or a controller end, and the tactile information is used for controlling the continuous movement of the mechanical arm; on the other hand, the upper computer continuously records the tactile information, reduces the information such as surface friction, hardness and outline of the object through an algorithm, and completes the reconstruction function of the object after clustering analysis.
Under the set static friction model, the normal force can be obtained according to the component of the contact point acting force on the normal vector n and the relation between the component and the contact point moment q
Then subtracting the three-dimensional acting force p vector measured by the tactile sensor to obtain a tangential force Ft＝p-Fn. According to the method, the static friction factor of the surface of the object can be obtained
The haptic information obtained by resolving certain one-time six-dimensional force and moment data returned by the haptic sensor is shown in fig. 3. Wherein the curved surface is the outer surface of the touch sensor. Wherein the blue arrows represent normal force vectors; the orange arrows indicate the tangential force vectors.
When measuring the hardness of an object, a measurement method similar to the hardness of a material is adopted. The interface diagram of a touch sensor touching the surface of an object at a time is shown in fig. 4. According to the force F on the object surface and the radius D of the stress surface of the fruit returned by the sensoriCan calculate the objectBody hardness
The unit is Pa.
When the size of an object is measured and the measured object is assumed to be nearly spherical, the spherical equation x is solved by adopting a mode of carrying out spherical curve fitting on the measured point2+y2+z2+ ax + by + cz + d ═ 0. Assuming that the measurement point matrix is a, the coefficient matrix is m, and the distance matrix is l, Am is l.
In the fitting process, an indirect adjustment and least square method is adopted, and a coefficient matrix m ═ A can be obtainedTA)-1ATl. And obtaining a coefficient matrix m of the ball fitting. The results of a certain fit according to the above method are shown in fig. 5. In this way, the diameter of the object can be calculated smoothly as
The coordinates of the center of the sphere are
The spherical data accumulation and the subsequent model reconstruction can be carried out according to the calculated data.
As shown in fig. 6, the principle of object recognition implemented by K-means clustering algorithm includes the following steps:
s201: the touch sensor touches the same kind of articles and collects related touch information;
s202: performing clustering analysis on the touch data by using a K-means clustering algorithm;
s203: establishing a touch information characteristic library corresponding to contact surfaces of different objects;
s204: when a new sample is tested, acquiring surface information by touch, and sending the surface information to an upper computer;
s205: sending the data into a feature library for matching and identification, and analyzing the data at an upper computer end;
s206: and analyzing and converting the identification information and the data into a visible three-dimensional object.
And aiming at the contact surface of the same kind of object, a large amount of effective data is obtained as a data sample through repeated anthropomorphic touch. And analyzing the extracted sample by using a K-means clustering algorithm, extracting the data characteristics after the touch is performed, and establishing a data characteristic library aiming at the data of the corresponding touch surface. In the subsequent testing link, the collected testing data is matched with the data in the library, and the object information under the current testing condition is identified.
And analyzing the sample relation of the collected data in the training set by using a K-means clustering algorithm, and describing the similarity degree between the data through the minimized square error of the Euclidean distance between the data so as to achieve a classification result. In view of the fact that the object can be represented by a continuous function, in the process of manufacturing the training set, attention should be paid to selecting samples with obvious differences for training, and accuracy of training results is guaranteed. When the touch sensor continuously moves on the surface of the contact surface, the relevant data such as the surface friction degree, the object hardness, the object size and the like can be obtained from the anthropomorphic touch method for training, and relevant attributes are extracted. The roughness information of the contact surface can be extracted by training the tangential force and the normal force in the data; training the ratio of the normal force to the area of the contact surface in the data to extract the characteristics of the hardness of the contact surface material; the diameter size of the specific round object can be calculated through coordinate information of the contact points in the data for training, the size information characteristic of the object can be extracted, and then the data of the attributes are subjected to cluster analysis by using a K-means clustering algorithm, so that a corresponding characteristic set of the similar object is established.
And drawing a three-dimensional model at the upper computer end according to the acquired information stored in the database of the hardness, the roughness and the size and the coordinate information at the mechanical arm end, so as to realize the three-dimensional reconstruction function of the object.
The embodiment also provides a cluster analysis-based anthropomorphic touch object reconstruction system, which comprises a processor, a touch sensor and a mechanical arm, wherein the processor is respectively connected with the touch sensor and the mechanical arm, the mechanical arm is also connected with the touch sensor, and the processor executes the steps of the cluster analysis-based anthropomorphic touch object reconstruction method.
The foregoing detailed description of the preferred embodiments of the invention has been presented. It should be understood that numerous modifications and variations could be devised by those skilled in the art in light of the present teachings without departing from the inventive concepts. Therefore, the technical solutions available to those skilled in the art through logic analysis, reasoning and limited experiments based on the prior art according to the concept of the present invention should be within the scope of protection defined by the claims.
Claims (10)
1. A personification touch object reconstruction method based on cluster analysis is characterized by comprising the following steps:
a step of anthropomorphic touch: driving a touch sensor to continuously move on the surface of an object, contacting the object through the touch sensor, acquiring elastic deformation, normal force and tangential force information, and acquiring a three-dimensional coordinate record with a contact point according to the contact point and a moving track of the touch sensor;
establishing an object attribute: extracting hardness attribute data of the object contact surface according to the relation between the elastic deformation and the normal force; extracting roughness attribute data of the object contact surface according to the information of the normal force and the tangential force; fitting object surface track information attribute data according to the three-dimensional coordinate record of the contact point; summarizing the object contact surface hardness attribute data, the object contact surface roughness attribute data and the object surface track information attribute data into object attribute data;
classifying the feature library: acquiring the object attribute data of a plurality of sample objects, extracting the object attribute data by adopting a clustering algorithm, establishing a data set, classifying the data set, and establishing a corresponding feature library for similar objects;
an object reconstruction step: and acquiring an object to be detected, sequentially executing the anthropomorphic touch step and the object attribute establishing step, constructing a plurality of pieces of object attribute data of the object to be detected, comparing the object attribute data with the data set data, acquiring the feature library corresponding to the object to be detected, acquiring optimal object attribute data from the plurality of pieces of object attribute data according to the feature library, and finishing object reconstruction.
2. The method of claim 1, wherein the normal force F is a normal forcenThe calculation expression of (a) is:
in the formula, q is the moment of a contact point measured by the touch sensor, and n is the normal vector of the contact surface of the touch sensor;
the tangential force FtThe calculation expression of (a) is:
Ft＝p-Fn
wherein p is the three-dimensional force measured by the tactile sensor.
3. The method for reconstructing the anthropomorphic touch object based on the cluster analysis as recited in claim 2, wherein the object contact surface roughness attribute data comprises an object surface static friction factor μ, and the calculation expression of the object surface static friction factor μ is as follows:
4. the method as claimed in claim 3, wherein the fruit stress surface radius D measured by the touch sensor is used as the basis for reconstructing the anthropomorphic touch object based on cluster analysisiCalculating the hardness of the object, and acquiring the hardness attribute data of the contact surface of the object, wherein the calculation expression of the hardness P of the object is as follows:
5. the method for reconstructing the anthropomorphic touch object based on the cluster analysis as claimed in claim 2, wherein the attribute data of the fitted object surface trajectory information is specifically that if the object is nearly spherical, a spherical surface fitting mode is adopted to solve a spherical equation to obtain object surface trajectory information;
the expression of the spherical equation is:
x2+y2+z2+ax+by+cz+d＝0
the fitting solving process of the spherical equation comprises the following steps:
assuming that the measurement point matrix is A, the coefficient matrix is m, and the distance matrix is l, then
Am＝l
Wherein x is an x-direction coordinate, y is a y-direction coordinate, z is a z-direction coordinate,
is a coefficient;
fitting by using indirect adjustment and least square method to obtain coefficient matrix m ═ (A)TA)-1ATl, so as to solve the coefficient matrix m and obtain the object surface track information.
6. The method of claim 1, wherein the touch sensor is a six-axis force-moment sensor.
7. The method for reconstructing anthropomorphic touch objects based on cluster analysis as recited in claim 1, wherein the clustering algorithm is a K-means clustering algorithm.
8. The method as claimed in claim 7, wherein the K-means clustering algorithm is used for classifying data by calculating the minimum square error of the euclidean distance between data and describing the similarity between data.
9. The method for reconstructing the anthropomorphic touch object based on the cluster analysis as claimed in claim 1, wherein in the anthropomorphic touch step, the touch sensor is driven by a mechanical arm to continuously move on the surface of the object.
10. A cluster analysis based personified touch object reconstruction system comprising a processor, a touch sensor and a robotic arm, the processor being connected to the touch sensor and the robotic arm, respectively, the robotic arm being further connected to the touch sensor, the processor performing the steps of the method according to any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110072638.6A CN112802182B (en) | 2021-01-20 | 2021-01-20 | Method and system for reconstructing anthropomorphic touch object based on touch sensor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110072638.6A CN112802182B (en) | 2021-01-20 | 2021-01-20 | Method and system for reconstructing anthropomorphic touch object based on touch sensor |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112802182A true CN112802182A (en) | 2021-05-14 |
| CN112802182B CN112802182B (en) | 2022-12-16 |
Family
ID=75810630
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110072638.6A Active CN112802182B (en) | 2021-01-20 | 2021-01-20 | Method and system for reconstructing anthropomorphic touch object based on touch sensor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112802182B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090132088A1 (en) * | 2007-04-24 | 2009-05-21 | Tairob Ltd. | Transfer of knowledge from a human skilled worker to an expert machine - the learning process |
| CN109543732A (en) * | 2018-11-09 | 2019-03-29 | 山东大学 | Assembly system and method based on category feature knowledge base |
| CN110956651A (en) * | 2019-12-16 | 2020-04-03 | 哈尔滨工业大学 | Terrain semantic perception method based on fusion of vision and vibrotactile sense |
| CN111079333A (en) * | 2019-12-19 | 2020-04-28 | 华南理工大学 | Flexible touch sensor deep learning sensing method |
| CN112197676A (en) * | 2020-10-14 | 2021-01-08 | 同济大学 | Method and device for acquiring object surface information, computer equipment and storage medium |
-
2021
- 2021-01-20 CN CN202110072638.6A patent/CN112802182B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090132088A1 (en) * | 2007-04-24 | 2009-05-21 | Tairob Ltd. | Transfer of knowledge from a human skilled worker to an expert machine - the learning process |
| CN109543732A (en) * | 2018-11-09 | 2019-03-29 | 山东大学 | Assembly system and method based on category feature knowledge base |
| CN110956651A (en) * | 2019-12-16 | 2020-04-03 | 哈尔滨工业大学 | Terrain semantic perception method based on fusion of vision and vibrotactile sense |
| CN111079333A (en) * | 2019-12-19 | 2020-04-28 | 华南理工大学 | Flexible touch sensor deep learning sensing method |
| CN112197676A (en) * | 2020-10-14 | 2021-01-08 | 同济大学 | Method and device for acquiring object surface information, computer equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 顾海巍: "机器人灵巧手的物体触摸识别及自适应抓取研究", 《中国优秀博硕士学位论文全文数据库（博士）信息科技辑》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112802182B (en) | 2022-12-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Bauza et al. | Tactile mapping and localization from high-resolution tactile imprints | |
| Gandarias et al. | CNN-based methods for object recognition with high-resolution tactile sensors | |
| Yi et al. | Active tactile object exploration with gaussian processes | |
| Villalonga et al. | Tactile object pose estimation from the first touch with geometric contact rendering | |
| Liu et al. | A computationally fast algorithm for local contact shape and pose classification using a tactile array sensor | |
| Bhattacharjee et al. | Haptic classification and recognition of objects using a tactile sensing forearm | |
| Sferrazza et al. | Transfer learning for vision-based tactile sensing | |
| Molchanov et al. | Contact localization on grasped objects using tactile sensing | |
| CN111507425A (en) | Mode identification method and system based on air bag array touch perception | |
| Fang et al. | A cross-modal tactile sensor design for measuring robotic grasping forces | |
| Palermo et al. | Automatic fracture characterization using tactile and proximity optical sensing | |
| Lima et al. | Classification of textures using a tactile-enabled finger in dynamic exploration tasks | |
| CN112802182B (en) | Method and system for reconstructing anthropomorphic touch object based on touch sensor | |
| CN116652971A (en) | Method for identifying objects with different rigidities and controlling grabbing force of manipulator | |
| Dong et al. | A review of robotic grasp detection technology | |
| CN113792792B (en) | Robot-based haptic data perception analysis method | |
| Dario et al. | An investigation on a robot system for disassembly automation | |
| Wu et al. | Vision-based tactile intelligence with soft robotic metamaterial | |
| Pohtongkam et al. | Object recognition using glove tactile sensor | |
| CN112893180A (en) | Object touch classification method and system considering friction coefficient abnormal value elimination | |
| Garrofé et al. | Virtual haptic system for shape recognition based on local curvatures | |
| Gu et al. | Model recovery of unknown objects from discrete tactile points | |
| Grover et al. | Under pressure: Learning to detect slip with barometric tactile sensors | |
| Rouhafzay et al. | A virtual tactile sensor with adjustable precision and size for object recognition | |
| Chang et al. | An Investigation of Multi-feature Extraction and Super-resolution with Fast Microphone Arrays |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |