JPH0798217A - Method and device for automatic inspection of three dimensional object - Google Patents

Abstract

PURPOSE:To automate various inspections by identifying and recognizing objects to-be-examined based on obtained design data, etc., deciding a measuring and inspecting method to be carried out, and controlling sensor and lighting systems. CONSTITUTION:The data on design, examination, etc., of an object 1 to-be- examined is given through communication means, and based on this, position and attitude of a sensor 7 and a lighting system 8 are controlled, so that the object 1 to-be-examined is identified and recognized to determine an inspection method. For example, the object 1 to-be-examined is transported by an autonomous unmanned transportation robot 2, and carried into/out of a measurement stage 6 by handling robots 3 and 4. A sensor 7 detects the object 1 to-be- examined lit by the lighting means 8. The sensor 7 and lighting means 8 are controlled in their position and attitude by robots 3, 10, 11, etc.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic inspection apparatus and method for a three-dimensional object which can be used for automatically inspecting the appearance of an inspection object having a three-dimensional shape in a manufacturing line. .

[0002]

2. Description of the Related Art In the conventional inspection apparatus used in such a production line, even a partially automated apparatus, which has a problem of requiring a lot of manpower and a long dead time, is for inspecting a specific object, and has various types. It is not possible to cope with the situation (multi-product, various postures, etc.). Even if the inspection detects defects in the inspection object, there is no particular loop for analysis, factor analysis, and design review based on the results.
Enthusiasm of the engineer in charge. Therefore, the design defect is not corrected or the correction is delayed.

[0003]

SUMMARY OF THE INVENTION An object of the present invention is to make it possible to automatically perform various inspections according to the situation, and to automatically analyze even a defect is detected. It is an object of the present invention to provide an automatic inspection device and method for a three-dimensional object, which is capable of

[0004]

SUMMARY OF THE INVENTION The present invention provides a means for handling an inspection object, a sensor / illumination system, a means for controlling the position and orientation of a sensor or an illumination device, a means for processing a sensor signal, and an object. Includes first calculation means for identifying / recognizing an object, means for communicating data of an object with the outside, and second calculation means for determining a measurement method or inspection method suitable for the object from the data and the recognition result. Is an automatic inspection device for three-dimensional objects.

Further, according to the present invention, there is provided a third arithmetic means for analyzing the cause of the defective portion from the inspection result by the sensor signal processing means, and a means for outputting the arithmetic result of the third arithmetic means to an external design means or manufacturing means. It is characterized by including.

The present invention also relates to shape data output means for outputting data relating to the shape of the inspection object, means for measuring at least one of the three-dimensional shape, position and orientation of the inspection object, and shape data. Responsive to the output of the output means and the measuring means, the inspection target object is recognized, the inspection command information for measurement is output, and the measurement means uses the inspection command information to measure the predetermined portion of the inspection target object. An automatic inspection apparatus for a three-dimensional object, comprising: a recognition / instruction means to perform; and a means for inspecting a defect or the like of an inspection object in response to an output of a shape data output means and a measurement means. .

Further, the present invention is characterized in that the recognition / command means determines the inspection method for inspecting the inspection object and the inspection tool of the measurement means for performing the inspection method, and operates the measurement means. And

According to the present invention, the inspection tool has an optical detecting means for optically inspecting the inspection object, a light source for irradiating the inspection object with light, and inspection conditions for predetermining the optical inspection means and the light source. And means for operating so that

Further, according to the present invention, the position and orientation of the object, the sensor, the illumination system and the like are determined so that the measurement necessary for identifying the object can be realized from the data of the inspection object registered in advance and the initial measurement data. Then, a step of generating a command to the handling means, a step of identifying the target object by comparing the registered data with the measurement data, and a target object so that the inspection required for the identified target object can be realized. An automatic inspection method for a three-dimensional object, comprising: a step of determining a position and orientation of a sensor / illumination system and the like, generating a command to a handling means; and a step of outputting an inspection result to the outside.

[0010]

According to the present invention, for example, a conventional sensor for inspection, a means for conveying an inspection object, an illumination system as a light source for irradiating the inspection object with light, and image processing for optically detecting the illumination system. Further, the device including the system is provided with data related to the design of the inspection object and other inspections through the communication means, thereby controlling the position and orientation of the sensor and the illumination system to identify and recognize the inspection object. Then, the inspection method of the inspection object is generated. In this way, the characteristics of the inspection object are extracted by the sensor, the illumination system, the image processing system, etc., and the inspection object is identified and recognized from the design data and the like obtained through, for example, communication means, and the identified inspection object is performed. The power measurement method and inspection method are determined, and the inspection is performed by controlling the position and orientation of the sensor and illumination system.

Further, according to the present invention, the factor of the defect detected by the inspection is estimated, and the calculation result is output to the portion for designing and production planning by the communication means, and thus the defect is detected from the inspection result. Perform factor analysis and send study results to design and manufacturing departments. Therefore, the design cannot be reviewed and the production cannot be examined, and, for example, a design defect can be corrected.

[0012]

【Example】

(1) The configuration of the entire system of one embodiment of the present invention will be described.

With the diversification of products and the shortening of delivery time, CAD /
The CAM system has been introduced to streamline design and manufacturing. Along with this, there is a demand for an organically connected system from design and manufacturing to inspection and processing in order to improve quality and productivity. Currently, CAD information is used to generate NC data for processing and It is only used to create an operation level program according to a predetermined procedure such as offline teaching of a coordinate measuring machine.
CA such as the shape of the inspection object is given by the instruction of the work level such as "inspect product A" or "assemble product B".
It is desirable to develop an operation plan for measurement / inspection autonomously and learning based on non-shape information such as D information, tolerances, inspection points, and measurement accuracy, and develop a system with flexibility and toughness. Be done.

As a solution to this problem, the applicant of the present invention has proposed an "active automatic inspection system" and stated the need for its early realization.

The following items a to d are listed as the functions required for this active automatic inspection system.

(A) In the machining / assembly manufacturing line, the object to be inspected is a three-dimensional object,
It is important to measure 3D shape, position / orientation, and recognize objects.

(B) Especially when a three-dimensional shape is targeted, the amount of data increases compared to two-dimensional, and a measuring method suitable for speeding up the inspection, generation of an inspection model, comparison between the measurement result and the inspection model. A matching algorithm is required.

(C) In addition, since the object is a three-dimensional solid, in the case of a complicated shape or a shape having steps such as unevenness,
It is impossible to measure the required shape of the target object due to the existence of blind spots etc. only by measuring from one direction, and autonomous measurement is performed based on CAD information and environment information of the measurement site for active measurement. It is desirable to have an inspection system that can draw up an operation plan for measurement and inspection.

(D) At present, programming for inspection is required for each inspection object, and in high-mix low-volume production, the cost required to create an inspection program becomes enormous.

Therefore, when an object to be inspected and design data are given, a function of autonomously generating a program for purposeful, economical and safe measurement / recognition and inspection, versatility and flexibility. A three-dimensional object measurement / recognition / inspection system is required.

FIG. 1 is a perspective view showing a specific structure of an embodiment of the present invention. The appearance of the inspection object 1 having a three-dimensional shape on the manufacturing line is automatically inspected. The inspection object 1 is transferred by the autonomous unmanned transfer robot 2,
Measuring table 6 using handling robots 3, 4, 5 etc.
The sensor 7, which is carried in / out of the TV camera, is a television camera and other optical detection means or other detection means,
Inspection object 1 illuminated by illumination means 8 which is a light source
To detect. The sensor 7 and the illumination means 8 are the robot 3
In addition, the positions, postures, etc. of the robots 9, 10, 11 are controlled.

FIG. 2 is a block diagram showing the outline of the overall construction of the active inspection and measurement apparatus of one embodiment of the present invention, and FIG. 3 is a block diagram showing the more specific construction of that embodiment. . A more detailed structure of the structure shown in FIG. 3 is shown in FIGS. 4 and 5. A more detailed description of the configuration shown in FIGS. 4 and 5 will be described in detail with reference to the drawings starting from FIG.

Automatic appearance inspection system A shown in FIG.
0 is a three-dimensional object recognition and inspection unit A shown in FIG.
1 and a three-dimensional object measuring unit A2. Figure 6
Resource is a physical means required when performing an inspection, and includes, for example, an inspection tool, a fixture, a transport device, and all other objects than an inspection object. The three-dimensional object recognition and inspection unit A1 is accompanied by the three-dimensional information A2a measured by the three-dimensional object measuring unit A2 as shown in FIG. 8 and the inspection information such as the shape sent from the CAD design system. A three-dimensional object recognition unit A1 for generating an inspection model required for inspection and recognizing a specific three-dimensional object from a plurality of models from the product model data A1a.
2 is provided, and the inspection system is automatically selected and generated by an expert system or the like according to the purpose of the inspection from the generated inspection model and the inspection program is automatically generated. A11,
The size / shape / defect inspection unit A13 is used to inspect sizes / shapes / defects by comparing an inspection model with measurement data.

Further, in the measuring section A2 for the three-dimensional object, FIG.
An active measurement unit A2 including a sensing device, a lighting device, a hand, a robot, and a transfer device in order to actively perform measurement based on the inspection command data like
3. A three-dimensional information acquisition unit A21 that generates a distance / tilt / edge / density image from the measured three-dimensional point sequence data,
It is a measurement unit A22 of a three-dimensional object that generates a three-dimensional shape from the acquired data.

From the next section, these components will be developed in more detail, and the flow of information and processing and the necessary elemental technology will be described.

(2) Description of each component (2-1) Automatic selection / automatic programming unit A11 of inspection method The automatic selection / automatic programming unit A11 of inspection method is
As shown in FIG. 8, various information is received from the three-dimensional object recognition unit A12 and the design system, an inspection work plan is prepared, and automatic selection / automatic programming work is performed with respect to the inspection method, sequence, tools, conditions, and routes. The result is the active measurement unit A2 in the measurement unit A2 of the three-dimensional object in FIG.
It is sent to the 3D and 3D information acquisition unit A21.

However, minor changes in the operating program of the manipulator or the like for performing information processing by sensor fusion and active inspection are performed as local processing in the active measuring section.

From the above policy, the automatic selection / automatic programming unit A11 for the inspection method is provided with the following six constituent elements a1 to a6 as shown in FIG.

(A1) Inspection / feature extraction work development A111 (a2) Inspection method determination A112 (a3) Inspection order determination A113 (a4) Inspection tool determination A114 (a5) Inspection condition determination A115 (a6) Inspection path generation A116 Information and The flow of processing and required elemental technologies are as follows.

As shown in FIG. 10, a three-dimensional object recognition unit A12
The information sent from is the feature extraction request A12a when the identification information ID of the inspected object is unconfirmed, and the inspection model, position / orientation created by the recognition unit A12 of the three-dimensional object after confirming the ID. Data, inspection accuracy, part, etc. are A12b. An inspection / feature extraction work expansion unit A111 is provided in order to draw up an overall inspection work plan from such information A12a and A12b.

In the inspection / feature extraction work expansion unit A111,
Information from the three-dimensional object recognition unit A12 is the feature extraction request A
In the case of 12a, the fact is confirmed, and the feature extraction request is sent as it is to the measurement unit A2 of the three-dimensional object as the inspection command data.

When the information from the recognition unit A12 of the three-dimensional object is the inspection model, position / orientation data, inspection accuracy / part, etc. A12b of the inspection object accompanied by the ID information confirmed from the extracted features. Confirms the intent to perform inspection work and makes an inspection work plan based on the information sent. Here, general-purpose knowledge such as visual information processing algorithms and control strategies has already been given, and the outline of the overall work plan is determined by giving information such as the object to be inspected and the purpose. . In this component, an AI (Artificial Intelligence) planner for planning inspection work is important.

When an inspection work plan is prepared in the inspection / feature extraction work development unit A111, an optimum inspection method, its information processing algorithm, and its application procedure are applied to the application of the object to be inspected or the purpose based on it. Will need to be generated automatically. Therefore, the inspection method determination unit A112, the inspection order determination unit A113, and the inspection tool determination unit A114 are provided.

The inspection method determination unit A112 automatically selects the optimum inspection method for the inspection site and its information processing algorithm from the information such as the inspection model and the position / orientation data.
Here, past cases are referred to from the inspection method database, and the inspection method such as contact type or non-contact type in the case of shape, size, accuracy of the inspection site, passive type or active type in the case of non-contact type and its information processing. The algorithm is determined. Here, past cases can be used, and case-based reasoning with a learning function is applied.

The inspection order determining unit A113 receives information such as the inspection model and the inspection method determined by the inspection method determining unit A112, and inspects a plurality of inspection parts in any order while considering the inspection method and the information processing algorithm. Decide what to do. Here, a scheduling function using an expert system is provided in consideration of the efficiency of inspection work.

The inspection tool determination unit A114 automatically selects an inspection tool based on the inspection method determined by the inspection method determination unit A112. Inspection tools such as multi-view (CCD (charge storage element) or PSD (photodetection element)), range finder, and touch probe are selected while taking into consideration information such as inspection accuracy. Here, an inspection tool selection system using rule-based inference is constructed from an inspection tool database by creating a rule for selecting an inspection tool by using information such as inspection accuracy and site.

In order to improve the ability of the visual system, not only the information processing ability of the input image is improved, but also the focus or aperture of the camera or its position or direction,
Furthermore, it is important for the system to autonomously control the position, direction brightness, etc. of the illumination light and input an image that is easy to actively process. Therefore, knowledge about imaging such as cameras and lighting systems (valid objects, purposes, functions, efficiency, etc.)
Is provided to the system, and an inspection condition determination unit A115 and an inspection path generation unit A116 are provided so that the active measurement unit A23 can obtain an optimum image according to the application purpose.

The inspection condition determining unit A115 uses the information such as the inspection model, the inspection order determined by the inspection order determining unit A113, and the information about the inspection tool determined by the inspection tool determining unit to determine the focus, aperture, lighting, etc. of the camera. To determine the optimum inspection conditions. Further, when the illumination system is installed in the manipulator and active writing work is performed, the position and direction of the illumination system are determined, and the route of the manipulator is generated. Here, it is considered that determination of inspection conditions by case-based reasoning and operation planning technology for manipulators are required as important technologies.

The inspection path generation unit A116 mounts an inspection tool such as a camera on the basis of information such as the inspection model, the inspection order determined by the inspection order determination unit A113, and the inspection tool information determined by the inspection tool determination unit. Generates the route of the manipulator. This process includes a series of operation plans in which the manipulator mounts the inspection tool and sets the inspection object. The manipulator motion planning technology is an important technology here, and the current technology concerning robot motion planning such as obstacle avoidance is applied.

The respective information thus determined are collected as inspection command data A11b and sent to the active measuring section A23 and the three-dimensional information acquiring section A21 which are the constituent elements of the measuring section A2 of the three-dimensional object. It will be.

(2-2) Three-dimensional object recognition technology section A12 The object to be inspected first passes through the recognition stage at the entrance of the inspection system. When the recognition unit A12 receives a signal that the inspection object has arrived, the recognition unit A12 automatically selects the inspection method.
A feature extraction request is issued to the automatic programming unit A11, and the feature amount data (measurement data) of the object to be inspected from the measurement unit A22.
In response to this, recognition is performed based on this. After the recognition, the type and position / orientation of the inspected object, which is the recognition result, together with the inspection model including the inspection information are automatically selected / automatically programmed for the inspection method A11 and the dimension / shape / defect inspection section A13.
Hand over to. The recognized inspection object is sent to the inspection stage by the transportation system.

The basic idea of the three-dimensional shape recognition is to compare the description of the inspection object obtained from the three-dimensional image of the scene with the description of the recognition model of the inspection object prepared in advance. To obtain the type of inspection object and the three-dimensional position / orientation. Therefore, 3D shape recognition is
As shown in FIG. 11, the recognition / inspection model creating unit A122, which creates a recognition model of the inspected object for recognition in advance, and the recognition model and the description of the scene are collated at the time when the measurement data comes in for recognition. It is composed of a three-dimensional object recognition unit A123.

By the way, the recognition / inspection model creating unit A12
In 2, the recognition model is often created based on the three-dimensional shape model of the inspection object. Considering that the object to be inspected is a product or a part of the product, it is assumed that the three-dimensional shape model is given from the upstream of the design process as shape data of the CAD model or product model, but there is no CAD model. Requires a three-dimensional shape model creating unit A121 that directly senses the surface of the object to be inspected or the clay model and creates a three-dimensional shape model from the obtained three-dimensional coordinate data.
Upon receiving a signal that the object to be measured is set on the measuring table, the three-dimensional shape model creating unit A121 issues a sensing command to the automatic selection / automatic programming unit A11 of the inspection method, and the measuring unit A22 measures the surface of the object to be measured. The three-dimensional coordinate data (x, y, z) is received and processed to create a three-dimensional shape model of the inspection object.

(2-3) Dimension / Shape / Defect Inspection A13 FIG. 12 shows the size / shape / defect based on the inspection model data after recognizing about 50 kinds of known three-dimensional inspection objects.
The reference model of the system for performing defect inspection and sorting / classification inspection is shown.

The information included in the inspection model includes the ID of the object to be inspected, information on the inspection command system such as inspection items and inspection parts, and basic information required for inspection such as dimensions, tolerances and inspection accuracy. Information such as the shape, position, and orientation indicating the state of the inspection object should be provided.

Based on the information of the inspection item in this inspection model, the inspection item selection unit A131 in the figure performs the inspection item selection process, and the selection result is used as the size / shape inspection unit A1.
32, or a defect inspection section A133 of appearance or an inspection section A134 of sorting / classification, a command is given as an inspection execution instruction.

Each of these inspections is performed individually in some cases, and when there are a plurality of inspection items, it is instructed by the inspection item selection unit A131 as an inspection execution instruction including the order of inspection processing, and the inspections are sequentially performed. Be seen.

In the execution process (A132-A134) of each inspection in response to the inspection execution instruction, the position / orientation information in the inspection model, the confirmation ID information of the inspection object, and further after the recognition of the inspection object are finished. A local coordinate system for inspection is set based on detailed data such as coordinate reference values included in the three-dimensional information.

Extraction of the portion to be inspected is performed by the combination of the coordinate values obtained by the setting processing of the coordinate system, the inspection site / inspection accuracy information in the inspection model, and the detailed measurement data in the three-dimensional information. .

When the coordinate value and the inspection accuracy of the portion of the inspection target portion are determined, the actual inspection is determined. this is,
It is executed according to the dimensions / shapes or tolerances included in the inspection model and the determination method / determination reference data stored in the inspection / evaluation database. In this case, the database also stores determination results as past inspection results and always feeds back the latest inspection as a determination reference value.

The inspection result is displayed and printed as inspection information together with the raw image data of the measurement and the data processed into numerical values and graphs, and at the same time, it is sent to the upstream process if necessary. It

(2-4) Three-dimensional object measurement A22 Various three-dimensional information obtained by the three-dimensional information acquisition unit A21 is input, and the measurement processing result is used as the three-dimensional object recognition unit A.
Output to 12. Depending on the contents of the input data, even a simple recognition process is performed and output to the dimension / shape / defect inspection unit A13.

As shown in FIG. 13, the data discriminating section A 225
Determines the input data and sets the data type ID necessary for subsequent selection of the processing method. The data is basically image data, but also dimension data and the like.

The types of image data are, for example, (b1) distance image (b2) height image (b3) tilt image (b4) grayscale image (b5) color image.

The processing selection unit A224 sets a processing ID required for subsequent measurement processing process selection according to the determination result of the input data. When the input is a monocular image (a configuration using one TV camera), a processing process for adding the missing three-dimensional information is performed. In the case of a compound eye image (a configuration using a plurality of cameras), the processing process of extracting the three-dimensional information is performed as it is.

The image dividing unit A223 selects an optimum image dividing method according to the selected process process ID and performs the process.

The image division method is, for example, (c1) binarization processing method (c2) edge detection method (c3) area method.

Data that does not need to be subjected to image division processing is passed as it is.

If no result is obtained by the selected processing, another processing method is selected to automatically retry.

The retry history data is added and output to the next process.

The feature extraction unit A222 selects the optimum feature extraction method according to the selected process process ID and performs the process.

The feature extraction is performed by the following method, for example.

(D1) Global feature extraction (d2) Feature focused on contour (d3) Extraction of positional relationship Data for which feature extraction processing does not need to be passed through as it is. Also, if the selected process does not produce a result, another process method is selected and the process is automatically retried.

The retry history data is added and output to the next process.

The data output unit A221 outputs the processed result according to the content. When performing the recognition process of the three-dimensional object, the measurement data obtained in this process is output to the three-dimensional object recognition unit A123. The measurement data obtained in this process is output to the dimension / shape / defect inspection unit A13 if it can be directly inspected and utilized. In the process of this process, the cases where data could not be obtained successfully are entered in the database and accumulated as know-how.

(2-5) Acquisition unit A21 of three-dimensional information In this block, the sensing data obtained by the active measuring unit A23 is processed, and the processed sensing data is combined into three-dimensional data to be measured by the measuring unit A22 of the three-dimensional object. Play the role of transfer.
The model development diagram will be described below with reference to FIG.

The sensing data includes information from the visual sensor and other information, and the sensing data classification unit A
It is classified by 212. The information from the visual sensor is normally input to the distance data generation unit A214 based on sensing data, but 1) monocular grayscale image, 2) depending on the acquisition method.
There are two types of data that have already been converted to the form of distance information, such as those based on triangulation and information from distance sensors. Regarding 1), it is processed into distance data by binarization and pattern matching (corresponding point determination). The monocular image needs some assumption in distance generation, that is, scene knowledge, which is prepared in advance in this part. When the three-dimensional data cannot be constructed only by the information in this block, the preprocessing may be performed only by the image preprocessing unit A213.

Information other than the visual sensor is 3) necessary for distance data generation as an element that constitutes a moving image such as operation of an optical system and conveyance of a product, and is input to the distance data generation unit A215 by relative position / motion. To be done. The inspection method expansion unit A211 extracts the inspection method from the inspection command information from the upper layer, and sends fixed parameters such as the positional relationship of the optical system, the visual field, and the projection pattern to the distance data generation unit A215 by relative position / motion. Here, together with the above 3), distance data generation is performed using information other than visual information.

Distance data generation unit A based on relative position / motion
215 and the distance data generated by the distance data generation unit A214 based on the sensing data complement each other such that if either one is one-dimensional coordinate data, the other is two-dimensional coordinate data, and finally 3 The three-dimensional data generation unit A216 generates three-dimensional data and sends it to the three-dimensional object measurement unit A22.

(2-6) Active measurement unit A23 FIG. 15 shows a model development view of active measurement. It is composed of five sub-modules that receive inspection command information from the host and output sensing data, scanning information, illumination information and the like.

The inspection command information input from the host is interpreted by the sensing work developing unit A231, and the conveyance control unit A235 for controlling the conveyance of the inspection object, the inspection object and the position / route of the sensor / illumination system, etc. Controlled object removal /
It is transferred to the inspection route control unit A234 as control information.
The types and number of lighting devices and sensors required for sensing, parameters unique to the sensing system, and the like are also simultaneously output to the lighting device selection unit A233 and the sensor selection unit A232. The sensor selection unit A232 to the transport control unit A235 perform a predetermined work according to the instructed information and return the sensing data and the state information of each unit to a higher level. As the sensing device, not only an image sensor such as a CCD or a PSD or a distance sensor, but also a sensor for measuring a physical quantity such as an electric signal or a mass is used auxiliary.

The object to be inspected is loaded on an inspection station by an AGV (Automatic Guided Vehicle, a moving body which guides and travels by electromagnetic waves and light) controlled by the transport control section A 235, a conveyor, etc. Etc. are set. If the object to be inspected is unknown, exploratory measurement is performed by the robot hand and the sensing system. After the host system recognizes the target based on this data, detailed inspection and measurement are performed again. After the measurement is completed, the object to be inspected is again conveyed by the transfer control unit A
It is conveyed to the next process by the control of 235.

An actuator such as a robot hand also constitutes a part of a sensing system for active measurement. Therefore, the sensing unit that actually performs the inspection includes three modules, that is, a sensor group, a lighting device group, and a robot hand.

Table 1 shows, as an example, the types of the inspection object and the inspection items in the embodiment.

[0075]

[Table 1]

[0076]

As described above, according to the present invention, the inspection of the inspection object can be performed automatically, efficiently, and in a necessary and sufficient manner.
It can be done, which saves manpower and wastes time.

Further, according to the present invention, it is possible to automatically obtain product design and manufacturing correction information from the inspection result of the inspection object, so that the design can be reviewed. Become.

[Brief description of drawings]

FIG. 1 is a perspective view showing the overall configuration of an embodiment of the present invention.

FIG. 2 is a block diagram showing the overall configuration of an active inspection / measurement apparatus according to an embodiment of the present invention.

FIG. 3 is a block diagram showing an outline of the overall configuration of an embodiment of the present invention.

FIG. 4 is a block diagram showing in more detail a portion of the configuration of the embodiment shown in FIG.

5 is a block diagram showing in further detail another portion of the configuration of the embodiment shown in FIG.

FIG. 6 is a block diagram showing the overall configuration of an automatic inspection system according to an embodiment of the present invention.

FIG. 7 is a block diagram showing the entire configuration of the embodiment in an expanded manner.

FIG. 8 is a developed block diagram of recognition and inspection unit A1 of a three-dimensional inspection object.

FIG. 9 is a block diagram showing a developed configuration of a measurement unit A2 of a three-dimensional inspection object.

FIG. 10 is a block diagram showing a developed configuration of an automatic selection / automatic programming unit A11 of an inspection method.

FIG. 11 is a block diagram showing a developed configuration of a three-dimensional inspection object recognition technology unit A12.

FIG. 12: Inspection part A1 for the size, shape, and defect of the inspection object
3 is a block diagram showing an expanded configuration of FIG.

FIG. 13 is a block diagram showing a configuration of a measurement unit A22 of a three-dimensional inspection object.

FIG. 14 is a developed block diagram of a configuration of a three-dimensional inspection object information acquisition unit A21.

FIG. 15 is a developed block diagram of a configuration of an active measurement unit A23.

[Explanation of symbols]

 1 Inspection object 2 Autonomous unmanned transfer robot 3,4,5 Handling robot 6 Measuring table 8 Lighting device 9,10,11 Robot A0 Automatic appearance inspection system A1 3D inspection object recognition and inspection unit A2 3D Measurement part of inspection object A11 Automatic selection of inspection method Automatic programming part A12 Recognition part of three-dimensional inspection object A13 Dimension / shape / defect inspection part A21 Acquisition part of three-dimensional information A22 Measurement part of three-dimensional object A23 Active measurement Department

[Procedure amendment]

[Submission date] October 26, 1993

[Procedure Amendment 1]

[Document name to be amended] Statement

[Correction target item name] 0011

[Correction method] Change

[Correction content]

Further, according to the present invention, the factor of the defect detected by the inspection is estimated, and the calculation result is output to the portion for designing and production planning by the communication means, and thus the defect is detected from the inspection result. Perform factor analysis and send study results to design and manufacturing departments. Therefore, the design can be reviewed and the production can be examined, and for example, the design defect can be corrected.

[Procedure Amendment 2]

[Document name to be amended] Statement

[Name of item to be corrected] 0023

[Correction method] Change

[Correction content]

Automatic appearance inspection system A shown in FIG.
0 is a three-dimensional object recognition and inspection unit A shown in FIG.
1 and a three-dimensional object measuring unit A2. Figure 6
Resource is a physical means necessary for performing an inspection, and includes, for example, an inspection tool, a fixture, a transportation device, and all other objects than an inspection object. As shown in FIGS. 7 and 8, the three-dimensional object recognition and inspection unit A1 includes three-dimensional information A2a and CA measured by the three-dimensional object measuring unit A2.
From the product model data A1a accompanied by the information about the inspection such as the shape sent from the design system by D,
A three-dimensional object recognition unit A12 for generating an inspection model necessary for inspection and recognizing a specific three-dimensional object from a plurality of models is provided, and an expert system is provided from the generated inspection model according to the purpose of inspection. Automatic selection and automatic programming of inspection method for automatically selecting inspection method and automatically generating inspection program by
1. By comparing inspection model and measurement data,
Dimension / shape / defect inspection unit A13 for inspecting defects
Is.

[Procedure 3]

[Document name to be corrected] Drawing

[Name of item to be corrected] Figure 5

[Correction method] Change

[Correction content]

[Figure 5]

 ─────────────────────────────────────────────────── ─── Continuation of the front page (71) Applicant 000005049 Sharp Co., Ltd. 22-22 Nagaike-cho, Abeno-ku, Osaka-shi, Osaka (71) Applicant 000003931 Niigata Iron Works Co., Ltd. 1-4-1, Kasumigaseki, Chiyoda-ku, Tokyo ( 71) Applicant 000113425 Honda Engineering Co., Ltd. 1-10-1 Shin-Sayama, Sayama-shi, Saitama Prefecture (71) Applicant 000003997 Nissan Motor Co., Ltd. 2-2 Takara-cho, Kanagawa-ku, Yokohama-shi, Kanagawa (72) Inventor Yasushi Otsuki Akashi, Hyogo-ken 1-1 Kawasaki-cho, Ichi, Kawasaki Heavy Industries, Ltd. Inside Akashi Plant (72) Inventor Sumihiro Ueda 1-1, Kawasaki-cho, Akashi-shi, Hyogo Prefecture Akashi Plant, Kawasaki Heavy Industries, Ltd. (72) Takao Kanemaru Akashi-shi, Hyogo Prefecture 1-1 Kawasaki Town Kawasaki Heavy Industries Ltd. Akashi Factory (72) Inventor Hirofumi Morita Yokohama City, Kanagawa Prefecture 1-14-1 Jigokusho JGC Corporation Yokohama Office (72) Inventor Satoshi Kimura Minami-Differential Office 1-14-1 JCM Corporation Yokohama Office (72) Inventor Yoshisuke Watanabe Osaka Prefecture Osaka Kubota Co., Ltd. 1-247 Shishizu Higashi, Naniwa-ku, Osaka (72) Inventor Atsushi Ohashi 1-247 Shikitsu Higashi-shi, Naniwa-ku, Osaka City, Osaka (72) Inventor Hideo Nakamura Osaka 22-22 Nagaikecho, Abeno-ku, Osaka-shi, Osaka Prefecture (72) Inventor Minamiichi Minamiichi 27, Shinisogo-cho, Isogo-ku, Yokohama-shi Kanagawa Prefecture Niigata Ironworks Yokohama Development Center (72) Inventor Wakahara Taketo, 27, Shinisogo-cho, Isogo-ku, Yokohama-shi, Kanagawa Stock company Niigata Ironworks Yokohama Development Center (72) Inventor Masatoshi Nomura 1-10 Shin-Sayama, Sayama-shi, Saitama Prefecture Honda Engineering Co., Ltd. (72 Inventor Haruo Koyanagi 2 Takaracho, Kanagawa-ku, Yokohama-shi, Kanagawa Nissan Motor Co., Ltd. In the expression company

Claims (6)

[Claims] 1. A means for handling an inspection object, a sensor / illumination system, a means for controlling the position and orientation of a sensor or an illumination device, a means for processing a sensor signal, and an identification / recognition of the object. One calculation means, means for communicating the data of the object with the outside, and second calculation means for determining a measurement method or inspection method suitable for the object from the data and the recognition result are included. Automatic inspection device. 2. A third arithmetic means for analyzing a factor of a defective portion from an inspection result by the sensor signal processing means, and a means for outputting the arithmetic result of the third arithmetic means to an external design means or manufacturing means. Claim 1 characterized by the above-mentioned.
The automatic inspection device for the three-dimensional object described. 3. Shape data output means for outputting data relating to the shape of the inspection object; means for measuring at least one of the three-dimensional shape, position, and orientation of the inspection object; shape data output means; In response to the output of the measuring means,
The inspection object is recognized, the inspection command information for measurement is output, and the measurement means uses the inspection command information to perform recognition / command means for measuring a predetermined part of the inspection object, and shape data output means. An automatic inspection device for a three-dimensional object, comprising: a device for inspecting a defect or the like of an inspection object in response to an output from the measuring device. 4. The recognition / command means determines an inspection method for inspecting an inspection object and an inspection tool of a measurement means for performing the inspection method, and operates the measurement means. Item 3. The automatic inspection device for a three-dimensional object according to item 3. 5. The inspection tool has optical detection means for optically inspecting the inspection object, a light source for irradiating the inspection object with light, and inspection conditions for predetermining the optical inspection means and the light source. 5. The automatic inspection device for a three-dimensional object according to claim 4, further comprising: an operating unit. 6. A handling means for determining the position and orientation of an object, a sensor, an illumination system, etc. so that the measurement necessary for identifying the object can be realized from the data of the inspection object registered in advance and the initial measurement data. To generate the command, to compare the registered data with the measurement data to identify the target object, and to implement the inspection required for the identified target object, sensor and lighting. An automatic inspection method for a three-dimensional object, comprising: a step of determining a position and orientation of a system etc. and generating a command to a handling means; and a step of outputting an inspection result to the outside.