三坐標測量機CAD尺寸檢測新算法

1、Int J Adv Manuf Technol (200016:107112©2000Springer-Verlag London LimitedA New Algorithm for CAD-Directed CMM Dimensional InspectionY.-J. Lin and P. Murugappan*Department of Mechanical Engineering, The University of Akron, Ohio, USAThe recent developments in computer integrated manufacturing (C

2、IMsystems have made the traditional dimensional inspec-tions bottlenecks in the production line. To overcome these bottlenecks, computer integrated dimensional inspection was proposed with the coordinate measuring machine (CMMbeing the key device. In this investigation, a framework for integrat-ing

3、the CMM into the computer-aided design/computer-aided manufacturing (CAD/CAMenvironment is developed to auto-mate the process of design, manufacturing and inspection. An algorithm to generate an optimum collision-free CMM probe path is proposed. This algorithm uses the ray tracing technique to locat

4、e the collision of the possible paths with the workpiece to be inspected, between the initial probe point and the target point. If there is a collision, the algorithm walks through the topological structure of the part and selects the midpoint of the edge shared by the face with which the path colli

5、des and the adjacent face nearest to the target point, as the next probe point. This procedure is followed until the target point is reached. The rst half of the proposed algorithm is implemented using Mechanical Desktop as the CAD system and AutoCAD Runtime Extension (ARXas the application programm

6、ing interface, running on a Windows NT 4.0plat-form. The effectiveness of the proposed algorithm is veried by the results of the implementation which demonstrate optimum collision-free dimensional inspection path generation for three prismatic parts.Keywords:Computer integrated manufacturing (CIM;Co

7、-ordinate measuring machine (CMM;Dimensional inspection 1. IntroductionA coordinate measuring machine (CMMis the key device used in computer-integrated dimensional inspection 1.The essential system components of a CMM are, the mechanical *Present address :The Structural Dynamics Research Corporation

8、, Cincinnati, Ohio, USA.Correspondence and offprint requests to :Dr Y.-J. Lin, Department of Mechanical Engineering, The University of Akron, Akron, OH 44325 3903, USA. E-mail:YL arrangement with the three axes and the displacement trans-ducers, a probe head with a touch trigger or analog

9、ue probe to probe the workpieces in all spatial directions, a control unit and a computer with peripheral equipment and software to calculate and display the results 2.Although CMM has a considerable level of automation compared with traditional methods, CMMs still have to be manually operated to ca

10、rry out the rst part inspection, in the case of on-line inspection planning. The CAD software then generates the DMIS (dimensionalmeasurement interface specication program. DMIS provides a neutral interchange format between CAD systems and the dimensional measuring equipment. The program is then rep

11、layed for inspection of the remaining parts, in a batch production environment. In the case of off-line inspection planning, the whole procedure is simulated by the CAD software. In order to achieve a higher level of automation, CAD-directed inspection planning is essential.In CMM inspection path pl

12、anning, optimisation of the inspection path plays a major role in increasing the throughput of the CMM. The path planning for a CMM is similar to that of a Cartesian robot path planning, the goal being to generate an optimum collision-free path. In the inspection process, the CMM probe has to move i

13、n the workspace to touch different areas of the part for measuring purposes. Since the probe works in the proximity of the part, the geometry of the part and xtures create obstacles that interfere with the movement of the probe. Hence, the generation of an efcient and optimum collision free CMM insp

14、ection path forms a major problem in the inspection planning.A new algorithm is proposed to generate an efcient CMM inspection path by accessing the CAD database for the part geometry and traversing through solid representations in the CAD system for collision detection and path generation. The obje

15、cts considered are simple prismatic solids. To lay the ground for the presentation of this new algorithm, past related work on path planning is briey reviewed and the overall framework and the goal of the proposed algorithm are described in Section 2. The proposed algorithm and its verication are di

16、scussed in Section 3. The results are shown in Section 4. Finally the conclusions and recommendations for future study are given in Section 6.108Y.-J. Lin and P. Murugappan2. Review of Related WorkCMM inspection path planning has received wide attention in recent years. There has been research using

17、 CAD databases for CMM inspection path generation. A computer-integrated environment for automated dimensional inspection was pro-posed 1which plans the inspection path by searching for functional features from the CAD database using the speci-cation module, and generates the path using the planning

18、 module. The path generated is checked for collisions using the verication module and is executed using the execution module. The measured data is then compared with the specied dimen-sions and tolerances using the analysis module.The specication module accepts various inspection attributes from use

19、rs to generate the inspection specication which is used by the inspection planning module to search for the features and divide them into single features and related fea-tures. Based on the geometry of each individual feature and its relationship with the inspection specication, the number of inspec

20、tion points, point locations, probing vectors, backoff distance, the probing sequence is determined. These compo-nents are checked for interference with part surfaces and a collision-free inspection path is generated. In this approach much effort is related to searching the CAD database using the sp

21、ecication given, rather than using the part geometry or the solid representation of the features and generating a colli-sion-free path.Graphic representation methods can be classied into sweep representation methods, conguration space methods and space subdivision methods. Sweep representation denes

22、 a new object by sweeping an object along a trajectory through space. Then the overlap or intersection between the swept volumes and the obstacles is determined, and a new path is proposed. Since simple surfaces are used to model the object accurately, it is difcult to determine whether two such mod

23、els overlap. More-over, the proposed path provides only local information about the potential collision, for example, the shape of the intersec-tions of the volumes, or the identity of the obstacle giving rise to the collision. This information suggests only local path changes, but not a radically d

24、ifferent path if that is required 3.This results in an expensive search of the space of possible paths.In a conguration space method 46the objects considered are polyhedrons. Conguration space characterises the position and orientation of an object. This is formed by shrinking the object to a point

25、while expanding the obstacles to account for the object dimensions. The remaining free space is left to translate the point object between the starting and nal pos-itions. This method uses the vertices of all the transformed obstacle polyhedrons to form every possible visible path. A vertex graph is

26、 then used to search for the shortest vertex path. The computational time increases exponentially as the number of obstacles increases 7.In a CMM, the part is xtured to its base, so this leads to a long search time as the xtures become obstacles.The space subdivision method can be divided into two a

27、pproaches, namely, spatial enumeration 8and octree database 9,10methods. Many algorithms have been developed based on an octree data structure. Octree is a recursive method of representing 3D objects in a hierarchical eight-array tree struc-ture. The moving object and the obstacles are modelled by s

28、weeping volumes in octree methods, which makes collision detection difcult. A common problem in these methods is the zigzag nature of the path which involves acceleration and deceleration of the probe. This, in turn, leads to low path efciency due to vibration induced measurement errors and longer t

29、ravelling time.Lu et al. 11proposed an algorithm which uses a modied 3D octree ray tracing technique to search for the colliding obstacles in an octree database on a selected path. It also uses the global information about the obstacle vertices to reduce the zigzag nature of the path generated by th

30、e octree based methods. This approach represents the part using an octree data structure, which is not widely used in CAD solid model-ling, since it approximates the geometry of the part with cells. The commonly used solid representations are B-reps (boundary representations and CSG (constructivesol

31、id geometry. More-over, since the CAD model contains complete information about the part dimensions, tolerances, geometry, orientation, etc., this information can be used directly for inspection path planning. Therefore, it avoids the process of representing the workspace and object in either specia

32、l functions or a graphic database, which is time consuming.This research develops a framework for CAD directed inspection planning and implements the basic structure of the framework 12,13.3. Proposed FrameworkThe main objectives of the proposed framework are:1. To integrate the CMM into the CAD/CAM

33、environment, so that the part designed and manufactured can also be inspected in a single cycle.2. To incorporate feature based inspection planning at the design stage.3. To generate an efcient and optimum collision-free CMM inspection path.4. To generate a DMIS (dimensionalmeasurement interface sta

34、ndard output for on-line analysis on the CMM machine itself.The framework mentioned above is illustrated in Fig. 1. The geometry of the part to be measured is rst queried in the CAD system database. The CMM probe path from the home position to the target point is generated by traversing through the

35、solid representation of the part in the CAD system, using a modied ray tracing technique to avoid collisions.As the ray advances a decision-making technique is used to generate an optimum collision-free path. The generated path is then displayed on the computer screen for the user to visualise. Init

36、ially, the algorithm is developed to measure a point on a feature. The algorithm is then extended for geometry-feature-based path planning, in which the user has the exibility of selecting geometric features of the part to be measured. The nal goal of the proposed methodology is to generate a DMIS o

37、utput for on-line inspection using the CMM.CAD-Directed CMM Dimensional Inspection 109Fig. 1. Framework of the proposed work.As the probe path is generated by checking for intersection or collision of the ray with the different faces of the part, traversing through the data structure of the solid re

38、presentations of the part in the CAD system becomes an important task. To obtain a better understanding of the algorithm developed, a brief introduction is given in the next section to the different representations and data structures in solid modelling.4. Proposed AlgorithmThe algorithm developed g

39、enerates a collision-free CMM inspection path, which plays a major role in the proposed framework. The development phase explained in this section can be divided into three major phases:1. Developing a general algorithm for path generation.2. Selection of a CAD system with an API (applicationpro-gra

40、mming interface.3. Implementation of the algorithm.The main goal of the present work is to develop a general algorithm for CMM inspection path generation, which can be implemented with any CAD system API. The major assump-tions in the algorithm are:1. The CMM probe is assumed to be a point object. T

41、his helps in converting collision detection of the moving probe with the part, into the simpler detection of collision of a single point with the part.2. The other important assumption is that the part is not xtured, i.e. xtures are not included in the collision detec-tion with the probe, for simpli

42、city.3. The user species the home position or starting point of the CMM probe.The algorithm developed has several steps and cases to be implemented. The complete owchart of the algorithm is shown in Fig. 2.The explanation of the algorithm is as follows:Step 1. A session between the CAD system and it

43、s API is initialised to run the application.Step 2. As the link between the CAD system and its API is established, the user is prompted to input the target point and the initial position of the probe, and to select the part to be measured.Step 3. As the user selects the part, the underlying solid ob

44、ject is opened from the CAD system database for obtaining the necessary geometric information for collision detection and path generation.Step 4. The collision between the probe path and the part measured is detected using a ray tracing technique. Ray tracing is used in image processing for surface

45、shading or hidden line removal. It determines the visibility of surfaces by tracing imaginary rays of light from the viewers eye to the objects in the screen. The imaginary ray is nothing but an imaginary line connecting the starting and the ending nodes. The intersec-tion of the ray with the object

46、 is determined, to set the required colour at that pixel.Fig. 2. Algorithm owchart.110Y.-J. Lin and P. MurugappanIn this algorithm the concept of ray tracing is used in determin-ing the collision of the probe path with the part. An imaginary ray connects the initial probe point and the target point.

47、 The intersection of this ray with the part is determined. If there is no intersection or collision of the ray with the part, then the algorithm chooses the line connecting the initial probe point and the target point as the nal collision-free path. If there is an intersection, the algorithm detects

48、 a collision and goes on to the next stage.Step 5. Once a collision is detected, a pointer is initialised to the B-rep face, which intersects with the ray. This face is called the base face. Using the pointer to the base face, the adjacent faces of the base face are traversed and checked for the tar

49、get point. If the target point is found on any of the adjacent faces, the midpoint of the common edge of the base face and the adjacent face is selected as a collision-free intermediate point. The algorithm selects the path connecting the probe starting point, the intermediate point and the target p

50、oint as the nal path.Step 6. If the target point is not on any of the adjacent faces, then depending on the proximity of the edges in the base face to the target point, two edges are selected to shoot rays to their respective midpoints from the initial probe point. Taking the endpoints of these two

51、rays as initial points of two new rays, the ray tracing procedure is carried out. Let the intersec-tions of these two rays be i 1and i 2. The algorithm developed involves three cases, depending on the number of intersections these rays make with the part before they reach the target point. By referr

52、ing back to Fig. 2, these three cases are elaborated as follows:Case 1. If the number of intersections is equal, then the path is selected so that it goes around the solid. The total distances travelled by the two rays are calculated and the shortest one is selected as the nal path.Case 2. If the nu

53、mber of intersections, i 1=1and i 21, or vice versa, then the ray with one intersection is selected and the algorithm goes back to step 5.Case 3. If i 11and i 21, then the rays proceed in the forward direction to the opposite edges on the adjacent faces of the base face. The midpoints of the opposit

54、e edges become the new starting points. The coordinates of the new starting points are compared with the target point. If both the starting points are farther away from target point, then the algorithm goes to step 5. If one of them is farther from the target point, then the respective ray is neglec

55、ted. The algorithm goes back to step 5with a single ray in its database.The algorithm repeats steps 5and 6until a collision-free path is obtained. Thus, a generalised algorithm for collision-free CMM inspection path generation is developed.The CAD system used in the implementation is Mechanical Desk

56、top from AutoDesk, which contains AutoCAD 13c4, AutoSurf and AutoCAD Designer modules. The structure of the CAD system and its APIs are shown in Fig. 3. The API used in the current implementation is ARX (AutoCAD Runtime Extension Software Development Kit.An ARX application is a dynamic link library

57、(DLLthat shares the AutoCAD address space and makes direct functionFig. 3. CAD system and its APIs.calls to AutoCAD. In this research the algorithm developed was implemented to step 5. The interface to the application developed and the results obtained are discussed in the next section.5. Results of

58、 Computer Implementations The algorithm developed was implemented to step 5, as explained in the previous section, i.e. the algorithm generates a collision-free path if the target point is on any of the faces adjacent to the base face. The algorithm handles simple pris-matic solids well. Solids, wit

59、h curved surfaces and free-form surfaces are not supported at present. The algorithm was implemented using the ARX SDK for AutoCAD and the API code was written in C using Microsoft Visual C . In the present implementation, users run the application like any other native AutoCAD command. The registered AutoCAD command for this application is INSPECT. Users have to load the ARX application