A new memory efficient tool path generation method for applying very large STL files in vector-by-vector rapid prototyping processes

Transcription

1 A new memory efficient tool path generation method for applying very large STL files in vector-by-vector rapid prototyping processes Bahram Asiabanpour, Ph.D. Candidate Behrokh Khoshnevis, Professor D. J. Epstein Dept. of Industrial & Systems Engineering University of Southern California Los Angeles, CA Phone: (213) Fax: (213) Abstract There are several Rapid Prototyping (RP) processes, such as Contour Crafting, SLS, and LOM, that fabricate physical models through vector-by-vector machine control. For these processes the tool path is generated from a CAD model that is saved in the STL file format. The common procedure for tool path generation from a CAD model includes vector generation through intersecting a XY plane and STL facets, and then by sorting individual vectors. Conventionally this procedure is conducted by reading entire facets into the computer memory. Then the facets are sliced and generated vectors are sorted. This procedure needs a considerable computer memory and it can not proceed for large STL files with tens of mega bites size. A new tool path generation algorithm is proposed that solves the computer memory bottleneck in the tool path generation from very large STL files. The system receives an STL file as input and works in two steps. The first step generates slices at each increment of Z by intersecting the XY plane with the facets within the parts. In this step one facet at a time is read from STL file. Step two includes sorting the intersection lines, recognizing of the closed loops and disconnections, generating tool path file, and generating simulation file. In this step, one vector at a time is read from a slice file and is written into an output file, therefore, there is no limitation for size of STL file. In this system only data of the start point of the loop, contour end point, and current vector are saved in the computer memory. The system has been successfully tested on many ASCII STL files as large as up to 200 MB. The tool path generation system has been customized for Selective Inhibition of Sintering (SIS) process by adding hatch path to tool path file. 383

2 1- Introduction Numerical Control is a method to control a machine by computer through a set of numeric and alphabetic instructions. This is the method to control a wide variety of automated machines, such as CNCs, robots, and Rapid Prototyping (RP) machines. For the robots and CNC machines, these codes have been mostly standardized in the form of a few languages. But there is no standard code for rapid prototyping processes, because of the differences in the natures of these processes. Instead, there have been several efforts to establish a standard CAD file format for the rapid prototyping processes. Therefore, each RP process, based on its characteristics and requirements, uses the standard CAD file format to extract the required data for the process. Depending on layer fabrication, object prototyping processes can be classified into two different groups: vector-by-vector processes and integral processes (Vogt, G. et. al., 2002). In vector-by-vector processes every layer is made in an incremental method. In these processes each contour path is formed from a set of individual vectors. In integral processes a complete layer of prototype is made in one step. Although integral processes have advantages over vector-by-vector processes, especially in term of their speed, the nature of many rapid prototyping processes, especially laser-based and extruding-based processes, makes the vector-by-vector processes popular. There are several RP processes, such as SLS, LOM, and Contour Crafting, that fabricate physical model through vector-by-vector machine control. For these processes the tool path is generated from a CAD model which is saved in the STL file format. The common procedure for tool path generation from a CAD model includes vector generation through intersecting a XY plane and STL facets and then by sorting individual vectors. Conventionally this procedure is conducted by reading entire facets into the computer memory. Then the facets are sliced and generated vectors are sorted. This procedure needs a considerable computer memory and it can not proceed for large STL files with tens of mega 384

3 bites size. To overcome the memory limitation Choi and Kwok (2002) have proposed storing one slice at a time into computer memory to sort the vectors. Their approach decreases the amount of required computer memory greatly. However, this approach will not guarantee unlimited size for STL files because a considerable amount of memory is used to save and sort vectors in one slice especially for very large STL files. In this paper a new tool path generation algorithm is proposed that solves the computer memory bottleneck in the tool path generation for very large STL files. The paper is organized as follows. In section 2, slicing and tool path generation algorithms are described. In section 3, algorithms implementations and an example are presented. System customization for the SIS process is explained in section 4. Finally, discussions and conclusions on the tool path generation algorithm are given in sections 5 and 6, respectively. 2- Slicing and tool path generation algorithms To generate a tool path, an STL file with the ASCII format is received as input and the data is processed in two steps. The first step generates slices at each increment of Z by intersecting the XY plane with the facets within the parts. The second step includes sorting the intersection lines, recognizing the closed loops and disconnections (a common STL file error), and generating tool path and simulation files. Figure 1 illustrates tool path generation procedure. Slice 1 STL file Step 1 Slicing Slice 2 : : Slice K Step 2 Tool path generation Figure 1.Two steps of slicing and tool path generation Tool Path file Simulation file 385

4 2-1- Step 1: Slicing algorithm In the first step the STL file is read as the input file. By executing the slicing algorithm, slice files are generated. Only the intersection of those facets that intersect current Z=z are calculated and saved. In this step, XY plane for specific increments of Z intersect surface facets. The slicing algorithm reads one facet at a time and after vector generation for all z increments, it reads the next facets; therefore, it does not require a large amount of computer memory and also there is no limitation for the model size. In this step each slice is saved in a separate file. This guarantees that step 2 is run faster than when we save all slices in one file Step 2: Tool path generation After the slicing process, a set of unsorted vectors is available in each z increment. These vectors are not connected and are not in sequence. Therefore, for running in the RP machine, these vectors should be arranged in an appropriate form.in step 2, sorted vectors and output condition are written into path file. Output condition could be On, when the laser beam is on in an SLS machine or there is extruded output from contour crafting machine, or Off, when the laser beam in an SLS machine or extruder nozzle in a contour crafting machine traces the part surface in the off condition. At the beginning, the system turns the output On and writes the first vector into a path file and tries to find the next connected vector to this vector by reading one vector at a time from a slice file. Then the system does the same for the newly found vector until reaching the start point of the first vector (in closed loop cases) or finding no attached vector (in disconnection cases). In either case the system will turn the output Off and start from another vector. After arrangement of all vectors in one slice (z increment), the process starts the arrangement of the vectors of the next slice. This process is continued until all vectors are sorted. 386

5 Start: k=1, K=. of files Temp_count=0 i=1 Read slice file k k=k+1 i=1 Read vector V i (S),V i (E) Slice file 1. Slice file k.. Slice file K Path file Printer Off, Write V i (S) into output files Printer On, Write V i (E) into output files i=i+1 Read the next vector V new (S) = V i (S) V new (E) = V i (E) Sim. file Find and read the nearest point in the rest of slice file. Printer On, Write the V i (E) into the output files V last (S) =V i (S) V last (E) =V i (E) V last (E) = =V new (S) Temp file V last (E) = V new (E) Reversing the vector V last (E) = =V new (E) Write the vector (V i (S), V i (E)) into the Temp file End of loop? Temp_count = Temp_count +1 End of the Slice file k? i=i+1 Temp_count = =0 Read all slice files? Delete slice file k Rename Temp file to slice file k Temp_count = 0 STOP There is a disconnection. V last (E) ==V last (End) V last (End) =V last (E) Figure 2.Tool path generation algorithm flowchart 3- Algorithm implementation Slicing and Tool path generation algorithms have been implemented with C programming language. This software has been successfully tested for several medium and large STL files on different PCs and laptops. To demonstrate the tool path, a script file is generated and is run in the AutoCAD environment. Table 1 and figures 3 and 4 show the results for an STL file in the size of about 8 MB on a Pentium II computer with 32 MB RAM in Windows

6 Sample File Size Step Height Ave. Time per Slice Truck 7.92 MB 0.1 mm sec. Table 1.Path generation time result Figure 3.A sample CAD model (truck) Figure 4.Visualization of the tool path in the AutoCAD environment 4- System customization SIS process is a new rapid prototyping system that fabricates parts by joining polymer powder particles through sintering in the part s body, and by sintering inhibition of some selected powder bed areas. The selected areas of the powder bed for sintering inhibition are wetted by a printer. A single nozzle printer moves 388

7 in the XY plane and prints the inhibitor liquid on the model slice pattern. A tool path file is required to control printer nozzle movement. Tool path file includes part boundary path and hatch path. The tool path generation system explained in this paper was customized for SIS process by adding hatch path generation option to the system. The system generates hatch path when it is required for the process. Figure 5 illustrated some parts made by SIS process by using customized path generator system. Figure 5.Sample parts made by SIS machine 5- Discussion In the slicing step, only the intersections of those facets that intersect current Z=z are calculated and saved. Also the algorithm reads one facet at a time and after vector generation, it reads the next facets, therefore, it does not require a large amount of computer memory and also there is no limitation for the model size. Also since the generated vectors of each z increment are written into a separate file, the vector sorting is faster than when all vectors are written into a single file. In the sorting algorithm, the algorithm reads the data of one vector from an input file and writes it in another file (temp file or path file). In other words, the system saves only data of the start point of the loop, last vector end point, and the current vector in the computer memory; therefore, the sorting process does not need a huge amount of memory to sort the data, and also there is no limitation on the number of vectors in a slice. However, because of the continuous file input/output 389

8 communications in this algorithm, the tool path generation is relatively slower than the commercial tool path generator systems. This disadvantage can be improved by using faster disk Input/Output controllers. 6- Conclusion In this paper a slicing and path generation method is presented that solves the memory shortage for very large STL files. The system has been successfully tested on many ASCII STL files as large as up to 200 MB. In this system since only data of the start point of the loop, last vector end point, and current vector are saved in the computer memory, even larger STL files can easily be processed to generate a tool path file even on computers with very low memory. By adding a hatch pattern generation system, this system is customized for SIS process, a new layered-based fabrication method, to make complex polymeric 3D parts. 7- References Choi, S.H., and Kwok, K.T., A tolerant slicing algorithm for layered manufacturing, Rapid Prototyping Journal, Volume 8 Number 3, Dolenc, A., Makila, I., "Slicing procedures for layered manufacturing techniques", Computer Aided Design, Volume 26, Number 2, PP , Khoshnevis, B., Asiabanpour, B., Mojdeh, M., Koraishy, B., Palmer, K., and Deng, Z., SIS A NEW SFF METHOD BASED ON POWDER SINTERING, 13 th International Symposium on Solid Free From Fabrication, Austin, Texas, August Vogt, C., Bertsch, A., Renaud, P., and Bernhard, P., Methods and algorithms for the slicing process in microstereolithography, Rapid Prototyping Journal, Volume 8 Number 3, Wai, H., "RP in art and conceptual design", Rapid Prototyping Journal, Volume 7, Number 4, PP ,