- College of Engineering
Contents
1. To remedy this major flaw the Voronoi method was considered The Voronoi method involves partitioning a region into sections based on a set of points which is shown in Figure 11 Everything within a particular section is closest to the point that generated that section 11 This method involves setting results in the entire student points in the same cell associated to the teacher point used to generate that cell For any number of student and teacher points the Voronoi method will return the appropriate result however the significant computational complexity of the Voronoi method makes it difficult to implement especially in three dimensions Ultimately the Voronoi method was abandoned INPUT OUTPUT Figure 11 Voronoi diagram Closest Point 2 0 The solution to the SimPooch grading issues is similar to the form of the closest point algorithm It provides an association between any student point and its closest teacher point It accounts for the case of more or less student points than teacher points and the case of more than one student point for any given teacher point The current successful SimPooch grading algorithm shown below in Figure 12 is as follows For student point find the closest teacher point Save the student point the teacher point and the distance between them into one row of an array Continue for each student point without removing the previous teacher point When the process is complete the results are returned in the2. could not be created a surface based model was implemented with reasonable success Aside from these advancement was made in the areas of improvement to the program structure functionality and a wealth of resources and direction provided to next year s team Overall the SimPooch team had a productive and effective year and sizeable contributions were made to the project as a whole Accomplishments One of the primary accomplishments this year is the development of a surface based haptic model The initial requirements specified that a model exhibit the sensation of layers which was created The magnetic spheres model gives a clear sensation of layered tissue and although it has some undesirable effects it is a major step in the right direction of haptic programming for SimPooch The second primary accomplishment of this year s SimPooch team was the development and implementation of the grading algorithm Previous teams were unable to implement an effective robust algorithm to associate the student and teacher points It was determined that the Voronoi method was unnecessary and needlessly complicated and a simple effective solution to the problem was determined A third accomplishment of the SimPooch team is the improvement of our simulation program Not only were all of the known bugs repaired but the program now has extensively improved functionality The additions to the simulation program functionality are as follows A conversion method
3. from virtual units to millimeters displays the grade in a more intuitive manner so grades can now be displayed on the virtual model alongside the teacher points In the competition mode users have may save their grades for comparison with other users Finally the simulation can be repeated without closing and reloading the program and grading and display algorithms have been separated for ease of expansion Failures The primary failure of this year s SimPooch team was the inability to compile VHTK VHTK would have dramatically improved the volumetric haptic model for the dog and much time was spent debugging source code and graphics drivers Two factors contributed to this failure inexperience with compiling binaries with many dependencies and with debugging C source code that was not written by us Future teams would be inclined to learn more about this process as we were inexperienced Another issue VHTK had was a general lack of documentation and support It was developed and released as is to the public with only a handful of developers offering support The failure to develop the GUI was another issue dominated by the lack of time at the end of the spring semester It was not put as a high priority by the team and thus was constantly put on hold by more pressing issues in SimPooch development Research was done and much still needs to be done in the development of a user friendly interface Project Continuation Introduction
4. The direction of the project for next year depends on the composition of the team members involved Mechanical engineering students would concentrate on the physical design stage Electrical Computer engineering students would focus on improving the haptic representations and the simulation program Haptics Our attempts to use the tools provided by H3D VHTK and MedX3D was unsuccessful Future team members skilled in compiling and debugging C code would be better equipped to handle this With VHTK or another working volumetric haptic library a greatly improved haptic model can be implemented Using MedX3d or another medical image reader a volumetric model could be implemented directly from the CT scan data Failing that a simple volumetric model could be designed based on force tissue force tests Simulation Program The SimPooch simulation program provides an effective base for haptic development but improvements to its functionality are still required The main considerations are improvements to the rotational sensor algorithm and implementation of a GUI There are several areas of improvement for the simulation program writing USB Drivers for the A D controller adding controls and functionality to the GUI with wxWidgets improving the X3D model filling holes reducing extraneous polygons etc in MeshLab and implementing volumetric haptic properties in Python to create realistic forces of layers with VHTK These suggestions for improve
5. data structure shown below in Figure 13 The solution to the grading issue comes in the form of no longer removing points when they have been associated With the correlation algorithm set the actual grading is simply comparing the distance between each point to the maximum allowable passing distance specified by Dr Robinson 2 5mm By making measurements on the physical mannequin and comparing the distances in virtual units a conversion method was developed The function convert in DogStart py maps the distance in virtual units to millimeters using the formula where m is the output in millimeters and v is the input in virtual units m v 1094 390522 All grades are displayed in the form of pass fail with the distance specified in millimeters grade grade_array closest_point min_dist 99 0 k 0 s student point i t teacher point k d dist s t k length teacher_points i length student_points Yes return grade_array Figure 12 Flowchart for grading algorithm student_point 1 student_point 1 student_point 1 teacher_point 1 teacher_point 1 teacher_point 1 dist 1 student_point 2x student_point 2 student_point 27 teacher_point 2x teacher_point 2 teacher_point 2z dist 2 student_point 3 student_point 3 student_point 3 teacher_point 3x teacher_point 3 teacher_point 3 dist 3 Figure 13 Data structure of grading algorithm
6. dog_main x3d file Dog_main x3d defines the primary structure for the SimPooch program It contains references to the initialization program DogStart py and the simulation program Dog py H3D works by continuously looping through the nodes defined in dog_main x3d and taking appropriate action such as rendering graphics and forces or executing program scripts this process is called traversal 10 DogStart py The execution of the Python scripts begins with DogStart py DogStart py provides a primitive user interface and initializes the simulator according to the user s selection Four modes of simulation are available to the user student teacher demo and contest The student mode prompts the user for a set or pre defined acupuncture points that the user will attempt to touch The student then begins to attempt to touch the points correctly without being able to see them displayed on the virtual model When finished the students score is displayed and he she is prompted to continue with another set of points or exit the simulation The teacher mode is designed for the creation of acupuncture point sets A teacher would define a name for the set and touch the points to be graded These points are then saved and can be accessed through the student mode The demo mode was developed for demonstration of the SimPooch project at events like Colorado State University s Engineering Days The difference between the demo mode and student mode is the appearance o
7. hrs gt 0 2 lbf 0 88 N sms X axis gt 7 3 Ib in 1 26 N mm Stiffness Y axis gt 13 4 Ib in 2 31 N mm Zaxis gt 5 9 lb in 1 02 N mm 0 101 Ibm 45 g Position sensing x y z digital encoders Stylus gimbal Pitch roll yaw 5 linearity Interface IEEE 1394 FireWire port 6 pin to 6 pin Corporate Headquarters 1993 2008 SensAble Technologies Inc All rights SensAble Technologies Inc reserved OpenHaptics PHANTOM PHANTOM 15 Constitution Way Desktop PHANTOM Omni SensAble and Woburn MA 01801 USA SensAble Technologies Inc are trademarks or ft 1 781 937 8315 registered trademarks of SensAble Technologies ff 1 781 937 8325 Inc Other brand and product names are email info sensable com tademarks of their respective holders Product Web www sensabie com specifications are subject to change without notice Appendix D Software AutoHotKey Scripting package for rapid development C Mid level language for general software development CH Web scripting language DisReal X3D graphical world editor H3D API Open source haptic development platform MedX3D Open source medical image reader designed by H3D MeshLab 3D graphics viewer and editor MYSQL Open source database management software OpenGL Open Graphics Language for cross platform graphics development Python Object oriented language for general development VHTK Volume Haptics Toolkit for representing haptics
8. http www h3dapi org Accessed May 01 2009 Python Programming Language Official Website Python Software Foundation online Available http www python org Accessed May 02 2008 H3D documentation 2 0 DOXGEN 1 5 7 November 7 2008 online Available http www h3dapi org uploads api H3DAPI_20 docs H3DAPI html index html Accessed May 2 2009 A Maciel A biomechanics based articulation model for medical applications online Available http vrlab epfl ch amaciel projects These 3360 Maciel pdf Accessed May 2 2009 Volume Haptics Toolkit MediaWiki August 17 2008 online Available http www h3dapi org modules mediawiki index php Volume Haptics Toolkit Accessed May 2 2009 MedX3D MediaWiki October 7 2008 online Available http www h3dapi org modules mediawiki index php MedX3D Accessed May 2 2009 H3D API 2 0 User Manual SenseGraphics AB October 31 2008 online Available http www h3dapi org uploads api H3DAPI 20 docs H3D 20API 20Manual pdf Accessed May 2 2009 J Wijk and A Telea Visualization of Generalized Voronoi Diagrams online Available www win tue nl alext ALEX PAPERS Vissym01 paper pdf Accessed May 2 2009 Autohotkey quickstart tutorial Tutorial and Overview online Available http www autohotkey com docs Tutorial htm Accessed May 2 2009 13 A basic H3D player with wxWidgets MediaWiki January 11 2009
9. implementation the spring effect surface model Spring Effect Surface Another conceptual model was designed but not programmed because a more efficient method was found This model generates the feedback force by creating surface objects and force effect nodes at a touched point and then removing them when there is no longer contact between the device and the surface The sensation of layers is modeled as a set of spring nodes Springs are force effects that apply to a specific point in the haptic environment 6 Spring effect nodes work similar to magnetic surfaces the force is only felt within a specified distance of the spring The forces generated by a spring effect are the same for the force on a spring This would model what is called a Hookean material 7 because it depends on Hooke s Law F kx The force of three springs with different constants and escape distances can be made to provide an effect similar to the design criterion shown in the force vs depth diagram of Figure 6 This model is a two surface model The springs are created on the outer surface and a solid inner surface exists to act as the bone Notice that the only major difference between this model and the design criteria is the force felt by the user is not constant throughout the different layers This model will however give the sensation of passing from one layer to another with a different density Bone Skin Fat Muscle Force Depth Figure 6 Sprin
10. online Available http www h3dapi org modules mediawiki index php A basic H3D player with wxWidgets Accessed May 2 2009 VI Bibliography 1 H3D API 2 0 User Manual SenseGraphics AB October 31 2008 online Available http www h3dapi org uploads api H3DAPI 20 docs H3D 20API 20Manual pdf Accessed December 12 2008 Volume Haptics Toolkit User Manual Norrk ping Visualization and Interaction Studio August 29 2006 online Available http www h3dapi org modules vhtkdoc VHTK_Manual pdf Accessed December 12 2008 VII Appendices Appendix A Abbreviations A D API CT GUI TCM VHTK Analog to Digital converter Application Programming Interface Computed Tomography Graphical User Interface Traditional Chinese Medicine Volume Haptics Toolkit Appendix B Budget Item Contributions Expenses Description Year Available Funds Narda Robinson 1 000 ECE Sr Design 50 Total 1 050 Appendix C Phantom Omni Data Sheet Model The PHANTOM Omni Device Force feedback workspace 64Wx4 8Hx2 8 Din gt 160 Wx 120Hx70D mm Footprint Physical area the base of device 65 8 Wx8Din occupies on the desk 168 W x 203 Dmm 3b 15 07 Hand movement pivoting at wrist gt 450 dpi Nominal position resolution 0 055 mm Backdrive friction lt l 0z 0 26 N Maximum exertable force at nominal orthogonal arms position 0 75 lbf 3 3 N Continuous exertable force 24
11. volumetrically wxWidgets GUI development toolkit X3D Standard file format for representing 2D and 3D graphics Appendix E Acknowledgements Thanks to those who helped contribute to SimPooch Dr Narda Robinson Project Sponsor Dr Peter Young Project Advisor Dr Regina Schoenfeld Tacher Project Sponsor Olivera Notaros ECE Senior Design Coordinator The 2007 2008 SimPooch Team Brendan Dahl Ben Cordova Eric Hall Jeff Barlett The 2006 2007 SimPooch Team Matt Cain Renata Voorhees Brandon Nino Kelly Galloway Michelle Dummer Tim Bradney Dr Sue James Terry Precht DATAQ Instruments The Developers at H3DAPI org and Markus at the H3D API Forums Guido van Rossum and the Developers of Python The Developers of MySQL Mercury Systems Inc provided AMIRA Trial Program Dr Sue Kraft and Mrs Betsy Sestina of the CSU Veterinary Hospital Mr Wes Womack Jay Oaks And so many more Appendix F Project Code The current code is omitted As we are using open source software please email the SimPooch team to obtain a working version of our program Thank You
12. which is suitable for our application but could be cause problems for users unfamiliar to the device While the Phantom Omni can measure orientation and deliver force with respect to the stylus orientation of pitch yaw and roll it cannot measure orientation or supply force to the torque orientation about the stylus axis The Phantom Omni is also susceptible to mechanical failure The fall 2007 team had to order a new Phantom Omni four days before E Days due to catastrophic failure of the actuators in the device The cause of this failure was due to the unexpected reaction of the stylus to being trapped in narrow corridor in a test program with very high stiffness causing the Omni to vibrate between the walls at a very high frequency The stylus snapped before the program could be terminated A less extreme case of mechanical problems is currently occurring to our Omni random button failures We have programmed our demo to terminate when a button on the stylus is clicked However due to poor connections or some other unexplained malfunction the buttons have been known to be unresponsive to a user input Simulation Program Overview At the center of the SimPooch project is the SimPooch test program A combination of Python X3D C and the H3D API the test program controls interaction between the physical system and the virtual model Initially developed by the fall 2007 Spring 2008 team the test program has evolved since its inception Curren
13. 9 OVENI W esc ees SS oh ias 9 Model Crta iaa 10 Solid Surface Canine Model 0 2 cee ceeeecceeseeceenceceeeeeeaeeeeaaeceeeeeceaeeeeaaeeneaeeseaeeecsaeeeeaaeseeaeeceaeeeeaeeneaeeesees 10 Magnetic Spheres POPE no ance cedeaeiccacch tan EE EEE E r a Eaa 11 Spring EE 4 es 6 1 OE ld or eet ee 13 Volume Based Model serden oc scdenened tale ven eunana iea e eaae aE Eaa aeania K adek iea iR edi 14 Limitations ot Haptic DEVICE sienne a e dia 15 Simulation Progra viii A a a 15 OCW diosa ii au acbeesingd debiec AA ay deccecaep Seven ceded 15 Program Structures cias cacedelorssectics E EEE 16 GV AGING iii A A eae cae aed 18 Closest Point LO aa 18 Voronol Algorta dado dotada dolia isis 19 Closest Point Dia ias 19 Ulsa rafa ia 21 HIDE decidi ada 21 AUTO HOY iia Ad 21 WII Si ii 22 A iON seeen A O A e a ceed 22 Year COCINA i 22 Accompli AmM nt Sinse ia dicta 23 eE E S o a ed aida 23 Project Continuation c cnbe recat eres ee eh aetna ele en are ieit 23 Intro dui na 23 o te Aras beta Micseeeete E Mate eh onred te Arcos Mates E 23 Simulation PROS AM iiseczeecectavedeeteses eee ad A LA RA AAA ches ti abate ees 24 Physical SYSteMissxcc ead ee alae Heide eek A aes sel ee ree ee 24 Long Term Continuation cion eek cee ss caeees eect E Wegceeshetacesc oddhese A coedee des tee cea laced dci 24 A O RA 25 VI BIBlO EPA A es iaa 26 Mill ApPendiceS vit A aa 26 Appendix A Abbreviations ui ca dd E E 26 APPeNdIxX Be BUE ii a A EE 26 Appendix C Ph
14. GUI The GUI is a core component in the virtual simulation design When H3D API was installed it was included with a simple scene graph loader H3DLoad To improve the functionality of the program for users unfamiliar with SimPooch a more intuitive GUI needed to be developed Various scripting languages were tested AutoHotKey and wxWidgets The GUI was not improved and remains as the basic H3DLoad interface H3DLoad H3dLoad would load up an X3D world or python script in a simple window for viewing models The settings for H3DLoad were limited to selecting haptic devices stylus models and display types To improve the appearance the stylus model was improved to more closely resemble an acupuncture needle This accomplished by editing the model in an X3D editor shown in Figure 14 Figure 14 Improved stylus model H3DAPI settings common stylus x3d However H3DLoad has substantial limitations a lack of controls textual feedback in H3DLoad exe and SimPooch s program output fed to a separate window AutoHotKey AutoHotKey was researched 12 and a beta program was developed complete with an installer and graphics It improved H3DLoad exe by adding a toolbar with File Edit View and Help menus with respective submenus The top of the H3DLoad window was lined with a status bar giving the user basic instructions and progress indicators These additions can be seen in a screenshot of the program shown in Figure 15 However a fatal
15. SimPooch Canine Medical Acupuncture Training amp Simulation System May 08 2009 Team Members Sean Thomas ECE sptbear engr colostate edu Bryan Landwehr ECE blandwehr engr colostate edu Advisors Professor Peter Young ECE pmyOengr colostate edu Sue James ME sjames engr colostate edu Sponsors Narda Robinson CVMBS nrobinso lamar colostate edu Regina Schoenfeld Tacher CVMBS reginast colostate edu Colorado State University Department of Electrical amp Computer Engineering amp College of Veterinary Medicine and Biomedical Sciences Available Online http www engr colostate edu ece sr design AY08 SIMPooch Documentation SimPooch Spring 2009 Engineering Report pdf Abstract Teaching acupuncture point location requires a combination of anatomical knowledge and palpation techniques in order ensure accuracy Because live dogs may not be immediately available and in order to limit stress on dogs with repeated point locating and needling practice we are working on developing a canine acupuncture simulator that will provide visual and force feedback as well as provide a mechanism for testing and grading We are implementing this simulator using physical and virtual model s A physical mannequin exists to provide the palpatory sensation of locating known acupuncture points while a virtual model supplies the sensation of inserting the acupuncture needle and grades the student on a variet
16. antom Omni Data Sheet cccccccccccsssssececececsssesssaeeeceesseeseeaaeeeceesseesesaeaeeeesesseeaaeaeeeesens 27 Appendix D SO WTO 0 A A AA ta a T EE 27 Appendix E Acknowledgements ccccecssssececeeecessessnaeceeecscesseseaaeeeeeessessesaeaeeeeeeeseeseaaeaeeeeseeseesaaaeaeeeeeess 28 Appendix F Project Code ii a osa So 29 Table of Figures Figure 1 Physical System Mannequin on top of the stand black box Inside the stand is a rotational Noi iria vee E aos 8 Figure 2 Conceptual Tissue MOE cccccsssssceceeeceesesenaececeesceeseaaaeseeecscessesaeaecececseseseuaeaeeeeeseseeseasaeeeesens 10 Figure 3 Solid Surface Model sccccccccsssesssssceceeecseseseeaecececscsuseaaaesececssecsesaeaesececsesensaaaeeeeeseseeseasaeeeesens 11 Figure 4 Magnetic sphere model supplied haptic force VS depth cccoconococconnnnoconononannnnncnncananononnnanonnnns 12 Figure 5 Magnetic spheres model conceptual model cccconocooconccncnononononnnnnonononononononnnnnncananononnnnnnnnnns 12 Figure 6 Spring effect node conceptual force vs depth ocoocoononccccnnonocncnnononncnnononccnnononcnnnononncnnnnnnnnnnno 13 Figure 7 Spring effect node cylindrical boundary surface oococccnonccccnnononncnnononcnonononccnnnnnnnnnnonnncnnnnnnnnnnn 14 Figure 8 Spring Eltecttowch tasa Td 14 Figure 97 Programi OW iist e a 16 Figure 10 Program output with sample grading poiNtS cccconocooconcnncnononononnnnn
17. e An additional program SimPooch exe allows the rotation of the physical mannequin to be read from a magnetic encoder trough a D A converter This process involves reading the value on from the encoder and converting the voltage to a rotational value This value is then saved in a file that is opened and read in the update function of Dog py Ultimately this function will be invaluable for the synchronization of the two systems however it is not required for the virtual simulation When a user chooses to run the program without the rotation function activated movement of the physical mannequin will do nothing The scene can instead be rotated using the mouse in the H3D environment window l 4 Output File c out tat Browse gt leR s 8 041 Sample Rate 60 pastoncess 1108 Comm Port COMS h Student Point 0 067 0 013 0 0081 Event Point 50 int 0 964 0 037 0 081 577 millimeters 0 062 0 039 0 00 Teacher Point 064 0 037 08 Distance 5 828 millimeters ass o Student Point 8 015 8 014 0 101 Teacher Point 8 008 8 009 0 1051 Distance 10 026 millimeters ints 4 2 1 gra 58 8 Percent BUTTON PUSHED EXITING ke i to continue Stat J _ stop sal The grading algorithm of the SimPooch project has evolved thought the years The tasks involves comparing a set of student points to a set of teacher points and determining which student points correspond to which teacher points Previ
18. f four points displayed on the virtual dog the goal being to hit the point with the stylus The demo mode was expanded further to a contest mode In contest mode users now have the option of entering their names which are matched to scores These scores can be compared to scores of other users The results of the E Days 2009 scores were posted on the 2009 SimPooch website enclosure x3d After the user has made their selection the simulation begins with the rendering of the enclosure x3d file This file is primarily a container for the data points that define the shape of the dog head Along with the data this file contains transforms to hold each of the three rendered objects the dog the markers and the needle The data for these objects are contained in nodes called transforms this is done to allow manipulation of properties like scale and rotation without changing the viewpoint Routes exist between the dog and marker transforms to ensure that when parameters are changed for one they are updated for both An example of this is that the change in the rotation of the dog also changes the rotation of the markers so that they do not move relative to the dog Every time the scene changes it is rendered again As H3D continuously loops through dog_main x3d it encounters the python scripts that make these changes and record the user interaction Dog py The function traverseSG is called in DogStart py the main function for the simulator After ini
19. ferent layers While it is important to model other aspects of tissue this is the primary sensation of inserting a needle and therefore the most critical From the chicken demonstration we were able to determine that the general flesh model will need four layers These four layers can be thought of as variable amounts of force required to push a needle through them Figure 2 This is a conceptual description of general flesh and it will be where we begin to construct the model The layer thickness and the force required to pass a needle through them are important parameters to consider However the initial model will focus more on replicating the effect of layers and less on the physical parameters Even so there are some parameters that remain constant throughout the model In all places there will be four distinct layers The last of these layers will represent a bone layer and will be completely solid Also for the general case the force required to pass through the skin is larger than the force required to pass through the muscle and that is larger than the force required in fat Two haptic models exist in the SimPooch program the solid surface canine model and the layered sphere model Each of these represents a part of what will become a complete model Skin Bone Mussle Fat Force Depth Figure 2 Conceptual Tissue Model Solid Surface Canine Model The SimPooch project requires the ability to map physical acupuncture points into a virtua
20. flaw in the AutoHotKey scripting language was soon discovered only one instance of H3DLoad exe could be running at once This was problematic when opening simulations from the program without closing and restarting the program This approach had to be abandoned SimPooch Canine Acupuncture Simulator amp Trainer Student Test Status Stopped Menu File Users Settings View Help Figure 15 AutoHotKey docking toolbar and status window with fish demo wxWidgets While researching the H3D API 13 and the methods for handling windows with classes we discovered a simple script using wxWidgets written in C which created a H3D shell that could open and close X3D worlds The resources and information were available on the H3D org website s wiki and other documentation for wxWidgets Unfortunately the demo program and contest programs required immediate attention and the development of the GUI was put on hold for next semester s team Development of the GUI will be a long term goal for future teams and will go through many revisions as Dr Robinson and other end users give feedback how to improve the SimPooch experience IV Conclusion Year Conclusions The SimPooch project this year focused on the development of realistic haptic models for canine tissue and implementing grading on those haptic models The team successfully developed and implemented a grading algorithm that meets all the requirements While a volumetric haptic model
21. g effect node conceptual force vs depth Additional design criteria specified that the needle should not slide around laterally while inserted in the skin To address this restriction of movement to one direction it was determined that the simplest solution was to create a boundary surface beneath the skin surface that would not allow movement along any other direction In the same method we generated a thin cylinder at the same point oriented in the direction of the haptics device upon contact As long as the haptic device remains in contact with the surface the cylinder will prevent it from moving in any other direction Figure 7 A flowchart of this entire modeling process is presented in Figure 8 no 1 reate at touched point 2 kin spring spring ussle spring 3 haptic device EA orientation ae gt a S wN Xx M e sx LS Y Figure 7 Spring effect node cylindrical boundary surface yes remove cylinder and J springs i Figure 8 Spring Effect flowchart Volume Based Model Although it would be possible to model the force profile of a canine flesh with a surface based model the resulting product would have little physical meaning Flesh is not made up of springs and cylinders it is made up of different layers of volumetric material with varying properties and a better model should reflect this H3D supplies a toolkit that is designed to represent volumetric data VHTK 8 VHTK is a relatively
22. l environment and onto a virtual dog model The solution to this is the solid surface model found in enclosure x3d and shown here in Figure 3 The main requirement of the solid surface model is its shape Ultimately the physical and virtual systems will be synchronized and calibrated and this will require a virtual model that accurately represents the physical one This model will also serve as a basis for future models The model development process begins with implementing an effect on a simple geometry and then translating it to more complex shapes All models will eventually be required to take the shape of the solid surface dog model The shortcoming of this model is that it is simply one solid surface It lacks the sensation of actually inserting a needle into anything To give the effect of layers a layered surface model was created the magnetic sphere model Figure 3 Solid Surface Model Magnetic Spheres The most advanced haptic flesh model that exists in the SimPooch program is the layered magnetic sphere model The primary goal of this model is to create the sensation of pushing a needle through layers To achieve this effect it was first required to develop the effect of pushing a needle through one surface The H3D API contains pre defined force effect nodes create simple force feedback applications 6 We have used one of these pre existing force effect nodes in an attempt to simulate this sensation The magnetic surface force effec
23. ment are explained in much more detail in the spring 2009 continuation report Obviously all of these issues require knowledge and experience with the respective software packages in Appendix D Physical System Mechanical engineering students would have two important subsystems of the physical system An improved stylus and a calibration arm The stylus has several proposed upgrades Dr Young suggested a retractable needle on the tip of the stylus which would give the user the impression of inserting an acupuncture needle into the dog head Dr Robinson suggested a finger sleeve in between the buttons and tip of the stylus for the pointer finger Doing so would make the stylus lighter and could make the stylus usage more versatile for both palpation and needle insertion The second improvement would be a calibration arm attached to the base unit which would in effect lock the rotational position of the head at zero at the start of each simulation This calibration effort would be complemented in software by reading the rotation value and assigning it to zero This important upgrade would allow more realistic correlation of physical and virtual systems and improve the user experience vastly Long Term Continuation In the long run several years down the road SimPooch can take several directions The initial constraints specified by Dr Robinson was to start with the head Once that worked the remainder of the body would be developed in
24. n rendered in graphics All of the CT data and volumetric data needed to be converted to a model that could be use with the software packages The painstaking process of obtaining those points was a semester long process that the fall 2007 spring 2008 team accomplished Details of the design and procedure of the virtual system can be found in the SimPooch Fall 2007 Engineering Report 3 III Virtual Simulation Design Overview Currently the SimPooch system is powered by the H3D API 2 0 Registration on the H3D org website is required for downloading the software as well as other toolkits available H3D API is an open source haptics software development platform that uses the open standards OpenGL and X3D with haptics in one unified scene graph to handle both graphics and haptics 4 Designed for use with the Python programming language H3D API greatly simplifies development by rendering scene data provided by Python scripts in a haptic environment with minimal programming and fast computation H3D uses X3D worlds for obtaining geometric input data X3D is a XML like file format for providing point sets for describing 3D models like our virtual mannequin head Python is a high level scripting language that is used to implement the routes that H3D provides It is also useful for its extensive libraries as an object oriented programming language easily interpreting H3D objects such as nodes surfaces and forces These objects can all be accessed
25. new software tool and as of May 1 2008 less than 200 people have downloaded VHTK 4 Several attempts were made to compile the source code for the VHTK library however despite technical support from the webmasters at H3D org Forums none were effective Problems in graphic rendering and compatibility with the Phantom device are the primary reasons for the inability to compile a working version of VHTK It is unfortunate we were not able to develop a volumetric model for canine flesh with VHTK however the concept of volume haptics will be valuable to future teams in one form or another because VHTK is designed specifically to simulate the sensation of tissue 8 Another toolkit provided from H3D would have been useful as well MedX3D MedX3D is an implementation of the volumetric rendering component of H3D 9 Had we been able to compile and operate VHTK MedX3D would have allowed us to translate the files taken from the CT scan into a volumetric representation of the dog s head Without VHTK we were unable to do this Ultimately volume haptics will be used to create an accurate model for canine flesh But it should be noted that a virtual model must still work within the confines of the physical system s limitations Limitations of Haptic Device While the Phantom Omni is a powerful tool for haptic interaction it two serious limitations force feedback and mechanical failures The Phantom Omni can deliver a maximum force of 3 3 N Appendix C
26. nnnnnnononocnnnnnnnnanonennnncnnnnnnns 18 Figure 11 Voron ol diagram dada 19 Figure 12 Flowchart for grading algorithm ccccccecssssscecccecessesseaeeeeecsceeseaeaeeeeecsssesneaeaeeeeeeesseseasaeeeeeens 20 Figure 13 Data structure of grading algorithm cccssccccccecssssssceeecececsssesseaecececessesesaeaececesesseseaeaeeeesens 20 Figure 14 Improved stylus model H3DAPI settings common stylus x3d ocoooconcccconccononnnononnnnnnnnnnnorn nos 21 Figure 15 AutoHotKey docking toolbar and status window with fish d MO cccsscccccecessesssteeeeeeens 22 I Introduction Background of Acupuncture Veterinary medical acupuncture is a medical discipline that approaches the ancient system of acupuncture or needling therapy from a scientific perspective Veterinarians who practice medical acupuncture find it to be an effective method of relieving pain and speeding recovery from a variety of medical conditions For example dogs who are experiencing pain from arthritis or neurologic compromise from disk disease often respond especially well to acupuncture Veterinary medical acupuncture differs from Traditional Chinese Medical TCM style acupuncture by placing emphasizing the neuro anatomic and neurophysiologic bases of acupuncture Acupuncture points contain nerve endings that when stimulated cause the body to release endorphins and other neurotransmitters that will counter the pain and promote recovery Other physiol
27. nside the stand is a rotational sensor The mannequin model allows the student to perform palpation a technique of pressing the tissue with the tips of fingers to identify and locate acupuncture locations on the head The mannequin was developed from CT data of a Chocolate Labrador Retriever obtaining a CAD model and fabricating the head out of silicone rubber The stand supports the mannequin head and is attached to a rotation senor housed in a black box the base unit The rotation sensor reads the angle of rotation of the mannequin converts this data to a digital representation using an A D converter and sends it to the computer via a USB port Details of the design and construction of the physical system can be found in the SimPooch Spring 2007 Engineering Report 3 Virtual The fall 2007 spring 2008 team was made up of ECE students under the direction of our current advisor Dr Peter Young Much of the groundwork of our current software was conceptualized and implemented by this team The virtual system had three major subsystems acquisition of A D data conversion of CT data to enclosure x3d file H3D API The rotational data obtained from the USB port was initially read and stored by a database using MYSQL It was later determined that this approach was not fast enough to keep up with the haptic processing requirements MYSQL was replaced with a homebrew rotation acquisition program written in C in spring 2008 which improved rotatio
28. ogic changes include improved circulation and better immune function Instead of selecting acupuncture points based on invisible energy blockages as a TCM style acupuncturist would medical acupuncturists determine a needling protocol based on their knowledge of the nervous system and which neural pathways require stimulation in order to bring the body back to health This demands in depth study of anatomy and specifically the nervous system but delivers more reliable and more effective results Knowing which nerves relate to which acupuncture points and acquiring skill at localizing the points through palpation comprise important first steps to becoming a successful veterinary medical acupuncturist Dr Narda Robinson Need for Acupuncture Simulator The main problem with training veterinarians in acupuncture is the difficulty associated with hands on training Recruiting canine participants for acupuncture point palpation and needling practice becomes time consuming expensive and poses a certain amount of stress for dogs A simulator specifically designed for acupuncture would equip teachers with necessary tools in veterinary education When a student is deemed competent enough to practice on live animals any mistakes could cause harm to the dog used for training To reduce the chance of harming or causing stress to live animals students training on our simulator would become as proficient without the emotional physical or financial cost of recr
29. or the desired feel and it is faster to reproduce a working virtual model on a large scale Still the strongest argument for the use of a virtual system is that it is expandable a simple model can be created and from those more complicated effects can be programmed without the expensive and time consuming molding and casting process 2 To develop a virtual model for acupuncture we need to be familiar with the sensation of inserting a needle into flesh With the assistance of Dr Robinson we were able to perform some basic acupuncture procedures on a canine head analog From this demonstration we were able to determine the necessary parameters for a suitable virtual model Two model options were considered a surface based haptic model and a volume based haptic model Both models use the open source software from H3D and the Python programming language Model Criteria With the assistance and supervision of Dr Robinson we were able to perform acupuncture techniques on a whole store bought chicken The chicken was chosen because no canine cadaver was available for the demonstration However the chicken is a suitable replacement because it does exhibit similar effects we intend to model The current model already exhibits the primary effect of a haptic model the shape The surface model is unfortunately only skin deep and a more advanced model is required The primary effect our new haptic model will be required to implement is the sensation of dif
30. ous teams compared two possible algorithms for grading but were unable to implement an effective scheme The two methods considered were a simple closest point algorithm and a more complicated Voronoi segmentation algorithm Ultimately the problem was solved this year by a modification to the closest point algorithm The first algorithm considered by previous SimPooch teams was a simple closest point method 3 This method involves finding the closest student point to a given teacher point and then removing both the student point and teacher point from the available points list This method results in a correct correlation between the two sets on the following conditions There are exactly as many teacher points as student points and each student point is sufficiently close to its intended teacher point so that there is no confusion which point the user intended to hit While these conditions are not entirely unreasonable it is necessary to provide a more robust method of grading Consider the possibility of a student attempting to hit the same acupuncture point twice The teacher point and the closest student point will be correlated and then removed from consideration The remaining student point could be associated to another teacher point and those two removed The process of removing possibly incorrect student and teacher point pairs will propagate through the point sets and the result is only guaranteed to be one correct match Voronoi Algorithm
31. t node behaves as the name suggests when a haptic device gets within a snap distance of the magnetic surface it feels a force pulling it in the direction of surface and when it travels past the snap distance it no longer feels the force By making the snap distance small the model gives the illusion of a needle pushing through a thin dense layer The development of another layer simply involves defining another magnetic surface inside the previous one The layer that will represent muscle and bone is developed by defining a frictional surface inside the two magnetic surfaces The inner surface provides a lower force feedback over a distance before being damped to the device limit Conceptual descriptions of the haptic model and feedback it supplies are shown below in Figure 4 and Figure 5 intothe model Force out of the model Depth Figure 4 Magnetic sphere model supplied haptic force vs depth Skin Fat Muscle Bone Snap distance Figure 5 Magnetic spheres model conceptual model While the model in Figure 4 does approximate the design criteria model in Figure 2 there are some undesirable effects present The most prominent of these is a force hovering above the outer surface that pulls the haptic device towards the object Also note that while the inner surface does approach the maximum device damping it is not constant throughout the region that would represent the muscle Another model was researched for haptic
32. the same fashion This would be a viable and quick approach once the head is completed to Dr Robinson s satisfaction using the same methodology the rest of the body can be designed and implemented in a rapid prototype taking very little time Ultimately the direction of this project will heavily depend on the motivation and willingness of future teams to develop a marketable prototype which can be put into production and distributed in a cost effective fashion The cost of haptic devices like the Phantom Omni will surely decrease in the future as haptics becomes more widespread and better supported as communities grow V References 1 10 11 12 Surgery rehearsal in virtual reality freedom to learn online Available http www comp leeds ac uk matthewb research FirstYearPresentation ppt accessed December 12 2008 SimPooch spring 2007 engineering report Canine Acupuncture Simulator amp Trainer online Available http www engr colostate edu ece sr design AY08 SIMPooch Documentation SimPooch Spring 2007 Engineering Report pdfengr report 06 07 Accessed May 2 2009 SimPooch spring 2008 engineering report SimPooch Canine Medical Acupuncture Training amp Simulating System online Available http www engr colostate edu ece sr design AY08 SIMPooch Documentation SimPooch Spring 2008 Engineering Report pdf Accessed May 2 2009 H3D org Open Source Haptics online Available
33. through a Python interface and thus can be modified created or removed The fact that Python is becoming a widely supported and powerful language and that H3DAPI was designed around this framework makes it the natural choice for central development 5 All of these components of the virtual system are integrated together in a cohesive collection of programs known as SimPooch The next task was to use this powerful API to realistically simulate accurate canine acupuncture techniques Haptic Representation of Canine Flesh Overview The central aspect of haptics is the sensation of touch As demonstrated by Dr Robinson touch is the primary tool to locate the exact position of a known acupuncture point This process is called palpation By feeling the plastic canine skull beneath the silicone canine flesh of the physical model a student will be able to accurately find the locations of acupuncture points However finding the points is only half the process Once the point is found the user needs to insert the needle this is where the physical model reaches its limit Although it may have been possible to simulate the sensation of needle insertion with a physical model similar to our existing canine head mannequin previous teams chose to pursue a virtual simulation 2 3 A virtual model was chosen for a wide variety of reasons It will not wear down and become damaged with prolonged use it is simpler change parameters to tune the virtual model f
34. tialization the scene is constantly being rendered When the python script nodes are encountered H3D calls traverseSG a method that has been overridden to perform the simulation The primary function of traverseSG is to call the update function in Dog py The update function has two tasks handling the changes to the rendered scene and the identification of touched points When a touch occurs its value is saved and a timer is initialized If another touch occurs within a short time it means that the stylus is still in contact with the surface and no additional touched point is saved The event of a touched point or change in rotation causes the scene to change Using a series of helper functions the markers are added and the orientation of the scene is updated the changes appear the next time the scene is traversed When the user has selected the points and chooses to proceed to grading or saving teacher points traverseSG is suspended At this point the scene is no longer being rendered and the finish function is called Depending on the user type finish will grade or save the points and display the results Displaying the results requires one more traversal of the scene graph so traverseSG is called again and then immediately paused The user now has the option of starting over again in the same mode or exiting the program Restarting clears all markers and the arrays storing the grade and touched points and restarts rendering of the scene SimPooch ex
35. tly the test program has the ability to render the surface based haptic model of the canine head detect a collision between the haptic device and the haptic surface grade a user on their ability to touch near an acupuncture point and display markers for the teacher and student points Sample output from the current SimPooch simulation program is shown at the end of the section in Figure 10 The current form of the program exists in two windows a command window used for entering user information and displaying grade statistics and an H3D environment window where graphics are rendered and haptic interaction occurs dog main x3d DogStart py H3D File that contains Provides User Interface file linkage information Calls Dog py s update function Main Executable on each scene refresh Performs Point Set Comparison Grading amp Saving enclosure x3d Defines H3D Object and Environment viewpoint Dog py dog stylus Main class for the haptic interaction Contains an update function that is called on each scene refresh Performs point touch acquisition Performs object rotation C Datag ActiveX Control Reads the rotation from the dog head and writes it to a file Rotation File Contains current rotational value of the Physical Dog Head Figure 9 Program flow Program Structure dog_main x3d A flow diagram of the program structure is included above in Figure 9 The program is run using H3D to open the
36. uiting live dogs 1 Project Description An acupuncture simulator would provide an efficient effective and cost saving solution for the aforementioned challenges as well as providing an accurate representation of the canine anatomy and exhibiting the many behaviors tissues express Our sponsor in the College of Veterinary Medicine and Biomedical Sciences Dr Narda Robinson created SimPooch to accomplish this need The SimPooch simulator is separated into two separate systems A physical system and a virtual system In this report the virtual system will be the topic of focus as we were tasked with its improvement II Summary of Previous Work Our work on SimPooch could not have been accomplished without standing on the shoulders of past teams Development of the physical system was accomplished by fall 2006 fall 2007 teams and the virtual system by a fall 2007 spring 2008 team Brief descriptions of prior work are given below for detailed descriptions please refer to their respective engineering reports Physical Sue James coordinated the design and construction of the physical system which was completed in fall 2006 spring 2007 by a Colorado State University mechanical engineering senior design team The physical system is comprised of several subsystems a silicone rubber mannequin a stand supporting the mannequin and a rotation sensor shown in Figure 1 Figure 1 Physical System Mannequin on top of the stand black box I
37. y of parameters relevant to performing effective canine acupuncture With the use of a haptic device open source haptic software and an object oriented programming language a virtual model can be created that simulates the properties of canine skin and flesh and measures the interaction between itself and a user The ultimate goal of the SimPooch project is to develop a process to manufacture a canine acupuncture simulator which has the ability to grade an acupuncture student s performance Table of Contents l INGROAUCTION cia lt 7 Background Of ACUPUNCTULE cccscecesscesssecseseecsseeecsaecessaecsenseceeeessuecsesueceesecsaeessaaeesessecseeeessasessaaeesenteses 7 Need for Acupuncture SimulatoF cccccccnnocoooonnnnononononononononnnnnnancnnnnnncnnnnnnnrnnnnnnnnnnnnr nn nnnnnnnnnnnnnrnnnnnnnnnnnns 7 Project DESCrIPTOM sxc scccdcesacdecseanceeadedcoaseaeecda aia dd sasucdcccasetdde sadecdeasaaed dda iaa dida 7 I Summary Of Previous Work cccccccccccessssessssecececesseseaeseceeecssseeesaaaeceeecescesesaeaeeeeecuseeseaaeseeeeeceeseseaeseeeeeens 7 PHYSICA ria o ns e en y e e ed 8 A NN 8 III Virtual Simulation DeSISN cccccccccccessssesssseceeecesseseauseceeecessesesaeaeeeeecusceseaaeseeeescussesaaaeseeeeecssseseaeaeeseeens 9 OVER VIG Wi sist o es ico poi iecis adds desea aereas 9 Haptic Representation of Canine FI SN cccccccccccssssececsssececssssececsessececsssseeecsssaeeecsssaeeecsesaeescsesaeeeeseaaes
Download Pdf Manuals
Related Search