SlideShare uma empresa Scribd logo
1 de 23
From CAD to Classroom
ITEC 2015
Matt Bowden (BAE Systems)
(matt.bowden@baesystems.com)
Nick Palfrey (Real Visual)
(nick.palfrey@realvisualgroup.com)
Aim
• Look at the training aspects of the use of CAD sourced
data to create a complex 3D model of a marine platform
• Outline the lessons identified through the design and
delivery of previous 3D models
• Look at:
– How the training effectiveness of the models can be
optimised through careful task analysis and
instructional design
– Development of the training requirement set
– How a single 3D model created from CAD might meet
multiple requirements
Background
CAD-based 3D visualisation is well established in
engineering:
• Real-time interaction with the design
– Design optimisation
– Removal routes
– Human Factors
• Collaborative work environment
– Multiple design sites
– Interaction between designers & builders
– End-user involvement and familiarisation
Background
Background
• Plenty of evidence that 3D virtual environments can
effectively support skills acquisition and transfer of
knowledge, particularly for platform spatial awareness
and navigation ([1][2][3])
• Lots of work done on determining how to build the best
3D virtual environments for training ([4][5][6])
• Things are getting interesting as the boundaries between
CAD tools and gaming tools blur ([7][8])
Background
Lessons Identified
• Engineering models do not make good training models
– Engineers have different requirements, targets, skillsets and
approaches
– As shown in next image, their output has no similarity to our
target output
– They utilise collaborative environment to assist in design
communication and review; not classrooms
Lessons Identified
Too much fine detail Missing data
Good conversion
Lessons Identified
Lessons Identified
• Needs to be developed with instructors
– Lack of ownership leads to shelfware
• Engineers are not as cost constrained
– Considering physical cost savings not virtual
– IT cost constraints such as large models
– IT platform (big machines)
Lessons Identified
• Trainee needs
– Navigation
– Presentation
– Need to be engaged
– Multiple users, multiple instances
– Dispel Belief
Lessons Identified
• Known Issues
– Security restrictions
– Accuracy; fidelity and dimensional
– Multiple data sources (CAD, LIDAR, Drawings & Photographs)
– IT trends and keeping up
– Cost
– Time constraints
Lessons Identified
• The view that games shouldn't be in the classroom
• Vice Versa that this is “cutting edge” and pioneering, it’s
a good use of a growing technology
• Lack of common toolsets
– Needs support tools
– Needs hardware solutions
Training Effectiveness
• Training model needs to meet training requirements
– So clear training requirement needed up-front (training
objectives)
• Training model needs to address trainees’ needs and
situation(s)
• Early engagement between the model creators and the
training analysts/designers
• Early engagement with the training delivery team
Development of a
Requirement Set for a New Platform
Model
• Valid Training Needs Analysis
– Target Audience(s) identified
• Input standards
• Throughputs
• Training venue(s)
– Training Objectives identified
• Individual training
• Collective training
– Existing Training System defined
Development of a
Requirement Set for a New Platform
Model
• Content
– What needs to be there?
– In what form does it need to be there?
– What needs to be active/interoperable?
• Support/Maintenance Approach
– Integration with platform ILS
– Update periodicity
– Single or multiple models for a class?
Single 3D Model from CAD
• Options based on what CAD (if any is being used)
• Particular outputs are required
• Should be optimised
• Keep a common dataset for as long as possible before
creating CAD for A,B & C
• Create additional value early
Single 3D Model from CAD
• Examples of value added
– Tech Pubs
– Mission Rehearsal
– Maintenance Tasking
– Embarking & Disembarking
– Structural Repairs
– Refits
– Decommissioning
– True Through Life Support
References
[1] Dalgarno, D. & Lee, M.J.W. (2010). What are the Learning Affordances of 3-D Virtual
Environments?. British Journal of Educational Technology, Vol 41 No.1, 10-32.
[2] Mantovani, F. & Castelnuovo, G. (2003). Sense of Presence in Virtual Training: Enhancing
Skills Acquisition and Transfer of Knowledge through Learning Experience in Learning
Environments. Being There: Concepts, effects and measurement of user presence in synthetic
environments. Ios Press.
[3] Hernan, James F. & Siegel, Alexander W. (1977). The Development of Spatial
Representations of Large-Scale Environments. University of Pittsburgh.
[4] Waller, D. Knapp, D. & Hunt, E. (2001). Spatial Representations of Virtual Mazes: The Role
of Visual Fidelity and Individual Differences. Human Factors, Vol 43, No. 1,147-158.
[5] Debattista, K. Chalmers, A. Gillibrand, R. Longhurst, P. Mastoropoulou, G. & Sundstedt, V.
(2007). Parallel Selective Rendering of High-Fidelity Virtual Environments. Parallel Computing
33, 361-376.
[6] Ruddle, R. A., & Péruch, P. (2004). Effects of Proprioceptive Feedback and Environmental
Characteristics on Spatial Learning in Virtual Environments. International Journal of Human
Computer Studies, 60, 299-326.
[7] Public Works Group Blog “Are Gaming Engines the new CAD?” (2012)
http://www.publicworksgroup.com/blog/2012/02/will-gaming-engines-replace-cad/
[8] Studica Blog “Unity Game Development & Autodesk CAD = a Winning Combination” (2012).
http://www.studica.com/blog/unity-game-development-autodesk-cad-a-winning-combination
Requirement Set
• Technical
– Data sources (confirmed?)
– Fidelity levels (visual, aural, functional,
dimensional)
– Content (eg valves, eqpt, active content,
etc)
– Lighting (controllable, eg torches)
– Federatable?
– Training target audience
– Additional non-training applications/uses
• Training Environment
– Max umber of trainees using it at once
– Individual or collaborative
– Constrained or free movement (eg God’s
eye view)
– Virtual crew members (AI)
• Scenarios
– Scenario creation facilities
– Scenario storage capacity
• Trainee Interface
– Mouse keyboard or controller
– Screen size
– Level of immersion (eg HMD)
• Feedback & Reporting
– Immediate trainee feedback
– Summary trainee feedback
– Summary instructor feedback (audit trail)
• Instructor Controls
– Scenario control (initiate events, change
lighting, monitor trainee(s), etc)
– Speed up, rewind, pause scenarios
– Free movement – collision avoidance off
• System & Hardware
– Use of COTS technologies
– Access at place of work
– Permit future expansion & upgrade
– Deployable (mobile devices)
CAD Export Guide
• Check it’s 3D – Not all CAD is 3D, if it cannot be rotated around with a realistic perspective this is
likely to be 2D CAD. If it is 3D CAD, ensure the 3D data can be exported from your software.
• Export to a common 3D format – Using a common format that is readable by many pieces of off
the shelf software is best. .FBX or .OBJ are often the preferred choice.
• Keep objects separate - often during export multiple objects are combined into one large object,
try to avoid this so the different objects can be selected and edited easily once in other software.
• Preserve Meta Data - Try to preserve as much model information as possible, especially: name,
material, position centre (pivot point), ID, layers, hierarchy, groups. This will aid scene
management. Real world engineering or mechanical data such as weight, capacity, and strength
are not important for 3d engines.
• Optimise - Real time graphics rely on processing the polygons fast in order to achieve 30 renders
per second. This means the models need to be optimised as possible. An average real time scene
would ideally not go over 100,000 triangles. If your CAD software has any mesh resolution,
density, or division settings, reduce these to as low as possible.
• Level of detail – as an expansion on the last point if control over the mesh density can be
maintained during export, this is better, as the engine can be configured to automatically change
the detail based on the camera distance.
• Textured information- Generally any details smaller than a few inches will be added into a
texture rather than be included in the mesh (this is to optimise the performance of the application).
Ideally, create two exports for the same objects – one at full resolution with all the mesh details
(polygon count does not matter for this one) and another low polygon count model with the same
silhouette. The information from the high can then be baked into the texture for the low version.
CAD Export Guide
• Consistent topology - As another expansion on the optimise point, ensure the mesh density is
consistent. Ideally models should be made of even sized quad (4 sided) polygons. With more
polygons for curved surfaces, and less for flat areas.
• Scale – ensure that your export and import maintains a consistent unit setup. Different software
can use different units of measurements. As a check, insert a reference object, such as a cube
that you know to be a standard scale.
• Orientation and position - For single objects, ensure models are positioned and rotated to 0,0,0
in 3D space. Or with a scene with multiple objects, ensure that their position and rotation in 3D
space is maintained during export.
• Non solid models - Real time 3D models are not solids – ie the inside of the shape you are
seeing is empty. It is worth considering that the model you are exporting will always be
represented as a shell rather than a solid that can be sliced through.
• Backface and double sided polygons - CAD software tends to use double sided polygons,
whereas most game engines (such as unity 3D) do not by default. This can result in “invisible”
areas of a model. If possible, view your model with backface culling on (or render 2 sided off) to
see how it could look in engine.
• Double sided Export - Similarly, CAD can be known to export 2 polygons on top of each other
(one for each side) for each single polygon. Try to disable this option, it will result in artifacts and
wasted polygons. It is possible to render single sided polygons as double sided in engine- this is a
more efficient way to do it.
• Only include what you will see – It is not necessary to include parts of the model that will not be
seen. For example if you want to show a car bodywork, there’s no need to include the engine.
Any Questions ??

Mais conteúdo relacionado

Destaque

T-cure final executive summary official
T-cure final executive summary officialT-cure final executive summary official
T-cure final executive summary officialMarwan Rashed
 
Places to visit in kailash mansarovar
Places to visit in kailash mansarovarPlaces to visit in kailash mansarovar
Places to visit in kailash mansarovarTHE BRAVEHEARTZ
 
Promocion Food Service verano 2016 [Hostelería]
Promocion Food Service verano 2016 [Hostelería]Promocion Food Service verano 2016 [Hostelería]
Promocion Food Service verano 2016 [Hostelería]Arc Ibérica
 
Zenix: El material más resistente para hostelería
Zenix: El material más resistente para hosteleríaZenix: El material más resistente para hostelería
Zenix: El material más resistente para hosteleríaArc Ibérica
 
¿Por qué elegir cristalería Chef & Sommelier?
¿Por qué elegir cristalería Chef & Sommelier?¿Por qué elegir cristalería Chef & Sommelier?
¿Por qué elegir cristalería Chef & Sommelier?Arc Ibérica
 
Lanzamiento copa Party Arcoroc
Lanzamiento copa Party ArcorocLanzamiento copa Party Arcoroc
Lanzamiento copa Party ArcorocArc Ibérica
 

Destaque (9)

T-cure final executive summary official
T-cure final executive summary officialT-cure final executive summary official
T-cure final executive summary official
 
Places to visit in kailash mansarovar
Places to visit in kailash mansarovarPlaces to visit in kailash mansarovar
Places to visit in kailash mansarovar
 
Kailas mansoravar
Kailas mansoravarKailas mansoravar
Kailas mansoravar
 
Fisiologia del estres
Fisiologia del estresFisiologia del estres
Fisiologia del estres
 
Promocion Food Service verano 2016 [Hostelería]
Promocion Food Service verano 2016 [Hostelería]Promocion Food Service verano 2016 [Hostelería]
Promocion Food Service verano 2016 [Hostelería]
 
Zenix: El material más resistente para hostelería
Zenix: El material más resistente para hosteleríaZenix: El material más resistente para hostelería
Zenix: El material más resistente para hostelería
 
¿Por qué elegir cristalería Chef & Sommelier?
¿Por qué elegir cristalería Chef & Sommelier?¿Por qué elegir cristalería Chef & Sommelier?
¿Por qué elegir cristalería Chef & Sommelier?
 
Lanzamiento copa Party Arcoroc
Lanzamiento copa Party ArcorocLanzamiento copa Party Arcoroc
Lanzamiento copa Party Arcoroc
 
CYCLONE
CYCLONECYCLONE
CYCLONE
 

Semelhante a From CAD to Classroom Final 17 Apr 15

2013 Lecture 5: AR Tools and Interaction
2013 Lecture 5: AR Tools and Interaction 2013 Lecture 5: AR Tools and Interaction
2013 Lecture 5: AR Tools and Interaction Mark Billinghurst
 
PLCM Module -5-Dr.GMS JSSATEB.pdf
PLCM Module -5-Dr.GMS JSSATEB.pdfPLCM Module -5-Dr.GMS JSSATEB.pdf
PLCM Module -5-Dr.GMS JSSATEB.pdfswamy62
 
Desktop Softwares for Unmanned Aerial Systems(UAS))
Desktop Softwares for Unmanned Aerial Systems(UAS))Desktop Softwares for Unmanned Aerial Systems(UAS))
Desktop Softwares for Unmanned Aerial Systems(UAS))Kamal Shahi
 
School of Computing, Science & EngineeringAssessment Briefin.docx
School of Computing, Science & EngineeringAssessment Briefin.docxSchool of Computing, Science & EngineeringAssessment Briefin.docx
School of Computing, Science & EngineeringAssessment Briefin.docxanhlodge
 
A discrete Event Simulation Model of Asphalt Paving Operations, Ramzi Labban ...
A discrete Event Simulation Model of Asphalt Paving Operations, Ramzi Labban ...A discrete Event Simulation Model of Asphalt Paving Operations, Ramzi Labban ...
A discrete Event Simulation Model of Asphalt Paving Operations, Ramzi Labban ...CCT International
 
Scaling & Transforming Stitch Fix's Visibility into What Folks will love
Scaling & Transforming Stitch Fix's Visibility into What Folks will loveScaling & Transforming Stitch Fix's Visibility into What Folks will love
Scaling & Transforming Stitch Fix's Visibility into What Folks will loveJune Andrews
 
Understanding and Addressing Architectural Challenges of Cloud- Based Systems
Understanding and Addressing Architectural Challenges of Cloud- Based SystemsUnderstanding and Addressing Architectural Challenges of Cloud- Based Systems
Understanding and Addressing Architectural Challenges of Cloud- Based SystemsCREST @ University of Adelaide
 
DSD-INT 2014 - OpenMI Symposium - Federated modelling of Critical Infrastruct...
DSD-INT 2014 - OpenMI Symposium - Federated modelling of Critical Infrastruct...DSD-INT 2014 - OpenMI Symposium - Federated modelling of Critical Infrastruct...
DSD-INT 2014 - OpenMI Symposium - Federated modelling of Critical Infrastruct...Deltares
 
Ectel nods v2
Ectel nods v2Ectel nods v2
Ectel nods v2nodenot
 
MODEL-DRIVEN ENGINEERING (MDE) in Practice
MODEL-DRIVEN ENGINEERING (MDE) in PracticeMODEL-DRIVEN ENGINEERING (MDE) in Practice
MODEL-DRIVEN ENGINEERING (MDE) in PracticeHussein Alshkhir
 
Design and implement a reality-based 3D digitisation and modelling project
Design and implement a reality-based 3D digitisation and modelling projectDesign and implement a reality-based 3D digitisation and modelling project
Design and implement a reality-based 3D digitisation and modelling project3D ICONS Project
 
Introduction_to_Auto_CAD_and_Civil_3D.pptx
Introduction_to_Auto_CAD_and_Civil_3D.pptxIntroduction_to_Auto_CAD_and_Civil_3D.pptx
Introduction_to_Auto_CAD_and_Civil_3D.pptxRaviShenker1
 
Unit 67 3_d_animation
Unit 67 3_d_animationUnit 67 3_d_animation
Unit 67 3_d_animationDamionVize
 

Semelhante a From CAD to Classroom Final 17 Apr 15 (20)

2013 Lecture 5: AR Tools and Interaction
2013 Lecture 5: AR Tools and Interaction 2013 Lecture 5: AR Tools and Interaction
2013 Lecture 5: AR Tools and Interaction
 
PLCM Module -5-Dr.GMS JSSATEB.pdf
PLCM Module -5-Dr.GMS JSSATEB.pdfPLCM Module -5-Dr.GMS JSSATEB.pdf
PLCM Module -5-Dr.GMS JSSATEB.pdf
 
Desktop Softwares for Unmanned Aerial Systems(UAS))
Desktop Softwares for Unmanned Aerial Systems(UAS))Desktop Softwares for Unmanned Aerial Systems(UAS))
Desktop Softwares for Unmanned Aerial Systems(UAS))
 
School of Computing, Science & EngineeringAssessment Briefin.docx
School of Computing, Science & EngineeringAssessment Briefin.docxSchool of Computing, Science & EngineeringAssessment Briefin.docx
School of Computing, Science & EngineeringAssessment Briefin.docx
 
A discrete Event Simulation Model of Asphalt Paving Operations, Ramzi Labban ...
A discrete Event Simulation Model of Asphalt Paving Operations, Ramzi Labban ...A discrete Event Simulation Model of Asphalt Paving Operations, Ramzi Labban ...
A discrete Event Simulation Model of Asphalt Paving Operations, Ramzi Labban ...
 
infiDOF - An Overview
infiDOF - An Overview infiDOF - An Overview
infiDOF - An Overview
 
Solid Works
Solid WorksSolid Works
Solid Works
 
NEXiDA at OMG June 2009
NEXiDA at OMG June 2009NEXiDA at OMG June 2009
NEXiDA at OMG June 2009
 
Lamar 3d Map
Lamar 3d Map Lamar 3d Map
Lamar 3d Map
 
Scaling & Transforming Stitch Fix's Visibility into What Folks will love
Scaling & Transforming Stitch Fix's Visibility into What Folks will loveScaling & Transforming Stitch Fix's Visibility into What Folks will love
Scaling & Transforming Stitch Fix's Visibility into What Folks will love
 
Understanding and Addressing Architectural Challenges of Cloud- Based Systems
Understanding and Addressing Architectural Challenges of Cloud- Based SystemsUnderstanding and Addressing Architectural Challenges of Cloud- Based Systems
Understanding and Addressing Architectural Challenges of Cloud- Based Systems
 
DE PPT.pptx
DE PPT.pptxDE PPT.pptx
DE PPT.pptx
 
DSD-INT 2014 - OpenMI Symposium - Federated modelling of Critical Infrastruct...
DSD-INT 2014 - OpenMI Symposium - Federated modelling of Critical Infrastruct...DSD-INT 2014 - OpenMI Symposium - Federated modelling of Critical Infrastruct...
DSD-INT 2014 - OpenMI Symposium - Federated modelling of Critical Infrastruct...
 
Ectel nods v2
Ectel nods v2Ectel nods v2
Ectel nods v2
 
MODEL-DRIVEN ENGINEERING (MDE) in Practice
MODEL-DRIVEN ENGINEERING (MDE) in PracticeMODEL-DRIVEN ENGINEERING (MDE) in Practice
MODEL-DRIVEN ENGINEERING (MDE) in Practice
 
Design and implement a reality-based 3D digitisation and modelling project
Design and implement a reality-based 3D digitisation and modelling projectDesign and implement a reality-based 3D digitisation and modelling project
Design and implement a reality-based 3D digitisation and modelling project
 
Wims2012
Wims2012Wims2012
Wims2012
 
Introduction_to_Auto_CAD_and_Civil_3D.pptx
Introduction_to_Auto_CAD_and_Civil_3D.pptxIntroduction_to_Auto_CAD_and_Civil_3D.pptx
Introduction_to_Auto_CAD_and_Civil_3D.pptx
 
Unit 67 3_d_animation
Unit 67 3_d_animationUnit 67 3_d_animation
Unit 67 3_d_animation
 
Solidworks software
Solidworks softwareSolidworks software
Solidworks software
 

From CAD to Classroom Final 17 Apr 15

  • 1. From CAD to Classroom ITEC 2015 Matt Bowden (BAE Systems) (matt.bowden@baesystems.com) Nick Palfrey (Real Visual) (nick.palfrey@realvisualgroup.com)
  • 2. Aim • Look at the training aspects of the use of CAD sourced data to create a complex 3D model of a marine platform • Outline the lessons identified through the design and delivery of previous 3D models • Look at: – How the training effectiveness of the models can be optimised through careful task analysis and instructional design – Development of the training requirement set – How a single 3D model created from CAD might meet multiple requirements
  • 3. Background CAD-based 3D visualisation is well established in engineering: • Real-time interaction with the design – Design optimisation – Removal routes – Human Factors • Collaborative work environment – Multiple design sites – Interaction between designers & builders – End-user involvement and familiarisation
  • 5. Background • Plenty of evidence that 3D virtual environments can effectively support skills acquisition and transfer of knowledge, particularly for platform spatial awareness and navigation ([1][2][3]) • Lots of work done on determining how to build the best 3D virtual environments for training ([4][5][6]) • Things are getting interesting as the boundaries between CAD tools and gaming tools blur ([7][8])
  • 7. Lessons Identified • Engineering models do not make good training models – Engineers have different requirements, targets, skillsets and approaches – As shown in next image, their output has no similarity to our target output – They utilise collaborative environment to assist in design communication and review; not classrooms
  • 8. Lessons Identified Too much fine detail Missing data Good conversion
  • 10. Lessons Identified • Needs to be developed with instructors – Lack of ownership leads to shelfware • Engineers are not as cost constrained – Considering physical cost savings not virtual – IT cost constraints such as large models – IT platform (big machines)
  • 11. Lessons Identified • Trainee needs – Navigation – Presentation – Need to be engaged – Multiple users, multiple instances – Dispel Belief
  • 12. Lessons Identified • Known Issues – Security restrictions – Accuracy; fidelity and dimensional – Multiple data sources (CAD, LIDAR, Drawings & Photographs) – IT trends and keeping up – Cost – Time constraints
  • 13. Lessons Identified • The view that games shouldn't be in the classroom • Vice Versa that this is “cutting edge” and pioneering, it’s a good use of a growing technology • Lack of common toolsets – Needs support tools – Needs hardware solutions
  • 14. Training Effectiveness • Training model needs to meet training requirements – So clear training requirement needed up-front (training objectives) • Training model needs to address trainees’ needs and situation(s) • Early engagement between the model creators and the training analysts/designers • Early engagement with the training delivery team
  • 15. Development of a Requirement Set for a New Platform Model • Valid Training Needs Analysis – Target Audience(s) identified • Input standards • Throughputs • Training venue(s) – Training Objectives identified • Individual training • Collective training – Existing Training System defined
  • 16. Development of a Requirement Set for a New Platform Model • Content – What needs to be there? – In what form does it need to be there? – What needs to be active/interoperable? • Support/Maintenance Approach – Integration with platform ILS – Update periodicity – Single or multiple models for a class?
  • 17. Single 3D Model from CAD • Options based on what CAD (if any is being used) • Particular outputs are required • Should be optimised • Keep a common dataset for as long as possible before creating CAD for A,B & C • Create additional value early
  • 18. Single 3D Model from CAD • Examples of value added – Tech Pubs – Mission Rehearsal – Maintenance Tasking – Embarking & Disembarking – Structural Repairs – Refits – Decommissioning – True Through Life Support
  • 19. References [1] Dalgarno, D. & Lee, M.J.W. (2010). What are the Learning Affordances of 3-D Virtual Environments?. British Journal of Educational Technology, Vol 41 No.1, 10-32. [2] Mantovani, F. & Castelnuovo, G. (2003). Sense of Presence in Virtual Training: Enhancing Skills Acquisition and Transfer of Knowledge through Learning Experience in Learning Environments. Being There: Concepts, effects and measurement of user presence in synthetic environments. Ios Press. [3] Hernan, James F. & Siegel, Alexander W. (1977). The Development of Spatial Representations of Large-Scale Environments. University of Pittsburgh. [4] Waller, D. Knapp, D. & Hunt, E. (2001). Spatial Representations of Virtual Mazes: The Role of Visual Fidelity and Individual Differences. Human Factors, Vol 43, No. 1,147-158. [5] Debattista, K. Chalmers, A. Gillibrand, R. Longhurst, P. Mastoropoulou, G. & Sundstedt, V. (2007). Parallel Selective Rendering of High-Fidelity Virtual Environments. Parallel Computing 33, 361-376. [6] Ruddle, R. A., & Péruch, P. (2004). Effects of Proprioceptive Feedback and Environmental Characteristics on Spatial Learning in Virtual Environments. International Journal of Human Computer Studies, 60, 299-326. [7] Public Works Group Blog “Are Gaming Engines the new CAD?” (2012) http://www.publicworksgroup.com/blog/2012/02/will-gaming-engines-replace-cad/ [8] Studica Blog “Unity Game Development & Autodesk CAD = a Winning Combination” (2012). http://www.studica.com/blog/unity-game-development-autodesk-cad-a-winning-combination
  • 20. Requirement Set • Technical – Data sources (confirmed?) – Fidelity levels (visual, aural, functional, dimensional) – Content (eg valves, eqpt, active content, etc) – Lighting (controllable, eg torches) – Federatable? – Training target audience – Additional non-training applications/uses • Training Environment – Max umber of trainees using it at once – Individual or collaborative – Constrained or free movement (eg God’s eye view) – Virtual crew members (AI) • Scenarios – Scenario creation facilities – Scenario storage capacity • Trainee Interface – Mouse keyboard or controller – Screen size – Level of immersion (eg HMD) • Feedback & Reporting – Immediate trainee feedback – Summary trainee feedback – Summary instructor feedback (audit trail) • Instructor Controls – Scenario control (initiate events, change lighting, monitor trainee(s), etc) – Speed up, rewind, pause scenarios – Free movement – collision avoidance off • System & Hardware – Use of COTS technologies – Access at place of work – Permit future expansion & upgrade – Deployable (mobile devices)
  • 21. CAD Export Guide • Check it’s 3D – Not all CAD is 3D, if it cannot be rotated around with a realistic perspective this is likely to be 2D CAD. If it is 3D CAD, ensure the 3D data can be exported from your software. • Export to a common 3D format – Using a common format that is readable by many pieces of off the shelf software is best. .FBX or .OBJ are often the preferred choice. • Keep objects separate - often during export multiple objects are combined into one large object, try to avoid this so the different objects can be selected and edited easily once in other software. • Preserve Meta Data - Try to preserve as much model information as possible, especially: name, material, position centre (pivot point), ID, layers, hierarchy, groups. This will aid scene management. Real world engineering or mechanical data such as weight, capacity, and strength are not important for 3d engines. • Optimise - Real time graphics rely on processing the polygons fast in order to achieve 30 renders per second. This means the models need to be optimised as possible. An average real time scene would ideally not go over 100,000 triangles. If your CAD software has any mesh resolution, density, or division settings, reduce these to as low as possible. • Level of detail – as an expansion on the last point if control over the mesh density can be maintained during export, this is better, as the engine can be configured to automatically change the detail based on the camera distance. • Textured information- Generally any details smaller than a few inches will be added into a texture rather than be included in the mesh (this is to optimise the performance of the application). Ideally, create two exports for the same objects – one at full resolution with all the mesh details (polygon count does not matter for this one) and another low polygon count model with the same silhouette. The information from the high can then be baked into the texture for the low version.
  • 22. CAD Export Guide • Consistent topology - As another expansion on the optimise point, ensure the mesh density is consistent. Ideally models should be made of even sized quad (4 sided) polygons. With more polygons for curved surfaces, and less for flat areas. • Scale – ensure that your export and import maintains a consistent unit setup. Different software can use different units of measurements. As a check, insert a reference object, such as a cube that you know to be a standard scale. • Orientation and position - For single objects, ensure models are positioned and rotated to 0,0,0 in 3D space. Or with a scene with multiple objects, ensure that their position and rotation in 3D space is maintained during export. • Non solid models - Real time 3D models are not solids – ie the inside of the shape you are seeing is empty. It is worth considering that the model you are exporting will always be represented as a shell rather than a solid that can be sliced through. • Backface and double sided polygons - CAD software tends to use double sided polygons, whereas most game engines (such as unity 3D) do not by default. This can result in “invisible” areas of a model. If possible, view your model with backface culling on (or render 2 sided off) to see how it could look in engine. • Double sided Export - Similarly, CAD can be known to export 2 polygons on top of each other (one for each side) for each single polygon. Try to disable this option, it will result in artifacts and wasted polygons. It is possible to render single sided polygons as double sided in engine- this is a more efficient way to do it. • Only include what you will see – It is not necessary to include parts of the model that will not be seen. For example if you want to show a car bodywork, there’s no need to include the engine.

Notas do Editor

  1. Good afternoon ladies & gentlemen. I’m Matt Bowden, and with my colleague Nick Palfrey I’ll be looking at the training aspects of the use of CAD sourced data to create a complex 3D model of a marine platform – a ship – for use within a training solution. [click] We’ll be looking at the lessons we’ve encountered through the recent design and delivery of a couple of 3D models for training, and then highlighting how we believe that the training effectiveness of such models can be optimised through careful task analysis and instructional design, supported by a robust training requirement set, and finally we’ll focus in on how a single 3D model created from CAD data might be used to meet multiple requirements. [click – next slide]
  2. CAD-based 3D visualisation is a tool that’s now well established in engineering, used across the range of projects from designing large chemical facilities to deciding what colour leather you want in your new Ferrari. The benefits that such toolsets bring – [click] here are a few – almost all have a training aspect associated with them. So we’ve been looking at how we can minimise or remove the gap between 3D CAD models for engineering visualisation, and 3D virtual models for training purposes, using the software toolsets available on the market today. [click - next slide]
  3. Our objective is to determine how we can use a single source of engineering CAD data to create engineering and training solutions, with minimum effort and cost for maximum effectiveness. We’re not quite there yet, but we’ve found some things out on the way which are worth sharing. [click – next slide]
  4. From a training viewpoint there is plenty of academic and practical evidence to show that 3D virtual environments can effectively support skills acquisition and transfer of knowledge, particularly for spatial awareness and navigation skills. This is of particular interest when dealing with environments that are relatively high risk and difficult to navigate, such as naval vessels, and also when dealing with people who need to work on them but who aren’t necessarily familiar with them, such as sub-contracted support personnel. [click] Lots of work has also been done on determining how to build the best 3D virtual environments for training, specifically in terms of content, interface and visual fidelity, but we’ve found that other constraints, specifically budgets and time, mean that managing priorities and getting the best ‘bang for buck’ becomes the order of the day. [click] And things are becoming ever more interesting, challenging and exciting as the boundaries between high-end technical CAD tools and gaming engines blur, with architects now often exporting CAD designs into gaming engines to bring their designs to life. I’ve cross-referenced some papers of interest here – there’s a full reference list on a slide at the end. [click - next slide]
  5. But we want to look at it from a practical point of view, looking at some real examples and some aspects not necessarily covered in the previous studies, specifically from recent experience developing 3D virtual environments for training for naval vessels. So I’ll pass you over to Nick to talk about some of the issues we’ve come across when developing 3D virtual models of naval platforms.
  6. Good afternoon Matt has already begun setting the scene in regards to the constraints we face when producing 3D walkthroughs from CAD sources. The same logic would in fact apply to any training, that’s an important distinction for me, as people see large environments and walkthroughs as being complicated, un achievable or worse computer games. Looking at the lessons we identified when building all three of the British submarine 3D Walkthroughs, of which most are entirely focussed on use of source data. Strait off the bat, with bold statement, it’s a fact engineering models do not make good training models. Most engineers have specialist skillsets, marine, structural etc which does not have a clear connection with skills that we use when producing our outputs. Their requirements for their output is different, what they need to communicate and show people. Fundamentally their approach and methodology is based on entirely different principles and requirements. Their day to day output has almost no similarity with ours.
  7. Not all 3D is created equal… To summarise, we require triangulated data, the scene is based on triangles to make shapes and patterns. CAD is based on parametric data and can produce a whole lot of triangles when converted, not always optimised for real time rendering at 30 or 60 frames per second. Good Cad Picture -  The CAD Mesh is visibly sound, but over triangulated when considering constraints that apply to real time rendering and can easily start to eat into memory.  BAD CAD The CAD can sometimes suffer from incorrectly welded vertices, which can result in the mesh having flipped faces, missing triangles, distorted normals and tangents (too much) resulting in a visibly incorrect object.  Game Ready Watertight mesh, consisting of good topology, correct silhouetting of the object with visibly appealing shading of the object.  A priority for an engineer might be a design reviews, functional elements of a ship for instance. Speed, performance, running costs. This is before we consider the impact that weaponising such a platform would have on engineers. They need to communicate design to multiple stakeholders with a high emphasis on cost, and rightly so but it doesn’t match with us.
  8. This diagram demonstrates a workflow we use to move from source data to our final output. It lists the softwares involved (go autodesk) and how we manage the flow of information. Towards the end of the slides, we have another diagram and chart showing settings we use to export data, this will be useful for everyone here in the future. This will most likely be relevant across most CAD, BIM or engineering packages, we can come back to that later.
  9. We talked a little bit about engineers, who got a tough rap, in terms of their work and how we can use it but the fact is we need them and look for opportunities to enhance relationships. We also need instructors, from day 1. We have been lucky in that we have never created shelf ware but no developer wants to see their hard work mothballed. All the instructors I have had the pleasure of working with see the value that our work can bring, for them. They want to take ownership of the product, its functionality, its usability and once finished, they feel as though its theirs. Matt will talk some more about this relationship later. For me, this leads us back to the differences they have with engineers. Who are not as cost constrained. Any engineers who have sat quietly so far may be ready to start throwing things now but hear me out. When making this comparison, an engineer would be considering the physical cost saving of real materials, not virtual efficiency. We rely on optimising large data, they neither have this luxury, nor do they need it because they use powerful and specially designed computers. We use COTS computers with cost efficiency in mind.
  10. Evaluating the final user group we work with but have little contact with, is the trainee. We talked about engineers and instructors, who help create the output for the end user, the student, so its worth considering their needs too. In all instance where our work has been used, navigation is a high priority, if not the highest. They want a presentation tool, that isn't death by PowerPoint. Users have to be engaged, a combination of self paced learning that can also be instructor led. Multiplayer is a big buzz word, with many of the platforms you see here at this event offering mission rehearsal, multiple instances and networked infrastructure. That’s all well and good in a sandbox environment where your using standard terrains, models and other functionality to program the experience. However, considering we do our work from scratch, many of you do too, we need to build networking functionality into the experience and it has to have a purpose. I would also prioritise the need to dispel belief that this type of training is games based, serious games anything to do with games in a negative sense. It can often be the elephant in the room. Maybe we use a games engine, but we use a games engine to produce effective training which requires the same consideration as any other training.
  11. All of this is well and good but what about outside issues? Security for instance and the detail we have, presents a security risk. The need to understand fidelity, whether this means super shiny and millimetre accuracy. The issues of using multiple data sources; CAD, LIDAR, engineers drawing and photographs. This is not optimum, yet leads to cutting edge training… Keeping up with IT trends, what happens 2 years down the line when you want to make a minor change, that requires an entire rebuild. I think most people know that we use unity and we are good at using it, but that’s only because we stay at the forefront of how it is used. Beyond defence, we do work in other industries such as healthcare and even games consultancy. This is important because they move faster than the defence industry allowing us to utilise this to our clients advantage. Cost, considering the above, people want the perfect blend of cheap, fast and high quality.. We all know that’s somewhat hard… Finally time constraints. I haven’t seen a project such as this that hasn’t taken years to commission and is then needed in months.
  12. Games shouldn’t be in the classroom… a myth and why not? I heard an australian rear admiral ask once why you could buy modern warfare on the shelf in a supermarket for 60 bucks but whenever he wanted similar training it cost millions. Is this resolved, do we know better or is there still a sense of mystery surrounding what we do? It the latest technology, sure… it looks pretty, sure… but its actually quite simple, we are just really good at it. Vice versa, I’ve heard sales teams talk about leading the trends and military personal echoing this. Having a genuine belief that they are leading the way. They are not, we are not… We use a growing technology, well and that’s just our business which has all the constraints of being small. Do I see this as a wider trend? Yes… Not using one platform, all standalone, not supported, no similar requirement. Im talking just software, Matt and all of you know about the whole training package. Prevents upgrading… I’m dead against only using one platform and paying shed loads of public money for a government license… What if needs change? How can we as providers, wherevever they are in the room, worktogether to do this… Where does hardware fit into this…
  13. There’s no rocket science associated with this next bit. It all sounds obvious, but I know of at least 2 virtual models that were provided as training solutions but which have ended up as shelfware. In their own right they looked great, but for a number of reasons they failed to become effective, integrated, useable training solutions. [click] The first, and most important point is that the training model needs to meet a set of defined and agreed training requirements. The solution needs to be driven by the training requirement, not the other way around. Clearly the functionality and fidelity of the model needs to be driven by the requirement, to ensure that the money is spent where it needs to be spent, rather than on things that look shiny but add no training value. [click] Secondly, the model needs to address the trainees’ needs and situation. This needs to include the hardware and infrastructure on which the model will sit, in the training environment. When we’ve been analysing the training target audiences for our models it’s quickly become apparent that we’re looking at a significant range of users, from trainees who will be undertaking the training in a proper learning environment, as a class and with an instructor, to people who will need to undertake the training at their place of work, preferably at their desks, without supervision. We’re looking at young and older people, from all backgrounds, who all need to be able to do the same thing – navigate their way around a naval vessel safely, and confidently, and deal correctly with any incidents that they may encounter on the way. It’s the new gaming engines that have opened this area up, removing the need for high-end number-crunching machines, and making the models available almost anywhere, using pretty ordinary hardware. We’re currently looking at a 3D virtual model for the new Queen Elizabeth class carrier, where our target audience ranges from cleaners, through sub-contractors who might only be on-board for a day, up to the support engineers who will be looking after the ship through-life, each with very different training needs but the same general training outcome. It might be a little more difficult to make training fun for everyone, but we’re going to give it a go …… [click] Thirdly, I’d strongly recommend early and continuous engagement between the model creators and the training analysts and instructional designers. For one model we (and in this case ‘we’ includes Babcock, who are represented in the audience) ended up with the model creators working in the same room as the instructional designers, and the continuous interactions between the two resulted in a better model and better training solution as each team better understood the needs and capabilities of the other. We also optimised re-use of the model as we used video and screenshots from it in the instructor’s Powerpoint presentations. [click] Lastly, I’d strongly recommend early engagement with the training delivery team. Nick has already covered how a lack of ownership by the instructors is likely to lead to any model becoming shelfware. This is particularly important because the model is almost certainly only going to be an element within a larger training solution. If the element isn’t designed to fit properly into the bigger machine, which is in turn applied by the instructors, then the machine won’t work effectively, or at all. If the instructor isn’t happy or confident with the model and its capabilities, then its training effectiveness will be compromised. [click – next slide]
  14. So – some things that we think are important to include within the requirement for any new 3D virtual model for training: Firstly some obvious but sometimes neglected ones [click]: Is it supported by a valid and stakeholder agreed TNA? And by TNA I mean: [click] Have the target audience or audiences been identified and agreed with the stakeholders, along with their competence levels at the start of the training, their throughputs and indication as to exactly where the training needs to be delivered – ashore, at sea, on-board but alongside, and so on. [click] Have the training objectives been clearly identified and agreed with the stakeholders? Clearly these are the most important elements of any requirement, so time spent working these through to as fine a level of detail as possible will reap returns. In particular the objectives need to be written in terms that everyone involved can understand, not just the trainers. It’s also necessary to consider both the individual and collective training requirements. Gaming engines enable the creation of models and environments that support collaborative and collective learning, with constraints. A full bridge team could be assembled, for example, with each team member taking part on his own device, and at his own location to enable full team training to take place – a port entry for example - but this requirement needs to be clearly defined and articulated at the outset in order to ensure that it can be met, and that it doesn’t blow the budget if it has to be added later on. [click] And has the existing training system been defined and explained? It’s fair to assume that any 3D model-based training solution won’t be running in a vacuum – it’ll need to integrate into existing training pipelines, with existing training solutions and existing training hardware and infrastructure. Being able to reduce costs by optimising synergies with any existing system will help to justify the use of a model in the first place. [click – next slide]
  15. The cost of building these models is effectively linear, ie the more compartments the higher the cost. The cost of a model for a small frigate or a submarine with perhaps 150-200 compartments is generally palatable to the bean counters. Work on a large platform such as a carrier, with upwards of 3000 compartment, and a price 20 times larger than that for a frigate becomes harder to justify – even if the requirement is all the more important and complex (imagine the impact of getting someone lost on a carrier ….). So, the larger the model the more important it becomes that it only includes what it needs to include. [Click] Working with the training analysts and the instructors is the only practical way to do this properly. For the models we’ve built we’ve inevitably then had to go back to the technical requirements to revise them on cost grounds, engaging compromises, to come up with the best solution possible for the given budget. For platform models there’s almost always the option of including time on-board the real platform as the means of ‘topping up’ the learning and experience in order to achieve the desired competence and confidence levels, so we haven’t had to achieve 100% of the learning transfer using just the model. This probably ended up being the main compromise we had to work with. [click] Another issue we encountered was that the training solutions were procured as ‘add-ons’ to a separate support solution rather than as an integrated part of the original integrated platform support solution. There are significant benefits to be gained from linking the training solution with the platform ILS solution from the outset, particularly from the use of common datasets (eg the CAD data). Naval platforms undergo refits and updates on a continuous and regular basis. Add in multiple vessels within a single class, and you quickly end up with a significant number of differences across the fleet at any one time. Clearly it would be best if individuals could be trained on a model that exactly matches the vessel they’ll be working on, but cost and time constraints often preclude having more than one version of a model. So the link between and approach for CAD updates and training model updates needs to be agreed up front. I don’t think that this will be such an issue in 5 or 10 years time, as the boundary between CAD and gaming toolsets disappears, but it is something that we’re having to consider now. We’ve summarised most of these on a slide at the end, which we’ll throw up during the Q&A bit. And, on that note, I’ll hand back to Nick to talk us through one aspect of this dilemma.
  16. I felt like the naysayer earlier, so now its time for me to offer some solutions. Getting model data from CAD is possible, but clearly we are basing this on an assumption of what data is being used. Tribon, Foran, Autodesk suites and other industry specific CAD packages all have differences but many common elements. At the end of this pack is a sheet of priority settings, receivable files and how to get them. Obviously, we cant do a manual for each software package but this is a good starting point. You need to get your head around the fact that a particular output is required. You don’t need everything… some of the worst CAD based training models I have seen are when a developer has used all the CAD unrefined and hidden behind the shapes is huge amount's of data. Optimisation is key… The day when there is a fully automated process for cleaning, optimising, texturing and lowering polygon data comes, a whole bunch of technical artists will become developers. For now, to achieve the results shown, this is a manual process that requires care and attention. One thing that is obvious to me and my colleagues but that we never see happening is keeping a common data set. We want to take our export from a master CAD file. Not a file that has been split out 10 times, is 6 months out of date and almost un-useable. This for me creates a dangerous precedence that changes the scope of the requirement. No longer are we being asked to use this data too make a walkthrough, but take this model, guess how it should look now and fix it. Next we have the buzz words, through life support… The easiest way too do this to create new value early. Yes we are talking about training but next we shall look at how this new value can be achieved and move us towards true, through life support.
  17. In our recent collaboration with BAE, Babcock and others, we see at every junction where we can add value. As part of our feasibility into QEC industrial SQEP training, we looked at where the work we produced could be used by others in other areas. This although changing scope, moves us towards through life support. Some of these could include… Integrating tech pubs, showing how to maintain, when maintenance last took place and any real time failures. Full mission rehearsal between users. Practicing the physical maintenance tasks Embarking and Disembarking, this would include VIP, foreign forces and local fire etc. Refits and Repairs Decomissioning which I am sure has little to no consideration And this ladies and gentlemen I believe gives us true through life support. Sign Off!