Goals

We have conducted a significant collaborative research review. From this review we have arrived at a draft agenda for R&D. This agenda represents our goals for CADLab R&D, as annually reviewed by our industry members and redirected as necessary. Projects will then be identified and assigned to specific researchers and graduate students.

Goal 1: Collaborative Understanding

Initial investigations should examine current tools for development collaboration, and how industries are applying these tools, particularly examining business practices when distributing IP through globally distributed product groups. The investigation should also consider expanding API access to the CAx kernel, identifying where the API could be modified for a multi-user environment. This would include direct access to the event buffer, and new API's that would extract the core parameters for the various design objects, rather than handles to proprietary modeling objects.

Goal 2: Practical CS Collaborative Architecture

Numerous prototypes are working on CS architectures, where the model is replicated at each site. If model security can be relaxed among a set of trusted product engineers, then the development of a few more tools will make CS collaboration practical. These include CS for localized collaboration, distributed server clustering, and cloud serving.

• Client-Server – In more localized environments where the network proves reliable and consistent, and assuming that several CAx companies are willing to expand/modify their API set to resolve some of the minor deficiencies mentioned earlier, a strict CS architecture could support multi-user collaboration. New multi-user software, such as the NC Connect prototype developed at BYU or some of the other notable prototypes, could be used to generalize an API set for multi-user collaboration, with capabilities to expand current CAx GUI's for multi-user observation and model activity updating.
• Distributed server clustering - Gaming technologies have made strategic advances in networked server distribution/clustering that could prove useful to collaborative CAx. Network latency is a critical dynamic to both successful gaming and CAx collaboration. Global companies will have development offices distributed among several global sites, and thus need to organize their facility servers by user density and development tasks. PLM systems need to incorporate a simultaneous access multi-user file mentality different from current versions, allowing a manager to assign simultaneous user file read/write access. Other PLM procedures will need to be developed to dynamically update part files in a simultaneous multi-user environment, with some type of manager alert, review and approval process.
• Cloud serving – Whereas a CAx application typically resides at each user PC/workstation computer, cloud serving technologies may offer even better multi-user architectures and provide for enhanced product distribution security. Windows sharing and RGS technologies provide for new methods to view application windows on the cloud servers.

Goal 3: Model Decomposition and Multi-user Constraints

CAx and PLM systems understand part assemblies and divided ownership. It seems that expanding these concepts to regional model decomposition would be practical using many of these same features. Whereas boundary conditions are used to segregate and join assemblies, constraint equations and other functional relationships could be used to decompose continuous parts regionally, by volume, surface, perimeter, functional relationships, etc. – see the figure. New methods would assign users to a specific model region, divide the regions by constraint features or functional relationships, monitor the screen cursor against the region boundaries, and expose/hide/lock a decomposed model region from other collaborators. The CAx GUI should enable only those menu options appropriate to the features within the user's editing region.

Goal 4: Multi-User Screen Dynamics

Generally, CAx API's provide for customizing GUI menus/panels for individual user needs, but are not designed for multi-cursor-user collaboration because computer OS's presume a single event cursor per computer screen and do not have API's to communicate part changes among a multi-user group. Newer OS's now incorporate multi-cursor capabilities that could show other user design edits, but for viewing purposes only. Regional blocking will not allow removed users to change a locally owned model. A strong need for new collaborative GUI's would be the interpretation of multi-user edit intent among collaborators, and its timely transmission. Multi-user awareness is critical to those engaged in collaborative design, along with appropriate data filtering. Agent software has been shown to be effective in capturing and transmitting local changes like these, and could be designed to capture the various cultural elements in global product development. Agent software could be incorporated into existing GUI's with appropriate filters to pass only the data relevant to the current user activities. We propose that agent software be strongly investigated as an add-on capability to existing CAx GUI's. Many gaming features such as avatars, lexicons, predictive actions, and regional dynamics, could prove useful in interpreting another user's design intent among several users.

Goal 5: Collaborative Layering

The figure shows the conventional product development process on the left side. Historically, the steps of specification (constraints), design, analysis, and process planning consume the Engineering Design Process activities. These activities are conducted serially, with the related tasks spread among dedicated departments and personnel. When downstream design or manufacturing errors are discovered, feedback modifications move the product/part back upstream. CAx applications are designed for this serial streaming and feedback process, and perhaps even encourage its serial nature. We would suggest that existing CAx architectures can be changed to encourage the process layering shown by the Venn diagram. Regional decomposition, when carefully considered, could encourage several CAx processes to be conducted simultaneously in partially or fully completed regions, such as laying down finite element models in completed regions, or tool paths. Dimensioning and manufacturing features, tolerances, tool and machine selection are examples of activities that could be applied immediately after the part geometry is reasonably well defined. This level of collaboration would encourage personnel with different expertise to provide more timely feedback and reduce the development cycle.

Goal 6: Collaborative Security

Product models contain sensitive information (trade secrets, patentable ideas) that must be protected to preserve competitive advantage. The new collaborative modes that we have investigated introduce new patterns of data sharing that must provide appropriate access control so that only authorized individuals and processes gain access to information and are authorized to make changes. It is necessary to explore fine-grained, scalable, easy-to-use, and flexible access control methods to ensure that all changes are consistent according to policy and only available to authorized entities. New methods must be suitable for access control in open systems where the participants are not in the same security domain. Traditional authentication approaches based on identity may not be suitable for an open environment. New policy-driven approaches that rely on trust management approaches for sharing across organizational boundaries should be considered, along with audio and video recognition.