In recent times, a significant trend has emerged in societal infrastructures: while new construction continues to take place, the aging of existing facilities has made the need for repairs and upgrading more acute. For instance, while the full cost of a new plant might not be justified, the economics of retaining competitiveness may best be fulfilled by upgrading existing facilities. This shift from an engineering to a re-engineering paradigm will further expand in the future. It generates new requirements in terms of intervention planning, optimization and execution.

In this re-engineering paradigm, it often occurs that the status of the facilities to be upgraded /repaired is not well known: man-made alterations have taken place and have not been documented in a cohesive manner; natural degradations have occurred over time and are difficult to assess because of hazards for workers or difficulties of stopping the plant (for instance cracks in dams/bridges, algae deposits on the surface of penstocks, zebra mussels on ship bottoms). Because the costs of the intervention may be considerable or its logistics quite complex the actual work to be done must be planned with care and rigorously executed.

There exists a broad class of emerging tasks which share common threads and for which the use of new information tools could bring considerable pay-off:

There is need to document actual installations in situ (i.e. gather geometric and other data) for the purpose of repair, inspection, certification, etc. The data may have to be brought in registration with the overall structure, and/or be linked with other a priori spatial information such as CAD records or engineering plans which are sometimes incomplete or out of date. Multiple resolutions may be needed, from full scene to detailed regions-of-interest.

These data must be transformed into workable computer models which document the status of the installations and which are cast in ways capable of supporting future intervention. Static features, such as shape and visual appearance must be encapsulated but...

These representations must also be augmented with behaviors describing the dynamic physical attributes of the sites and of the relevant phenomenons. There must also be accurate geometric and physical models of the tools (robot, ROV, actuators, etc.) that are envisioned for the intervention.

There must be means whereby task planning and optimization can be achieved, typically through simulations based on the modeled sites and intervention tools.

These means must be engineered in such a way that they also come into play during the execution of a task, for instance by a combination of supervisory tools that assist the operator managing the task through teleoperation of remotely controlled vehicle and complex semi-autonomous tools.

The above problems are yet to be solved both at the theoretical and industrial levels.

VERTEX aims at the development of resources which emphasize interactive use and effective integration through virtual and / or augmented environments. The planning phase and simulations are geared toward defining optimal scenarios while the supervisory tools help the operator to direct and monitor the actual accomplishment of the task.

While performing complex tasks on environments which are not initially well documented or easily accessible is a broad application area, VERTEX focuses its objectives on specific sub-topics. The conceptual level is concerned with providing a generic approach for exploiting the integration, through simulation and interactive virtual environments, of geometric and behavior models of actual sites and intervention tools in order to generate optimal scenarios of intervention. The application level concentrates on implementing the designed scenarios in situ under operator supervision.

VERTEX is proposed as a direct sequel to "VRES: Tools for the Rapid Prototyping of Virtual Representations of Existing Sites", an on-going Project funded through the NSERC-NRC RPP. VRES is carried out with the industrial collaboration of IREQ and Innovmetric Software Inc and in partnership with NRC / IIT. The 3-year VRES Project, which has started Nov. 1 1996, aims at using computer vision and human-machine interface technology for minimally-constrained acquisition of in situ geometric and photometric visual data and for the efficient and rapid operator-guided generation of 3D representations of existing sites at several levels of detail (LOD). VRES is limited to documentation aspects. However VRES has adopted modelling paradigms which allow the inclusion of behaviors. There are, therefore, strategic conceptual and temporal connections between VRES and VERTEX both of which rely on research accomplished in IRIS I/II.

In this context, at the conceptual level,VERTEX seeks the following objectives:

1. develop a generic approach for integrating behaviors and physical properties with geometric/photometric models of existing sites on which an upgrading or re-engineering task must be performed.
2. develop a generic approach for modeling the geometric and functional properties of tools which are used for the re-engineering task
3. develop authoring software modules for generating scenarios using the above behaviors (for scene components and tools) that will enable simulations of the task to allow operator training for a particular intervention mission and will allow to define the constraints for the selection of intervention tools and operational procedures
4. develop a strategy for efficient interactive planning of a task in simulation mode
5. develop a suitable human-machine interface (HMI) which exploits augmented reality for allowing an operator to participate in the simulation and planning.