| Return to RESOLV page | ISR/IST Laboratório de Robótica Móvel |
Note(in April 1999): This text was originally published in the Fall of 1995 and updated during 1996. Naturally, the project has changed since then and this text may be now outdated. It is kept public for the sake of project documentation.
The purpose of RESOLV is to produce a cost-effective system for creating realistic 3D models of the environment. These will be in a form that is suitable for enhancing the sense of 'being there' in telepresence and virtual reality applications. The models will also be sufficiently accurate for real estate, construction and other industrial applications.
A portable unit known as an EST (Environmental Sensor for Telepresence) is taken round the environment that is to be captured. The EST includes a scanning laser rangefinder for capturing the 3D structure of the surroundings and a video camera for adding the textures.
The environment is scanned from a number of capture positions but the reconstructed model can be viewed from any position. Surfaces are recognised by processing the range data and are textured from the camera images. By combining what is seen from neighbouring capture positions surfaces that would be occluded from one position are represented.
The EST is taken from one capture position to another on a trolley or an autonomous vehicle. The location of the new position is referred to the previous one by using key points in the surroundings. The environment is reconstructed as the EST progresses and the partial reconstruction is used to determine future capture positions.
The EST is optimised for human scale applications. It is not intended for detailed reconstruction of small objects or for capturing an external landscape. Both the trolley shown in the picture and the autonomous vehicle are designed to support capture at two heights - eye level when sitting and standing. The size of the unit is comparable with that of a person to ensure that it can be taken to all the places where people are likely to pause when looking around a building. The data will be held in a form suitable for CAD systems and for viewing on a WWW browser. Stereo viewing is possible. Live video of people can be inserted into the reconstruction for telepresence applications. In this way remote parties can look around the people they are talking to as if they were all in the same location.
RESOLV is related to products such as Quicktime VR from the Apple Corporation in creating a virtual copy of a real location. However the observer will have an enhanced sense of presence through being able to move freely within the reconstructed environment. The pictorial quality will be exceptionally good (equal to existing visual panorama systems) when the observer is near the environment capture point and will gracefully degrade in moving away from capture points.
An existing location will be captured in sufficient detail for viewing through a VRML viewer. Peoples bodies will then be identified from live video and inserted into the reconstruction. The video can include people who are actually at the location and others who are only telepresent. In this way it will be possible to support the illusion of all members of a distributed meeting being in the same place.
Telepresence in a reconstruction of a real location would provide added impact for any collaborative virtual environment. For instance a WWW home page could include a (possibly enhanced) reconstruction of the owners real home in which they could entertain telepresent friends. The social telepresence trial will take place between project partners and between staff in factories addressing similar factory organisation and layout issues.
Visualisations derived from CAD models are becoming an essential tool for the construction industry. However reconstructing the dimensions and appearance of existing interiors can be a time-consuming and expensive process. This trial will evaluate the cost- effectiveness of capturing relevant details of existing interiors for refurbishment, maintenance and facilities management applications. The trial will be undertaken by the subsidiary Facilities Management company of the lead partner.
The key technical issue is establishing how accurately the surroundings need to be scanned in order to produce a reconstruction of sufficient realism for telepresence, and adequate accuracy for industrial applications. This will indicate the potential cost-effectiveness of the service and hence the size of the potential market.
The credibility of the reconstruction is critically dependent on the speed of the processor in the viewing machine. Dramatic performance increases in PC 3D and graphics cards indicate that mass market systems would be able to make use of RESOLV reconstructions as soon as they could become widely available.
The output from the project will be a set of algorithms for producing 3D reconstructions of entended environments requiring the fusion of range data and texture information from many capture points.
These algorithms will be implemented in two types of environmental capture device, one portable and the other autonomous. An assessment will be made of the nature of the markets for the overall methodology. Technical results will include an assessment of which applications suit the RESOLV approach of integrating laser scanning with digital images.
RESOLV is expected to be one of many catalysts in encouraging a universal move to more natural three dimensional user interfaces. The most obvious demand is for truly spatial activities such as building design. However there is increasing interest in providing familiar spatial contexts in which more abstract issues can be discussed.
The continuing increase in the realism of virtual environments has created a desire to make the images look more and more like particular real remote locations. However conventional methods of capturing structure and texture are too slow and expensive for most locations that people wish to be telepresent within and which might include offices, factories and homes. Both RESOLV applications will improve the usability of future multi-way multimedia services by providing the environmental realism needed to feel telepresent in remote locations. As such it will contribute to European competitiveness by lowering the cultural and psychological barriers to communication between people in different countries.
More detailed information may be found on the Web pages of the various partners,
|
Return to the RESOLV main page. |
 |
 |
 |
 |
 |
 |
|---|---|---|---|---|---|
| ©1999 LRM, Maria Isabel Ribeiro, João Gomes Mota | 16-April-1999 | ||||