Search |  Contact |  SRI Home Do not follow this link, or your host will be blocked from this site. This is a spider trap. Do not follow this link, or your host will be blocked from this site. This is a spider trap. Do not follow this link, or your host will be blocked from this site. This is a spider trap.A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A ASRI International.  333 Ravenswood Avenue.  Menlo Park, CA 94025-3493. SRI International is a nonprofit corporation.

Publication in EndNote Format

%0 Report %@ 46 %A Duda, R.O. %T Some Current Techniques for Scene Analysis %C 333 Ravenswood Ave, Menlo Park, CA 94025 %I AI Center, SRI International %D 1970 %K Image processing, Scene Analysis, Robot vision, Shakey %X The purpose of the visual system is to provide the automaton with important information about its environment, information about the location and identity of walls, doorways, and various objects of interest. By adding new information to the model, the visual system gives the automaton a more complete and accurate representation of its world. The role of vision is not independent of the state of the model. If the automaton has entered a previously unexplored area, the visual scene must be analyzed to add information about the new part of the environment to the model. In this situation, the model can provide so little assistance that it is often not referenced at all. If, however, the automaton is in a thoroughly known area, the role of vision changes to one of providing visual feedback to correct small errors and verify that nothing unexpected has happened. In this situation, the model plays a much more important role in assisting and actually guiding the analysis. Until recently our attention has been directed primarily at the general scene-analysis problem. Every picture was viewed as a totally new scene exposing completely unknown area. More recently we have addressed the problem of using a complete, prespecified map of the floor area to update the automaton’s position and help in tasks such as going through a doorway. Another use of this kind of visual feedback would be the monitoring of objects being pushed. In trying to solve these problems, we have tended to take one or the other of two extreme approaches. Either we tried to develop general methods that can cope with any possible situation in the automaton s world, or we tried to exploit special facts that allow an efficient special-purpose solution. The first approach involves the more interesting problems in artificial intelligence, but it provides more capabilities than are needed in many situations, and provides them at the cost of relatively long computation times. The second approach provides fast and effective solutions when certain (usually implicit) preconditions are satisfied, though it can fail badly if the conditions are not met. Eventually, of course, some combination of the two approaches will be needed, since the automaton actually operates is a partially known world, rather than one that is completely unknown or completely known. However, we have decided to concentrate on the two extreme situations before addressing the intermediate case. The remainder %O SRI Project 8259 %U http://www.ai.sri.com/pubs/files/tn046-duda70.pdf

SRI International
©2014 SRI International 333 Ravenswood Avenue, Menlo Park, CA 94025-3493
SRI International is an independent, nonprofit corporation. Privacy policy