"For a human or a bird, the task is trivial: visually track an object moving through a cluttered scene. But computer vision can’t do this- there are too many visual ambiguities for even the most advanced artificial intelligence programs. Michael Acheson Isard is working to get past these obstacles. For his doctoral thesis at Oxford, he devised an algorithm called CONDENSATION, for CONditional DENSity PropagATION. Isard’s thesis supervisor, Andrew Blake, says Isard’s technique promises a ""revolution in the design of intelligent machines."" The premise is that as a computer watches a scene, it must continually reweigh alternative interpretations of what is signal and what is noise-mimicking human visual perception. Isard’s work, says Blake, has ""sparked a whole new area of activity by other researchers.""
This fall, following a three-year postdoc at Oxford’s Magdalen College, Isard took a research position at Compaq’s Systems Research Center (SRC) in Palo Alto. According to Blake, this Anglo-American, born in England to American parents, is ""set to have a major impact on the way machine intelligence develops over the next decade."