We report on AVE's research on a new family of smart analog vision chips and ancillary software capable to adaptively select image processing parameters and regions of interest int he field of view. The project aims to adapt the wedge- and-strip (WS) position-sensitive configuration to real- time, multi-object detection. To that end, we use semiconductor WS detector (WESD) arrays with spatial prefiltering implemented via image algebra. The WESD have the ability to locate the centroid of an incident light spot and, with appropriate prefiltering, to locate the centroids of multiple light spots. A unique analog mode centroid computation scheme facilitates on-chip data processing, while an overall undemanding geometry simplify electronics fabrication. Based on results of preliminary design and analysis, very significant cost reductions over existing massively parallel vision processors ar foreseen. Superior optical resolution results even when the sensor is manufactured with 'coarse line width' semiconductor processing technology, which may lead to detectors with large photosensitive areas. Departing from photosensitive element miniaturization provides the opportunity to launch a new class of vision chips, with active area limited only by the semiconductor wafer size used to manufacture the array. We suggest several applications of this new technology, including an efficient and robust WDM demultiplexing device.