MISSING IMAGE DETAIL
We live in a complex world and take for granted our seamless ability to process thousands of objects around us of all sizes and shapes in wide open spaces like highways, parking lots, warehouses, etc. Current vision systems are limited to well defined objects within restricted fields of view, for example people in a corridor or cars nearby on the road ahead. Covering larger spaces or detecting smaller objects is far too expensive and complex using existing technology.
Vision systems that have unlimited budgets will still get too large and complex as the “law of diminishing returns” comes into play, additionally mixing in unproven technologies --like Lidar-- further reduce reliability and compromise safety.
Example of vehicle not detecting critical objects at distance ahead.
Tesla AutoPilot Not Detecting Stopped Traffic on Highway Ramp
video credit: Scott Kubo