Distributed smart cameras for hard real-time obstacle detection in control applications

This paper describes the integration of distributed image analysis in a smart camera network with a control application. The distributed smart cameras analyze the environment of the controlled object and transfer accurate position and orientation information of all detected obstacles to the controller within guaranteed time bounds. With this information the controller can optimize the trajectory of the load. We present two different approaches for estimating the bounding boxes for detected obstacles and compare their benefits and drawbacks.