Skip to Main Content
One of the fastest growing next generation consumer products is a robot technology industry and it is rapidly forming a huge market. A depth extraction is one of the key techniques to adopt a home assistant robot for avoiding obstacles and looking for a transfer route. However, a home assistant robot mounted stereo vision systems produce unpredictable changes in video sequences when the robot is walking and needs hard computation to produce an accurate estimation. In this paper, we propose hardware based realtime stereo depth extraction method for an intelligent home assistant robot. For fast adaption of the external environment, we present depth map extraction algorithm by using preprocessing, parallel prediction searching employing median filter. Experimental results show that the proposed method reduces the processing time and the energy consumption compared to the conventional methods. This implementation is suitable for adaptive real-time depth extraction that is compatible with current robot applications.