High performance reliable obstacle detection and height measurement by stereo camera for Intelligent Home Service Robot

Abstract

This paper proposes a new dimensional intelligent robotic application that can track an obstacle throughout its way of moving by using active stereo camera. The central part of the proposed method is the resilient back-propagation algorithm which can intellectually find out the ground even though there have a complex surface. After detecting the obstacle, proposed method measures the height of the object so that it becomes possible for the robot to get on top of the obstacle if its height is negligible. The system is mainly proposed for an Intelligent Home Service Robot to work perfectly in different environment on different condition. A searching algorithm is proposed for the system to detect obstacle by neural-network which can search fast enough to apply it in real-time application. In best case its cost is only a constant and in worst case it also needs short time to find different obstacles.