Humanoid Navigation: An Intelligent Computer Vision Based Approach

With an increasing demand towards use of humanoids in industrial automation and other related applications, navigation and path planning has emerged as one of the most promising area of research since last few decades. The current investigation is devoted towards the use of a computer vision technique integrated with the sensory network of a humanoid robot for navigation in a complex environment. Here, humanoid NAO has been used for navigational purpose. The basic problem that arises in the use of only a sensory network of a humanoid is that the ultrasonic sensors mounted on the humanoid NAO can detect and measure the distance between the obstacles and the humanoid but not their exact location. This leads to a deadlock situation when both the sensors detect the same obstacle simultaneously, or a very large sized obstacle comes exactly in front of the humanoid. Therefore, the camera mounted on the NAO itself is integrated with the sensors to provide a decision making ability when a deadlock situation arises. The proposed technique is tested with the humanoid NAO in a real-time experimental setup, and satisfactory results are obtained.