Human Following Robot using Kinect in Embedded Platform

Abstract

This paper presents an embedded robotic system using Kinect sensor technology that is utilized to detect an individual target and track its movement in the surrounding. The developed system integrates both the features of computer vision and embedded robotics simultaneously. In this paper, a skeleton-based tracking algorithm is used because tracking plays an important role in localization and mapping. This system uses a gesture-based recognition that relies on Microsoft Kinect XBox 360 instead of a standard touch-based control system that can be utilized as a stand-alone system or subsystem to integrate into a larger system. The Kinect sensor captures the 3-dimensional information of the surroundings and recognizes the human body by retrieving the depth information that does not require wearing any kind of intrusive sensors. Firstly, this robotic system follows an individual by detection of torso point that is required for steering and maintaining a fixed safe distance for localization and mapping and provides a robust and reliable system. The proposed system can be utilized in a wide variety of applications such as work assistants, luggage carrying carts, etc.