Transformed Stereo Vision and Structure Sensor for Development 3D Mapping on "FLoW" Humanoid Robot in Real Time

Abstract

In this paper, we proposed a method for building 3D Mapping environment in real time. The purpose of this model is to be able to approach the human ability to quickly know the environment. The problem of building a 3D mapping in real time is dependent on the efficiency of the algorithm to transform 2D images into depth data forms, which then transformed into a 3D model. This paper shows a method of transformation which has built an efficient algorithm because it can complete the entire sequential algorithm in real time. The algorithm has successfully run the whole system that only has a depth pixel error average of 18,10% and an average error of the system running in real time.