Human recognition in a cluttered indoor environment by sensor fusion

Author(s):  
Sang-Yoon Kim ◽  
Sang-Roan Lee ◽  
Tae-Yang Kuc
Author(s):  
Ritwik Murali ◽  
Dhivya Nachimuthu ◽  
Dhansri Varsha SenthilKumar ◽  
Malarvizhi Shanmuga Pandian ◽  
Dhareni Krishnen

2010 ◽  
pp. 22-30
Author(s):  
Julian Lategahn ◽  
Frank Kuenemund ◽  
Christof Roehrig

In this paper a method for estimation of position and motion of a mobile robot in an indoor environment is introduced. The proposed method uses WLAN signal strength to estimate the global position of a mobile robot in an office building. Thus signal strengths of the received access points are stored in the radio map in calibration phase. In localization phase the stored values are compared with actually measured one’s. Therefore a fingerprinting algorithm, that was introduced before, is used. The improvement of the presented work is the multi sensor fusion using Kalman filter, which enhances the accuracy of fingerprinting algorithms and tracking of the robot. For this reason odometric and gyroscopic sensors of the robot are fused with the estimated position of the fingerprinting algorithm. The paper presents the experimental results of measurements made in an office building.


2009 ◽  
Vol 21 (1) ◽  
pp. 28-35 ◽  
Author(s):  
Songmin Jia ◽  
◽  
Jinbuo Sheng ◽  
Daisuke Chugo ◽  
Kunikatsu Takase

In this paper, a method of human recognition in indoor environment for mobile robot using RFID (Radio Frequency Identification) technology and stereo vision is proposed as it is inexpensive, flexible and easy to use in practical environment. Because information of human being can be written in ID tags, the proposed method can detect the human easily and quickly compared with the other methods. The proposed method first calculates the probability where human with ID tag exists using Bayes rule and determines the ROI for stereo camera processing in order to get accurate position and orientation of human. Hu moment invariants was introduced to recognize the human being because this method is insensitive to the variations in position, size and orientation. The proposed method does not need to process all image and easily gets some information of obstacle such as size, color, thus decreases the processing computation. This paper introduces the architecture of the proposed method and presents some experimental results.


Sign in / Sign up

Export Citation Format

Share Document