Loading...
Asymmetry regional boundary localization for indoor robot navigation
Author(s)
Wanadecha, N.
Date Issued
January 1, 2018
Type
Conference Paper
Abstract
This paper presents the new methodology of indoor localization based-on Received Signal Strength Indicator (RSSI) of Wireless Sensor Network (WSN) using Asymmetry Regional Boundary Localization (ARBL) which is designed to support mobile robot navigation. since many types of obstruction such as walls, poles, and traps in the environment may confuse robot trajectory and avoidance strategy, our main objective is to analyze our desire test bed in the form of asymmetry region with the benefit of RSSI map fingerprinting technique. The regions are created from multiple anchor nodes’ boundary intersection which is predetermined by the working range that provide the finest repeatability value. Each region is classified by supervised neural network. When the robot senses the existence of obstacle, that particular region is declared as occupied zone which is not allowed to enter during path planning process. Once the entire virtual map is acquired, suitable path is obtained by Q-learning algorithm, thus the robot can travel through the field. The experiments are performed with 2 different types of test bed. Each of which shows region classification performance of 71.9% and 85.8%, respectively. Finally, to demonstrate the effective of our method, dealing with obstacles in the field, the comparison between conventional Q-learning algorithm and our proposed method is shown. The proposed method not only gives better learning progress but also spends less computational resources, reducing the complexity of the navigation problem.
Citation
Lecture Notes in Engineering and Computer Science, 2238, 568-573, 2018