Universiti Teknologi Malaysia Institutional Repository

Enhanced localization with adaptive normal distribution transform Monte Carlo localization for map based navigation robot

Lim, T. Y. and Yeong, C. F. and Su, E. L. M. and Shithil, S. M. and Chik, S. F. and Duan, F. and Chin, P. J. H. (2019) Enhanced localization with adaptive normal distribution transform Monte Carlo localization for map based navigation robot. Journal of Electrical Engineering, 18 (3-2). pp. 17-24. ISSN 01298-4428


Official URL: https://elektrika.utm.my/index.php/ELEKTRIKA_Journ...


Map-based navigation is the common navigation method used among the mobile robotic application. The localization plays an important role in the navigation where it estimates the robot position in an environment. Monte Carlo Localization (MCL) is found as the widely used estimation algorithm due to it non-linear characteristic. There are classifications of MCL such as Adaptive MCL (AMCL), Normal Distribution Transform MCL (NDT-MCL) which can perform better than the MCL. However, AMCL is adaptive to particles but the position estimation accuracy is not optimized. NDT-MCL has good position estimation but it requires higher number of particles which results in higher computational effort. The objective of the research is to design and develop a localization algorithm which can achieve better performance in term of position estimation and computational effort. The new MCL algorithm which is named as Adaptive Normal Distribution Transform Monte Carlo Localization (ANDT-MCL) is then designed and developed. It integrates Kullback–Leibler divergence, Normal Distribution Transform and Systematic Resampling into the algorithm. Three experiments are conducted to evaluate the performance of proposed ANDT-MCL in simulated environment. These experiments include evaluating the performance of ANDT-MCL with different path shape, distance and velocity. In the end of the research work, the proposed ANDT-MCL is successfully developed. It is adaptive to the number of particles used, higher position estimation and lower computational effort than existing algorithms. The algorithm can produce better position estimation with less computational effort in any kind paths and is consistent in long journey as well as can outperform in high speed navigation.

Item Type:Article
Uncontrolled Keywords:Localization, Map based navigation, MCL, NDT-MCL, AMCL
Subjects:T Technology > TK Electrical engineering. Electronics Nuclear engineering
Divisions:Electrical Engineering
ID Code:85281
Deposited By: Fazli Masari
Deposited On:17 Mar 2020 08:10
Last Modified:17 Mar 2020 08:10

Repository Staff Only: item control page