TY - JOUR

T1 - Optimization of 3D laser scanning speed by use of combined variable step

AU - Garcia-Cruz, X.M.

AU - Sergiyenko, Oleg Yu

AU - Tyrsa, Vera

AU - Rivas-Lopez, Moises

AU - Hernandez-Balbuena, Daniel

AU - Rodríguez-Quiñonez, Julio C.

AU - Basaca-Preciado, Luis C.

AU - Mercorelli, P.

PY - 2014/3

Y1 - 2014/3

N2 - The problem of 3D TVS slow functioning caused by constant small scanning step becomes its solution in the presented research. It can be achieved by combined scanning step application for the fast search of n obstacles in unknown surroundings. Such a problem is of keynote importance in automatic robot navigation. To maintain a reasonable speed robots must detect dangerous obstacles as soon as possible, but all known scanners able to measure distances with sufficient accuracy are unable to do it in real time. So, the related technical task of the scanning with variable speed and precise digital mapping only for selected spatial sectors is under consideration. A wide range of simulations in MATLAB 7.12.0 of several variants of hypothetic scenes with variable n obstacles in each scene (including variation of shapes and sizes) and scanning with incremented angle value (0.6° up to 15°) is provided. The aim of such simulation was to detect which angular values of interval still permit getting the maximal information about obstacles without undesired time losses. Three of such local maximums were obtained in simulations and then rectified by application of neuronal network formalism (Levenberg-Marquradt Algorithm). The obtained results in its turn were applied to MET (Micro-Electro-mechanical Transmission) design for practical realization of variable combined step scanning on an experimental prototype of our previously known laser scanner.

AB - The problem of 3D TVS slow functioning caused by constant small scanning step becomes its solution in the presented research. It can be achieved by combined scanning step application for the fast search of n obstacles in unknown surroundings. Such a problem is of keynote importance in automatic robot navigation. To maintain a reasonable speed robots must detect dangerous obstacles as soon as possible, but all known scanners able to measure distances with sufficient accuracy are unable to do it in real time. So, the related technical task of the scanning with variable speed and precise digital mapping only for selected spatial sectors is under consideration. A wide range of simulations in MATLAB 7.12.0 of several variants of hypothetic scenes with variable n obstacles in each scene (including variation of shapes and sizes) and scanning with incremented angle value (0.6° up to 15°) is provided. The aim of such simulation was to detect which angular values of interval still permit getting the maximal information about obstacles without undesired time losses. Three of such local maximums were obtained in simulations and then rectified by application of neuronal network formalism (Levenberg-Marquradt Algorithm). The obtained results in its turn were applied to MET (Micro-Electro-mechanical Transmission) design for practical realization of variable combined step scanning on an experimental prototype of our previously known laser scanner.

KW - Engineering

KW - Dynamic Triangulation

KW - Neural Network

KW - Scanning step

KW - Scene

KW - Technical vision system

U2 - 10.1016/j.optlaseng.2013.08.011

DO - 10.1016/j.optlaseng.2013.08.011

M3 - Journal articles

VL - 54

SP - 141

EP - 151

JO - Optics and Lasers in Engineering

JF - Optics and Lasers in Engineering

SN - 0143-8166

IS - March

ER -