Using neural networks for high resolution distance measurements in pulsed laser radar

We have developed a new distance measurement method which can obtain distance information directly from the output waveform of pulsed laser radar (PLR). A simple digital signal processing technique and multilayer perceptrons (MLP) have been used to recognize the pulse shape and to obtain the distance information. The method has been implemented in a real PLR for high resolution distance measurements to improve the resolution and to decrease the nonlinearity error. Because of the ability of neural networks in decreasing the noise and preprocessing of the noisy input pulse shapes to the neural network, resolution and nonlinearity were greatly improved. Distance deviation of 53 /spl mu/m-168 /spl mu/m, full width at half power (FWHP) of 70 /spl mu/m-190 /spl mu/m and non-linearity of 187 /spl mu/m have been achieved. All the measurements in the same situation has been performed by using the standard method to extract the distance information from time interval between the reference pulse and the reflected pulse. In comparison with the standard method, resolution in the best case and non-linearity were improved by 86% and 6.5% respectively. In this method if the PLR system is reasonably stable during the measurement, it is possible to use only the reflected pulse from the target to extract the distance information and this makes PLR simpler in hardware. Because the neural network decreases noise, it is possible to make the measurements with the same resolution of standard method but with the lower averaging in sampling unit and this dramatically increase the speed of the measurement.

[1]  Simon Haykin,et al.  Classification of radar clutter using neural networks , 1991, IEEE Trans. Neural Networks.

[2]  Timothy C. Strand.  Optical Three-Dimensional Sensing For Machine Vision , 1985 .

[3]  Qi-Jun Zhang,et al.  Neural Networks for RF and Microwave Design , 2000 .

[4]  Akira Uritani,et al.  New position sensing method for position-sensitive proportional counter with neural network algorithm , 1997 .

[5]  Simon Haykin,et al.  Neural Networks: A Comprehensive Foundation , 1998 .

[6]  G. Kompa,et al.  Powerful picosecond-pulsed laser radar with micrometer ranging resolution , 1996, 1996 26th European Microwave Conference.

[7]  Thomas F. Edgar,et al.  Constructing a reliable neural network model for a plasma etching process using limited experimental data , 1994 .

[8]  Risto Myllylä,et al.  Acquisition Of Three-Dimensional Image Data By A Scanning Laser Range Finder , 1989 .

[9]  G. Kompa,et al.  High Resolution Hip-joint Socket Imaging Using Pulsed Laser Radar , 2000, 2000 30th European Microwave Conference.

[10]  Akira Uritani,et al.  Energy resolution enhancement of CdTe semiconductor detector spectra by using a neural network algorithm , 1996 .

[11]  Clark S. Lindsey,et al.  Filtering data from a drift-chamber detector using neural networks , 1993 .

[12]  Gunter Kompa.  Extended Time Sampling for Accurate Optical Pulse Reflection Measurement in Level Control , 1984, IEEE Transactions on Instrumentation and Measurement.

[13]  Hiroaki Ikeda,et al.  Compact and high-precision range finder with wide dynamic range and its application , 1992 .

[14]  Gary S. May,et al.  An optimal neural network process model for plasma etching , 1994 .

[15]  Jeffrey R. Carter,et al.  Generalizing from a small set of training exemplars for handwritten digit recognition , 1991 .

[16]  M. Kawato,et al.  Estimation of generalization capability by combination of new information criterion and cross validation , 1991, IJCNN-91-Seattle International Joint Conference on Neural Networks.