Extraction of precise object orientation and position from LIDAR data using maximum-likelihood methods

LIDAR-based measurement systems can overcome several limitations in comparable technologies for the measurement and mapping of 3D static and dynamic objects in any given reference frame. As a result, they present distinct advantages for the determination of target velocity, acceleration, roll, pitch, yaw and position from long distances. Continuous, precise sensing and monitoring of remote targets has applications in various areas including military and commercial systems from ground, air or space. In this manuscript, we present the use of Maximum-Likelihood Estimation (MLE) methods for the extraction of precise object orientation and position information from a "waveform-sensing"LIDAR detector, where the finely-sampled (> GHz) temporal waveform of the signal generated by the diffuse-reflected laser pulse (i.e., laser pulse reflected off of the object and returned to collection optics) is used. In this method, multiple waveforms generated by the return pulse from various detectors stationed at optimized specific positions are collected. The time-of-flight (TOF), shape and the duration of waveforms indicate the radial extent of the object and distance to the receiver. Position and orientation are then extracted from the waveforms using MLE. First, we describe the forward-model simulation tool to generate LIDAR waveform data for an arbitrary object position and orientation. Next, we present a brief introduction into MLE followed by the application of this method to the extraction of position and orientation parameters from the simulated LIDAR data. Finally, results are presented to demonstrate the accuracy of the proposed method in recovering the input object orientation and position under presence of noise.