P.E. An

Vehicle detection and recognition in greyscale imagery

Abstract This paper details a novel two-stage vehicle detection and recognition algorithm by combining an image-processing region of interest (ROI) designator to cue a secondary recognition process implemented using principal component analysis (PCA) as input to a Multi-Layered Perceptron (MLP) classifier. Both the image-processing system and the MLP classifier have been designed for real-time implementation and data-fusion with other information sources.

Enhancement of the inertial navigation system for the Morpheus autonomous underwater vehicles

This paper presents the design and development of an enhanced inertial navigation system that is to be integrated into the Morpheus autonomous underwater vehicle at Florida Atlantic University. The inertial measurement unit is based on the off-the-shelf Honeywell HG1700-AG25 3-axis ring-laser gyros and three-axis accelerometers and is aided with ground speed measurements obtained using an RDI Doppler-velocity-log sonar. An extended Kalman filter has been developed, which fuses together asynchronously the inertial and Doppler data, as well as the differential Global Positioning System positional fixes whenever they are available. A complementary filter was implemented to provide a much smoother and stable attitude estimate. Thus far, preliminary study has been made on characterizing the inertial navigation system-based navigation system performance, and the corresponding results and analyzes are provided.

An intelligent driver warning system for vehicle collision avoidance

This paper describes the basic architecture of an intelligent driver warning system which embodies an adaptive driver model for indirect collision avoidance. In this study the driver modeling objective is focused only on longitudinal car-following, and the model inputs are chosen to be the past history of throttle angle, controlled vehicles speed, range and range rate to the front vehicle whereas the model output is chosen to be the current throttle angle. An artificial neural network called cerebellar model articulation controller (CMAC) and a conventional linear model (CLM) are independently applied to model the real driver data taken from test track and motorway environments. The CMAC model is chosen because of its nonlinear modeling capability, on-line learning convergence and minimum learning interference characteristics, whereas the linear model is chosen as a control benchmark to examine the nonlinear characteristic of the drivers behavior. The modeling capabilities are then evaluated based on one-step ahead prediction error performances over the training and testing sets, learning curves and correlation based model validation techniques. Modeling results suggest that the past history of throttle angle plays a critical role in reducing the deviation of the error correlation, which in turn suggest that the throttle dynamics are generally slow for road driving. Also, the time scale dependency of the model on the drivers behavior varies significantly from the test track to motorway environment. In the driver modeling experiment, the time scale was chosen such that the deviation of the error correlation was minimized. The test track results suggest that the chosen inputs are indeed relevant variables for modeling the drivers behavior. Unlike that of the CLM, the degree of the error deviation of the CMAC model was found to be acceptable for the test track scenario, implying a significant nonlinear coupling of the throttle output with the speed, range and range rate data. Whereas for the motorway data, the modeling performance for both models is comparable, and the time scale of the driver model is approximately three times longer than that used in the test track data.

Modeling and simulation of autonomous underwater vehicles: design and implementation

Autonomous underwater vehicles (AUVs) have many scientific, military, and commercial applications because of their potential capabilities and significant cost-performance improvements over traditional means for performing search and survey. The development of a reliable sampling platform requires a thorough system design and many costly at-sea trials during which systems specifications can be validated. Modeling and simulation provides a cost-effective measure to carry out preliminary component, system (hardware and software), and mission testing and verification, thereby reducing the number of potential failures in at-sea trials. An accurate simulation can help engineers to find hidden errors in the AUV embedded software and gain insights into the AUV operations and dynamics. This paper reviews our research work on real-time physics-based modeling and simulation for our AUVs. The modeling component includes vehicle dynamics, environment and sensor characteristics. The simulation component consists of stand-alone versus hardware-in-the-loop (HIL) implementation, for both single as well as multiple vehicles. In particular, implementation issues with regard to multitasking system resources will be addressed. The main contribution of this paper is to present the rationale for our simulation architecture and the lessons learned.

Very high-frequency radar mapping of surface currents

An ocean surface current radar (OSCR) in the very high frequency (VHF) mode was deployed in South Florida Ocean Measurement Center (SFOMC) during the summer of 1999. During this period, a 29-d continuous time series of vector surface currents was acquired starting on 9 July 1999 and ending 7 August 1999. Over a 20-min sample interval, the VHF radar mapped coastal ocean currents over a 7.5 km /spl times/ 8 km domain with a horizontal resolution of 250 m at 700 grid points. A total of 2078 snapshots of the two-dimensional current vectors were acquired during this time series and of these samples, only 69 samples (3.3%) were missing from the time series. During this period, complex surface circulation patterns were observed that included coherent, submesoscale vortices with diameters of 2 to 3 km inshore of the Florida Current. Comparisons to subsurface measurements from moored and ship-board acoustic Doppler current profiles revealed regression slopes of close to unity with biases ranging from 4 to 8 cm s/sup -1/ between surface and subsurface measurements at 3 to 4 m beneath the surface. Correlation coefficients were 0.8 or above with phases of - 10 to - 20/spl deg/ suggestive of an anticyclonic veering of current with depth relative to the surface current. The radar-derived surface current field provided spatial context for an observational network using mooring-, ship- and autonomous underwater vehicle-sensor packages that were deployed at the SFOMC.

Asynchronous data fusion for AUV navigation via heuristic fuzzy filtering techniques

This paper presents a heuristic fuzzy position estimation technique for autonomous underwater vehicle navigation. The heuristic estimator performs asynchronous data fusion of all sensor measurements based on their relative confidence levels, and then nonlinearly combines the fused information with the INS estimates via fuzzy filtering techniques. In this paper, the basis and implementation of the estimator will be described, and navigation results will be presented based on the heuristic estimator. In addition, performance comparison based on the heuristic estimator and those based on extended Kalman filters will be reported in our companion paper, and the results are expected to provide insights into the pros and cons of individual methods in terms of computational cost, steady-state and convergence characteristics for bias estimation.

Vehicle Detection and Recognition in Greyscale Imagery

This paper details a novel two-stage vehicle detection and recognition algorithm by combining an image processing region of interest (ROI) designator to cue a secondary recognition process implemented using principal component analysis (PCA) as input to a Multi-Layered Perceptron (MLP) classifier. Both the image processing system and MLP classifier have been designed for real-time implementation and data-fusion with other information sources.

An AUV survey in the littoral zone: small-scale subsurface variability accompanying synoptic observations of surface currents

A survey of small-scale subsurface variability within the synoptic observational field of an ocean surface current radar (OSCR) using an autonomous underwater vehicle (AUV) is described. The survey involved observation of a developing upper mixed layer in a littoral zone off southeast Florida, on the edge of a strong Florida current during the summer of 1999. Complimentary in situ observations from a bottom-mounted acoustic Doppler current profiler (ADCP), conductivity-temperature (CT) chain arrays, atmospheric measurements from a surface buoy, and CTD and ADCP observations from a surface ship provided the background to the survey. The AUV, the Ocean Explorer, equipped with a CTD, downward and upward looking ADCPs, and a small-scale turbulence package, was used to conduct a continuous 12-h survey of small-to-fine-scale variability within a few grid cells of the surface current radar field. The vehicle repeatedly sampled the same grid in a set pattern at a fixed mid-water depth. Maps of developing spatial distribution of current, salinity, temperature, and rate of dissipation have been developed using the AUV-based observations. The observed features in the current field compare well with the OSCR and the bottom-mounted ADCP measurements.

Design Improvements in Associative Memories for Cerebellar Model Articulation Controllers (CMAC)

A number of recent improvements to the design of associative memories for CMAC systems are described. These are (i) an improved scheme for allocating C weights to a given input vector in R^n, (ii) design of receptive field shapes within the hypercube associated with an indvidual weight (including some experimental evaluations of these shapes), (iii) matching the field shapes to the hypercube itself using the concept of superspheres, (iv) speeding up the convergence of the weight training procedure.

Strategies for simultaneous multiple autonomous underwater vehicle operation and control

Under sampling of the coastal oceans remains a persistent problem for standard oceano-graphic measurement practice wherein an instrument package is tethered to a research vessel. The overhead costs associated with operating a large research vessel impose a strict minimum on the cost of data collected. Owing to the overheads, significant improvements in sampling technology on the tethered platform can only produce modest gains in the cost effectiveness. In contrast, untethered vehicles if operated simultaneously have the potential to increase cost effectiveness significantly by distributing the overhead costs over several sampling platforms. Furthermore, synoptic and pseudosynoptic data can be collected with multiple autonomous underwater vehicles (AUVs), thereby providing the type of information critical to dynamic process modeling unattainable with non-synoptic data. While the goal of simultaneous multiple-vehicle operation has been espoused over the last few years, AUV technology and practice have until...