This paper presents the results of a study designed to determine robustness of the performance improvements attributable to the use of space-based sensors for cueing a Navy Theater Ballistic Missile Defense (TBMD) Systems. The key system parameters of interest in the study are (1) the time delay, defined as the time interval between the threat missile burnout and the time at which the cue (threat missile state vector at burnout and covariance matrix) arrives at the TBMD shooter, and (2) the accuracy of the state vector. Performance characteristics of the space-based sensors and TBMD shooter subsystems were investigated parametrically. This study, with the exception of an excursion to consider targeting of the interceptor directly from the state vector and covariance matrix supplied by the space-based sensor, assumes that a cued weapon system will search for the threat missile, locate and track it, and provide the trajectory for the fire control solution. One of the metrics used to evaluate the effect of these critical parameters is the size of the defended area footprint. Conclusions drawn from this study support the proposition that the utilization of space-based sensors to cue a Navy Theater- Wide TBMD system can significantly increase the performance of the system.

Active tracking of a ballistic missile during the boost phase is a very challenging problem. The airborne laser (ABL) is one of several directed energy weapon programs that is interested in active track since the ABL design may use this technique. The Phillips Laboratory in response to this technical challenge has embarked on a project to verify the feasibility of active tracking over a long horizontal path through the atmosphere. The project is composed of two independent phases. The first phase is investigating tracking through a turbulent atmosphere using a scaled range with a static target at Lincoln Laboratory. The second phase of the project will demonstrate active tracking of boosting theater ballistic missiles using the SeaLite Beam Director at the High Energy Laser System Test Facility at White Sands Missile Range. This paper will present some of the tracking data and review the progress of the tests at both sites.

High fidelity simulation of active tracking systems requires the integration of optical, imaging, control and structural models for the payload with representative target and engagement models. The dependent relationship of actively illuminated target return on beam control system pointing commands particularly motivates an integrated simulation. Simulations of tracking and pointing systems have been developed to allow system algorithm refinement and performance prediction for several acquisition, tracking and pointing experiments. The simulation includes high fidelity, two axis control system models, 2D imaging sensor models, and 3D target geometry and reflectivity models. Key issues addressed include the effects of illuminator jitter coupling into the track error estimates, speckle effects, target reflectivity variations and control system interactions on residual pointing errors. The simulation has been implemented with commercial PC-class hardware and signal processing tools, using databased for specific target geometry and reflectivity maps as a function of engagement timelines. The simulation approach makes it particularly easy for control system and sensor system engineers to integrate discipline-specific models into a system simulation.

A naturally-occurring, enhanced backscatter appears whenever an object being obscured by a turbulent medium is actively illuminated and imaged by a monostatic transmitter/receiver. After making a double passage through the same turbulent eddies, reciprocal scattering paths, which encounter an identical phase delay, create a returning conjugate wave resulting in an enhanced illumination along the boresight of the telescope. Utilizing a dual aperture and orthogonal polarization to isolate the reciprocal paths, the backscatter enhancement occurs in the form of Young's interference fringes. For high visibility and stability of the fringes in the presence of time- varying turbulence, the width of an individual aperture is small compared to the atmospheric coherence diameter. With the separation and the width of the two apertures fixed and known, interferometric sensitivity of the displacement of objects was attained even when viewing through a turbulent atmosphere. Laboratory experimental data is compared with computer simulations and to analytical models. The results demonstrate the possibility of using this technique in a closed-loop pointing and tracking system, which would have potential applications in ground-to-space laser communications, laser power beaming to satellites and theater missile defense scenarios.

In the optimal guidance laws of proportional navigation, the problems are mostly formulated to obtain the optimal magnitude of applied commanded acceleration under pure or true proportional navigation. In this paper, the optimal direction of applied commanded acceleration is derived under conventional proportional navigation with a maneuvering or nonmaneuvering target, in which the line-of-sight deflection angle and the energy expenditure until intercept are minimized. Some related important characteristics, such as capture criterion and energy expenditure, are investigated and discussed in detail. Also, a typical example of target maneuver is introduced to illustrate the effect of target maneuver easily. It shows that the target maneuver will decrease the capture area and increase the energy cost for effective intercept of target.

This paper presents a case study where a comprehensive computer simulation is developed to determine the driving factors contributing to spacecraft pointing accuracy and stability. The simulation is implemented using XMATH/SystemBuild software from Integrated Systems, Inc. The paper is written in a tutorial manner and models for major system components are described. Among them are spacecraft bus, attitude controller, reaction wheel assembly, star-tracker unit, inertial reference unit, and gyro drift estimators (Kalman filter). THe predicted spacecraft performance is analyzed for a variety of input commands and system disturbances. The primary deterministic inputs are desired attitude angles and rate setpoints. The stochastic inputs include random torque disturbances acting on the spacecraft, random gyro bias noise, gyro random walk, and star-tracker noise. These inputs are varied over a wide range to determine their effects on pointing accuracy and stability. The results are presented in the form of trade-off curves designed to facilitate the proper selection of subsystems so that overall spacecraft pointing accuracy and stability requirements are met.

This paper presents a detailed description of the precision pointing system and the image stabilization system (ISS) for the Transition Region and Coronal Explorer (TRACE) satellite mission. The TRACE spacecraft is the fourth in NASA's small explorer series of missions and is scheduled for launch in September 1997. The primary TRACE science objective is to explore the relationship between the fine scale magnetic fields in the solar surface and features in the photosphere, chromosphere, transition region and corona. Quantitative images of these regions will be collected and used to study the structure and evolution of the sun's magnetic field with a spatial and temporal resolution of one arc-second and one second, respectively. TO meet the science objectives, the instrument payload and the spacecraft attitude control system are coupled using a guide telescope. The guide telescope provides both the targeting mechanism and pointing error signals for the spacecraft feedback control system. In addition, the guide telescope generates signals used to control the active mirror of the ISS. Simulation results show that precision target pointing is maintained to less than 5 arc-seconds, while analysis indicates that the ISS reduces image motion jitter below the 0.1 arc- second level.

NASA's Satellite Laser Ranging Network was originally developed during the 1970's to track satellites carrying corner cube reflectors. Today eight NASA systems, achieving millimeter ranging precision, are part of a global network of more than 40 stations that track 17 international satellites. To meet the tracking demands of a steadily growing satellite constellation within existing resources, NASA is embarking on a major automation program. While manpower on the current systems will be reduced to a single operator, the fully automated SLR2000 system is being designed to operate for months without human intervention. Because SLR2000 must be eyesafe and operate in daylight, tracking is often performed in a low probability of detection and high noise environment. The goal is to automatically select the satellite, setup the tracking and ranging hardware, verify acquisition, and close the tracking loop to optimize data yield. TO accomplish the autotracking tasks, we are investigating (1) improved satellite force models, (2) more frequent updates of orbital ephemerides, (3) lunar laser ranging data processing techniques to distinguish satellite returns from noise, and (4) angular detection and search techniques to acquire the satellite. A Monte Carlo simulator has been developed to allow optimization of the autotracking algorithms by modeling the relevant system errors and then checking performance against system truth. A combination of simulator and preliminary field results will be presented.

In the past decade, simulation has played a key role in order to sustain the development of weapon systems at AEROSPATIALE MISSILES. To achieve this goal, a fully object-oriented framework has been developed by a multi-disciplinary team. This framework is used at every level of the life cycle of the products (feasibility, specification, development and validation). In particular, this approach has been applied to the ROLAND weapon system, a multi-sensor ground- air weapon system including radar, laser and IR technology. According to the needs, different parts of the systems can be activated by the different members of the team for particular studies using the strength of the object oriented approach. The simulation has already been used in development case to study a kind of single IR tracking, to share performance allowances between components (sensor and signal processing hardware for example), to validate the stability of the tracking with multi-band information as well as to study future evolutions of the system using different data fusion configurations.

The LPMA (limb profile monitor of the atmosphere) instrument is a Fourier transform infrared (FTIR) spectrometer of the BOMEM DA2.01 type. It is used to perform balloon-borne remote sensing measurements of stratospheric (and in some case tropospheric) trace species. As an instrument working in absorption against the sun it is accommodated on a stratospheric gondola with pointing capabilities. The primary pointing system is used to put (and to maintain) the sun into the field of view of a sun-tracker (heliostat) acting as a fine pointing system to compensate for the residual motions of the gondola. This configuration allows to feed in a stable manner the infrared solar radiation along the optical axis of the spectrometer (through a germanium window used to reject the unused visible-UV flux). The complete system i.e. gondola, pointing-system, heliostat, FTIR spectrometer, on board processing and recording as well as telemetry unit has flown several times (from Aire-sur- l'Adour in France, from ESRANGE in Sweden and from Leon in Spain) under stratospheric balloons operated by CNES. A description of the various subsystems will be presented together with samples of representative ground tests and in flight measurements.

An important function for a fire control radar consists of the simultaneous tracking of multiple objects in a missile complex. A challenging tracking scenario is encountered when dense clusters of targets are viewed at large stand-off ranges. In this case, multiple target detections are found within a single radar beam. Because such highly populated complexes are not resolved in angle (azimuth, elevation), separation of the individual objects must rely on the use of wideband waveforms with good range resolution. A multi-object tracking algorithm suitable for real-time implementation has been designed to exploit range-only correlations. Good track initiation efficiency is achieved through the use of short radar observation times combined with applications of Hough transforms. Tracking filters are subsequently used to update all existing tracks with metrics and signature histories. A brief description of the algorithm is presented with focus on the specific tracking techniques appropriate for dense clusters of slowly separating objects on ballistic trajectories. The performance of the tracking algorithm has been evaluated and results are presented for several tracking strategies and different radar sampling rates.

A simulation program is presented for studying the problem of tracking a low- flying target over the sea with an amplitude comparison monopulse radar. The simulation models the essential physical and logical processes involved in the tracking environment, which include specular and diffuse reflections from a spherical sea surface, target RCS scintillation, channel noise. Realistic monopulse angle measurements are generated by means of a four-lobe antenna model. The simulator allows easy substitution of the default radar-pointing and tracking algorithms by alternatives that the user may construct. Target trajectories and radar size, height, and frequency can be selected to study of the relative merits of various trade-offs. A common application is to put the simulator under the control of a driver that carries out multiple Monte Carlo experiments to test different parameter sets.

When tracking a manned aircraft with a phase array radar, detecting a missile launch (i.e., a target split) is particularly important because the missile can have a very small radar cross section (RCS) and drop below the horizon of the radar shortly after launch. Reliable detection of the launch is made difficult because the RCS of the missile is very small compared to that of the manned aircraft and the radar typically revisits a manned aircraft every few seconds. Furthermore, any measurements of the aircraft and missile taken shortly after the launch will be merged until the two targets are resolved in range, frequency, or space. In this paper, detection of the launched missile is addressed through the detection of the presence of target multiplicity with the in-phase and quadrature monopulse measurements. The probability of detecting the launch using monopulse processing will be studied with regard to the tracking signal-to-noise ratio and the number of pulses n the radar waveform.

There has been increasing interest in the extraction of data from the frequency (doppler) domain for such purposes as target identification and kill assessment. An initial aspect of this process is to determine if there are signals in the doppler domain. There have been a number of techniques developed for signal detection in the time domain including various constant false alarm rate (CFAR) algorithms, but not many techniques have been translated or developed for the frequency domain. One reason for this lack of techniques results from an inability to accurately classify the statistics of the noise found within given regions of interest. Using the central limit theorem and some reasonable assumptions, it has been possible to develop two techniques that can be utilized as adaptive CFAR algorithms in the frequency domain.

An algorithm is presented for absolutely aligning a sensor that measures range, bearing, and elevation. The alignment (or bias) errors are determined in this algorithm by comparing the sensor-reported positions for a target to an absolute or true standard. For example, the global positioning system (GPS) could be used as an absolute standard because it provides positions that closely approximate truth. A possible technique for absolute alignment using GPS is described in this paper.

The heart of any tracking system is its data association algorithm where measurements, received as sensor returns, are assigned to a track, or rejected as clutter. In this paper, we investigate the use of genetic algorithms (GA) for the multiple target tracking data association problem. GA are search methods based on the mechanics of natural selection and genetics. They have been proven theoretically and empirically robust in complex space searches by the founder J. H. Holland. Contrary to most optimization techniques, which seek to improve performance toward the optimum, GA find near-optimal solutions through parallel searches in the solution space. We propose to optimize a simplified version of the neural energy function proposed by Sengupta and Iltis in their network implementation of the joint probability data association. We follow an identical approach by first implementing a GA for the travelling salesperson problem based on Hopfield and Tank's neural network research.

A neural based tracking scheme is proposed for the problem of direction-of- arrivals tracking using an adaptive array. Simulation results show that the proposed technique is efficient for mobile communication tracking applications.

The vast majority of gimbaled line-of-sight stabilization systems involve inertial angular rate sensors located on the gimbal assembly. These sensors are mounted to follow the line-of-sight as the gimbals move, thus allowing them to serve as direct feedback elements for the stabilization rate-control loops. For systems in which on-gimbal configurations are not possible, feedforward control configurations using strapdown inertial sensors are required. The gimbal control commands in these systems are calculated in real- time form the inertial sensor outputs and the transformation matrices between the sensor, gimbal, and line-of-sight reference frames. A testbed for the development and test of a feedforward line-of-sight stabilization systems is described. The system includes a three-axis, mirror-stabilized gimbal, a servoamplifier unit, and inertial measurement system, a digital controller, and a personal computer for software development.

For various reasons small spacecraft are becoming more appealing. Because of their smaller inertias, these spacecraft are more sensitive to disturbances and likely to have more attitude jitter than the bigger units. These jitter levels are unacceptable for some scientific instruments and need to be compensated. In the case of line-of-sight type instruments, the attitude jitter can be mitigated by incorporating a fast steering mirror into the system. To take full advantage of these devices, the spacecraft attitude needs to be measured at sufficiently high bandwidth, well beyond what is commonly provided by inertial reference units. This paper explores various ways to obtain higher bandwidth attitude measurements for the purpose of jitter control, and provides a practical solution to the problem.

The ability to compensate for disturbances resulting form nonlinear phenomena such as Coulomb friction in inertial stabilization systems has been demonstrated to be feasible when the plant dynamics of the system are accurately modeled. However, for cases where the plant in unknown or changing, a self-tuning control algorithm is desired to prevent instability. This paper formulates such a self-tuning control algorithm with specific application to systems with inherent nonlinearities. The ability of the algorithm to self- tune and compensate for nonlinear induced disturbances is demonstrated for an inertial stabilization gimbal control system with Coulomb type bearing friction.

A frequent problem in inertial stabilization control systems is the rejection of disturbances associated with moving components. Very often such disturbances are nonlinear and time varying. A prime example is the relative motion of components within a gimbal; in this case, nonlinear bearing friction induces a de-stabilizing torque from base motion to the component being stabilized. This paper presents an LQG algorithm, based on a simple first order linear stochastic differential equation, for estimating and compensating in real time a particular class of disturbances that can be modeled as a plus or minus unknown slowly changing random value such as is characterized by nonlinear Coulomb friction. Results of computer simulations testing the control algorithm are presented along with actual measurements from a laboratory brassboard system. The results reveal a noteworthy improvement in disturbance rejection as compared with a conventional PI controller with notch filters.

The problem of active control in linear elastic structures with piezoceramic actuators has been investigated using both a theoretical and experimental approach. An active structure consisting of a simple beam element with a bonded piezoceramic actuator and a strain gage as sensor has been made. To model the active estructure for maximum increase in system damping, a simple finite element model has been taken from the classic literature. The model incorporates the electromechanical coupling of the transducers as well as bonding effects. The model control has been experimentally demonstrated by conditioning the output of the sensor with a data acquisition system and a digital signal processing that implements adjustable bandpass filtering, fast Fourier transform for real time spectrum estimation and the cancellation algorithm that supplies the voltage reference to the boost circuit that provides the high voltage level required by the piezoactuator. The experiment primarily consisted of investigating the behavior and performance of the individual unimorph transducer, analysis of the active estructure behavior and searching the best cancellation strategy. Focus has been pointed on some special characteristics of piezoelectric elements, such as their dynamic behavior, as well as the digital signal processing for fast modal frequencies identification avoiding the pitfalls of aliasing and spectral leakage by means of the design of a real multirate system. The active damping augmentation in the low frequency range of flexible beam-shaped structures makes it suitable for placement of electro-optical devices where high-precision pointing is an essential feature. The time reduction to stabilize the structure, that is total vibration cancellation, demonstrates that the piezotechnology is a powerful means to increase the performance of a mechanical structure.

A vehicle real-time attitude estimation systems (VRAES) is described which uses multiple optical sensors to detect and track moving targets. The pointing and shape information from each optical sensor is combined in real-time to estimate the 3-space location and attitude (orientation) of the target. This paper describes the system architecture, the algorithm,a nd the real-time implementation. Test results indicate that the system can operate in real- time, on a wide range of data and provide precise estimates of position and orientation.

Design optimization of airborne IR target acquisition and tracking systems is a complex process involving tradeoffs between program goals, systems performance, cost, physical characteristics, reliability, growth adaptability and numerous other factors. The key performance goals are usually detection range, search area, search time and false alarm rate. Constraints on the design are available physical volume and the state of the art of IR detectors, inertial sensors, optics, signal processors, and recorders. Funding constraints often restrict the design to use off the shelf hardware. Design objectives for an airborne sea surface target IR tracking system are defined. The system design process is described and several passes through this process are discussed. An optimized custom design resulting from the first pass through design process with relaxed constraints on the problem and an 'off the shelf' design to meet the design objectives are defined and discussed.

A novel method for a multitarget detection and measurement system based on the coordination of objective and subjective estimation (COSE) is described. This system is composed of three components in cascade. The first stage implements available objective knowledge, namely target position and velocity, to form a subset of statistically likely target solutions via validation. The second stage of the system utilizes available heuristic (subjective) information, namely return area and brightness, to form a smaller subset of returns. The returns in this subset are similar to the target in terms of area and brightness. If only one return is available in this subset, it is chosen as the true target. If more than one return is available in this similar subset,the third state implements a minimum azimuth angle criterion, in which the return yielding the smallest azimuth angle between itself and the target's predicted position vector is selected as the true target. The causes and effects of clutter returns are described. Simulated target tracks are performed and discussed. Finally, the COSE algorithm is applied to a sequence of forward-looking infrared images provided by Texas Instruments, Inc. The results obtained will be subjectively compared to results of other tracking algorithms.

Several designs of Kalman filters and the interacting multiple models algorithm were used in real tracking tasks involving high dynamic targets. The data were obtained through the joint effort of the defense departments of Canada and the US. Their performance, measured in terms of positional deviation and the number of track losses, are rather unsatisfactory even though they perform particularly well when using simulated data. To identify the reasons behind, we compared and analyzed the differences between the model assumptions behind the design of these Kalman filters and the model required for accurate tracking of these targets. In this paper, we discussed our findings. Moreover, based on the characteristics of real tracking data, we present an alternative methodology for measuring the effectiveness of various Kalman filter based trackers in stressful environmental. It can also be used to explain the well known characteristics of Kalman filter. A lower bound for the deviation, obtained from this equation, shows that deviation could be too large to manage if noise bandwidth is as high as the real data instead of a pre-assumed magnitude. Instead of having to redesign many existing Kalman filters to suit for stressful environment, we developed a design-independent module that can be added to different types of Kalman filters based trackers to enhance their performance in the tracking high dynamic targets. The module is called adaptive systems noise covariance estimation. It is not only safe (i.e. almost no negative effect) but it can sometimes even double the performance of trackers in stressful environment.

In this paper, a new method for image stabilization with a three-axis image- stabilizing reflecting prism assembly is presented, and the principle of image stabilization in this prism assembly, formulae for image stabilization and working formulae with an approximation up to the third power are given in detail. In this image-stabilizing system, a single chip microcomputer is used to calculate value of compensating angles and thus to control the prism assembly. Two gyroscopes act as sensors from which information of angular perturbation is obtained, three stepping motors drive the prism assembly to compensate for the movement of image produced by angular perturbation. The image-stabilizing device so established is a multifold system which involves optics, mechanics, electronics and computer.

We have developed a robotic unmanned ground vehicle (UGV) that performs reconnaissance, surveillance, and target acquisition. THis vehicle has been used in a number of tactical training exercises at Fort Hood, TX with US Army scouts from the 1st Armored Cavalry Division. The UGV, built around a high- mobility, multi-purpose, wheeled vehicle, is designed to be supervised by an operator at a control station located 10 km or more away from the UGV. The UGV's real-time automatic target acquisition system uses an infrared sensor to automatically detect and track moving ground vehicles out to a range of 5 km. When commanded by the operator, the UGV will engage a particular target with a laser designator. The ATA system is the topic of this paper. We describe the requirements of the ATA system, the algorithms used, their implementation, and the system's performance.

Tank main gun systems must possess extremely high levels of accuracy to perform successfully in battle. Under some circumstances, the first round fired in an engagement may miss the intended target, and it becomes necessary to rapidly correct fire. A breadboard automatic miss-distance indicator system was previously developed to assist in this process. The system, which would be mounted on a 'wingman' tank, consists of a charged-coupled device (CCD) camera and computer-based image-processing system, coupled with a separate infrared sensor to detect muzzle flash. For the system to be successfully employed with current generation tanks, it must be reliable, be relatively low cost, and respond rapidly maintaining current firing rates. Recently, the original indicator system was developed further in an effort to assist in achieving these goals. Efforts have focused primarily upon enhanced image-processing algorithms, both to improve system reliability and to reduce processing requirements. Intelligent application of newly refined trajectory models has permitted examination of reduced areas of interest and enhanced rejection of false alarms, significantly improving system performance.

In a real world application an image from a stabilized sensor on a moving platform will not be 100 percent stabilized. There will always be a small unknown error in the stabilization due to factors such as dynamic deformations in the structure between sensor and reference Inertial Navigation Unit, servo inaccuracies, etc. For a high resolution imaging sensor this stabilization error causes the image to move several pixels in unknown direction between frames. TO be able to detect and track small moving objects from such a sensor, this unknown movement of the sensor image must be estimated. An algorithm that searches for land contours in the image has been evaluated. The algorithm searches for high contrast points distributed over the whole image. As long as moving objects in the scene only cover a small area of the scene, most of the points are located on solid ground. By matching the list of points from frame to frame, the movement of the image due to stabilization errors can be estimated and compensated. The point list is searched for points with diverging movement from the estimated stabilization error. These points are then assumed to be located on moving objects. Points assumed to be located on moving objects are gradually exchanged with new points located in the same area. Most of the processing is performed on the list of points and not on the complete image. The algorithm is therefore very fast and well suited for real time implementation. The algorithm has been tested on images from an experimental IR scanner. Stabilization errors were added artificially to the image such that the output from the algorithm could be compared with the artificially added stabilization errors.

An algorithm for the rapid identification of star patterns is presented. The paradigm utilizes a novel star selection technique and a new binary tree- search technique to identify star configurations from large catalogues in a minimal time with low data storage requirements. Performance claims are supported by test data and identification results from a ground-based experimental star camera, the system consistently demonstrating a capability to identify numbers of stars in a single star camera image. Features include a search time scaling to the logarithm of the number of stars in the identification set, a reduced dependence on star magnitude data, and a method of accounting for camera measurement inaccuracy. The method is suitable for in-space application and offers the possibility of complete sky identification coverage using high accuracy star cameras. The approach produces a reliable, orientation independent star identification faster than conventional serial techniques and adaptable to many types of mission and application.

This paper describes a fast algorithm that tracks face region in dynamic images. We adopt the color histogram backprojection algorithm of swain-Ballard in order to find face region in subsequent frames. The algorithm has the advantages of being invariant to object rotation, small deformation, and noise. The following problems are addressed: building a model histogram, backprojecting model histogram, searching the peak response location, and updating the model histogram. The experimentation shows the possibility of realtime face tracking on PC or low-performance workstation.

Imaging using optical parametric amplification (OPA) and optical parametric up-conversion has been performed since 1960s. Various problems associated with the pump laser, such as beam quality, size, weight, and power consumption, and optical quality of crystals have limited the utility of these techniques. While in recent times, new as well as good optical quality crystals with high damage resistance are being frown, there is a need to incorporate efficient laser pump configurations for developing field deployable imaging systems. Based on the development of a highly efficient and very compact semiconductor diode pumped OPO, the present paper discusses the design concepts of a self- pumped optical parametric amplifier and mixer for imaging using mid-IR wavelengths.

In this paper, we describe a robot endeffector tracking system using sensory information from recently-announced structured pattern laser diodes, which can generate images with several different types of structured pattern. The neural network approach is employed to recognize the robot endeffector covering the situation of three types of motion: translation, scaling and rotation. Features for the neural network to detect the position of the endeffector are extracted from the preprocessed images. Artificial neural networks are used to store models and to match with unknown input features recognizing the position of the robot endeffector. Since a minimal number of samples are used for different directions of the robot endeffector in the system, an artificial neural network with the generalization capability can be utilized for unknown input features. A feedforward neural network with the generalization capability can be utilized for unknown input features. A feedforward neural network trained with the back propagation learning is used to detect the position of the robot endeffector. Another feedforward neural network module is used to estimate the motion from a sequence of images and to control movements of the robot endeffector. COmbining the tow neural networks for recognizing the robot endeffector and estimating the motion with the preprocessing stage, the whole system keeps tracking of the robot endeffector effectively.

The aim of the present paper is to report preliminary results o : development of the image post-processing techniques, called Multiple Object DEconvolution (MODE) and Super Blind MODE (SBMODE), which are intended for joint restoration of the specified number of consec .tive distorted images of a highly-evolving unknown extended object being observed in the presence of unkno: 'n phase distortions (PSF). The solution to the problem is treated in the context of ir:oherent shift-invariant imaging and it is conceptually based on the generalized projecting onto convex ss (POCS), requiring only the qualitative knowledge (isoplanaticity, linearity, finiteness, positivity). Mathematically, the stated task is reduced to solving a class of inverse optical problems being described in form of a system of M (M 2) convolution-type Fredhoim integr ti equations containing a common unknown kernel (PSF). All the derived versions of the MODE are self-sufficient and Lhey do recover fairly reliably the blurred images for general compact objects in a relatively small numb. r of iterations. Brief theoretical analysis establishes uniqueness of a feasible solution and its close proxin iity to a true one under tolerance of poor (arbitrary) starting guesses. This versatile technique of few operators gives encouraging re e ults ,providing a basis for accurate imagebased ATP, target 3D-shaping and behavioural classification in n ax-real-time. The (SB)MODE technique is open to the imagery application ;in astronomy and defense plus biomedical imaging, including microscopies. Keywords: nonstationary space object; imaging telescope; imag sequence recovery; deconvolution; projection onto convex sets

Based on angle measurement data obtained by a monopulse tracking radar tracking the aircraft target in the short range, the statistical characteristics of the angle measurement data are analyzed. Moreover, the important influences and the proportion of the target glint in the entire measurement errors are evaluated in this paper. THen, a new method to suppress the angle measurement errors is developed based on the wavelet multiresolution analysis and the statistical analysis of the aircraft glint. Meanwhile, the explanation of the effect that glint can be suppressed by wavelet processing is developed in this paper. The results show that the precision of angle measurement can be improved about 30 percent. FInally, it is showed that the wavelet package algorithm can be used as a preprocessor of the Kalman filtering.

The complex of requirements providing successful usage of star trackers of the space vehicles, on the one hand, and the principles of design and technical implementation which allow to meet the requirements, on the other hand, are elaborated during developing of technique of star trackers for space vehicle attitude control. The main characteristics of star trackers are: accuracy, interference protection, capability of the information updating, dimensions and mass, stability under external impacts. The design principles include, in the first place, the proper choice of the main components (optics, photo detector, processing circuits) as well as the methods of the information processing in the device. The signal source and noise models are of great importance at the case. The approaches, which provide achievement of high technical characteristics of star trackers and which are the traditional ones at the development of novel instruments in 'GeoKos', are discovered in the report as well as new opportunities of the approaches implementation available at present. Description and the characteristics of new generation of the unified start trackers being developed in 'GeoKos' are presented.

This report describes some results of novel developments, that are provided in 'GeoKos', on the precise sun sensor with wide field of view. This device provides wide field of view and high accuracy by means of using CCD-array and special slit-raster. This raster reduces the errors components, arising from CCD features, such as discrete structure, pixels nonuniformity etc. High reliability, small dimensions and mass of this device are determined by simplicity of the optics, minimum of electronics and absence of moving components.

The main feature of a large ten-meters telescope is application of 84 segment primary mirrors which are united in one reflecting system. This decision makes the problem of the primary mirror designing easier yet brings another task: the designing of every mirror segment high precision (nanometer accuracy) drive, which provides perfect coincidence of segment mirrors constantly. There are suggested two variants of the drive construction based on magnetic rheology liquid. This drive provides nanometer precision and millisecond quick action.

One of the basic claims for new-developed opto-electronic devices intended for attitude control of space vehicles is the mass and dimensions minimizing with preservation of high accuracy and reliability. Novel Earth horizon sensor for the 2-axis oriented vehicles is developed, based on 4 pyroelectric arrays, that are obturated by the 4-slits modulator. Such sensor can be used for precision measuring of the satellite attitude angles. The sensor is protected against the Sun and the Moon interference. It can be modified for any orbit altitude.