SEARCH · Search NASA
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
In image matching applications such as tracking and stereo matching, it is common to use the sum-of-squared-differences (SSD) measure to determine the best match for an image template.
Titan’s dense atmosphere, low gravity, and high winds at high altitudes create descent times of >90 minutes with standard entry/descent/landing (EDL) architectures and result in large unguided landing ellipses, with 99% values of 110x110 km and 149x72 km in recent Titan lander proposals. Enabling precision landing on Titan could increase science return for the types of missions proposed to date and make additional types of landing sites accessible, opening up new possibilities for science investigations. Precision landing on Titan has unique challenges, because the hazy atmosphere makes it difficult to see the surface and because it requires guided descent with divert ranges that are one to two orders of magnitude larger than needed for other target bodies, i.e. up to on the order of 100 km. It is conceivable that such a divert capability could be provided economically by a parafoil or other steerable aerodynamic decelerator deployed several 10s of km above the surface. The long descent times lead to large inertial navigation errors, hence a need for terrain relative navigation (TRN). This would require a TRN capability that can operate at such altitudes, despite challenges of seeing the surface sufficiently clearly and of depending on map products that are two orders of magnitude lower in spatial resolution than those for Mars and airless bodies. We then develop algorithms for map matching and feature tracking with descent images and test these with synthetic images created from Cassini/Huygens data sets and our radiative transfer model. We also introduce new possibilities for TRN based on the potential to discriminate some specific types of terrain onboard in descent imagery, such as lake vs adjacent ground and dune vs interdune. We use sensor measurement noise models in simulations of state estimation with an extended Kalman filter that includes coordinates of a set of tracked features in the state vector. Case studies were done for two notional landing sites, one in a site with only dry ground and one in a Titan lake district. In both cases, the filter error model shows 3 position error at touchdown on the order of 2 km. More work is needed to validate these results with higher fidelity camera models and larger data sets, but this is very promising.
Since the Voyager encounters in 1979, it has been known that Jupiter's cloud-top zonal winds violate the barotropic stability criterion. A vortex-tube stretching analysis of the Voyager wind data indicates that the more general Charney-Stern stability criterion is also violated. On the other hand, the zonal winds determined by tracking cloud features in Hubble Space Telescope images taken in 1991 precisely match the zonal winds determined by tracking cloud features in Voyager images, and it is hard to understand how a complicated zonal wind profile like Jupiter's could be unstable and yet not change at all in 12 years. In fact, there are at least two unknown ways to violate the Charney-Stern stability criterion and still have a stable flow. The better known of these is called Fjortoft's theorem, or Arnol'd's 1st theorem for the case of large-amplitude perturbations. Although the Fjortoft-Arnol'd theorem has been extended from the quasi-geostrophic equations to the primitive equations, the basic requirement that the potential vorticity be an increasing function of streamfunction is opposite to the case found in Jupiter, where the Voyager data indicate that the potential vorticity is a decreasing function of streamfunction. But this second case is precisely that which is covered by Arnol'd's 2nd stability theorem. In fact, the Voyager data suggest that Jupiter's zonal winds are neutrally stable with respect to Arnol'd's 2nd stability theorem. Here, we analyze the linear stability problem of a one-parameter family of sinusoidal zonal wind profiles that are close to neutral stability with respect to Arnol'd's 2nd stability theorem. We find numerically that the most unstable mode is always stationary, which may help to explain the slowly moving mode 10 waves observed on Jupiter. We find that violation of Arnol'd's 2nd stability theorem is both necessary and sufficient for instability of sinusoidal profiles. However, there appears to be no simple extension of Arnol'd's 2nd stability theorem to the primitive equations. Nevertheless, the primitive growth rates are small, and the primitive system is still governed by the quasi-geostrophic neutral-stability configuration.
Abstract—The autonomous approach of a spacecraft to an asteroid or comet (a small body) relies heavily on visual feature tracking to aid in estimating relative trajectories and the properties of the small body. Feature tracking for small bodies brings several challenges, including changing lighting, poor visual texture, and a concentration of features in a small part of an image. Six existing, open-source algorithms for feature tracking were tested on a simulated dataset and compared to the ground truth in the path of features. The main finding is that none of the algorithms provide all of the desired characteristics of long feature tracks with low errors and few outliers. Instead, there is a trade-off between long feature tracks and low error. The feature-matching algorithms SIFT, and BRISK provide good error characteristics, but short feature tracks, whereas the optical flow algorithm KLT provides long feature tracks, but with many features of large error. Given the challenges in feature tracking, it is recommended to focus development on each component of a feature tracking system: detection, description, and outlier rejection.
A system of electronic hardware and software, now undergoing development, automatically estimates the location of a robotic land vehicle in an urban environment using a somewhat imprecise map, which has been generated in advance from aerial imagery. This system does not utilize the Global Positioning System and does not include any odometry, inertial measurement units, or any other sensors except a stereoscopic pair of black-and-white digital video cameras mounted on the vehicle. Of course, the system also includes a computer running software that processes the video image data. The software consists mostly of three components corresponding to the three major image-data-processing functions: Visual Odometry This component automatically tracks point features in the imagery and computes the relative motion of the cameras between sequential image frames. This component incorporates a modified version of a visual-odometry algorithm originally published in 1989. The algorithm selects point features, performs multiresolution area-correlation computations to match the features in stereoscopic images, tracks the features through the sequence of images, and uses the tracking results to estimate the six-degree-of-freedom motion of the camera between consecutive stereoscopic pairs of images (see figure). Urban Feature Detection and Ranging Using the same data as those processed by the visual-odometry component, this component strives to determine the three-dimensional (3D) coordinates of vertical and horizontal lines that are likely to be parts of, or close to, the exterior surfaces of buildings. The basic sequence of processes performed by this component is the following: 1. An edge-detection algorithm is applied, yielding a set of linked lists of edge pixels, a horizontal-gradient image, and a vertical-gradient image. 2. Straight-line segments of edges are extracted from the linked lists generated in step 1. Any straight-line segments longer than an arbitrary threshold (e.g., 30 pixels) are assumed to belong to buildings or other artificial objects. 3. A gradient-filter algorithm is used to test straight-line segments longer than the threshold to determine whether they represent edges of natural or artificial objects. In somewhat oversimplified terms, the test is based on the assumption that the gradient of image intensity varies little along a segment that represents the edge of an artificial object.
The Mars Exploration Rovers (MER'03), Spirit and Opportunity, represent the state of the art in rover operations on Mars. This paper presents validation experiments of different visual tracking algorithms using the rover's navigation camera.
We present observations of Saturn's central flash obtained from Palomar and McDonald Observatories during the 3 July 1989 occultation of 28 Sgr. As the star passed close to the geometric center of Saturn's shadow, the focusing of the incident starlight by the planet's atmosphere formed multiple stellar images along the limb which were detected in infrared images obtained at wavelengths of 3.9 (Palomar) and 2.1 microns (McDonald). These are the first reported observations of a central flash due to Saturn, and the first of any planet in which the signal flash from each stellar image could be determined separately, permitting a comparison of both intensity and position for each image with model predictions. Four separate flashes were observed from each observatory, corresponding to points on the limb where the starlight passed through the Cassini Division and the relatively transparent C Ring, with maximum brightness reaching 1-2% of the unocculted stellar intensity. The timing of the flashes is quite sensitive to the shape of Saturn's limb, which depends in turn on the planet's zonal gravity harmonics and on the zonal wind profile in the lower stratosphere, near the 2.5-mbar pressure level. The locations of the images along the limb, as well as the timing, shapes, and amplitudes of the individual flash light curves, are well matched by a smoothed model based on a tropospheric zonal wind profile obtained from tracking cloud features in Voyager images and the Saturn ring optical depth profile obtained from the Voyager Photopolarimeter experiment. The smoothing required to give the best match to the data exceeds that attributable to the finite angular extent of the occulted star and may be due to refractive scattering by turbulence or wave structure in Saturn's atmosphere. There is no evidence of significant atmospheric absorption at the observed wavelengths, which correspond to spectral regions of weak methane absorption.
Feature detectors have been considered for the role of supplying additional information to a neural network tracker. The feature detector focuses on areas of the image with significant information. Basically, if a picture says a thousand words, the feature detectors are looking for the key phrases (keypoints). These keypoints are rotationally invariant and may be matched across frames. Application of these advanced feature detectors to the neural network tracking system at JPL has promising potential. As part of an ongoing program, an advanced feature detector was tested for augmentation of a neural network based tracker. The advance feature detector extended tracking periods in test sequences including aircraft tracking, rover tracking, and simulated Martian landing. Future directions of research are also discussed.
We have developed a combined feature based and shape based visual tracking system designed to enable a planetary rover to visually track and servo to specific points chosen by a user with centimeter precision. The feature based tracker uses invariant feature detection and matching across a stereo pair, as well as matching pairs before and after robot movement in order to compute an incremental 6-DOF motion at each tracker update. This tracking method is subject to drift over time, which can be compensated by the shape based method. The shape based tracking method consists of 3D model registration, which recovers 6-DOF motion given sufficient shape and proper initialization. By integrating complementary algorithms, the combined tracker leverages the efficiency and robustness of feature based methods with the precision and accuracy of model registration. In this paper, we present the algorithms and their integration into a combined visual tracking system.
The field of computer vision is surveyed and assessed, key research issues are identified, and possibilities for a future vision system are discussed. The problems of descriptions of two and three dimensional worlds are discussed. The representation of such features as texture, edges, curves, and corners are detailed. Recognition methods are described in which cross correlation coefficients are maximized or numerical values for a set of features are measured. Object tracking is discussed in terms of the robust matching algorithms that must be devised. Stereo vision, camera control and calibration, and the hardware and systems architecture are discussed.
This software finds feature point correspondences in sequences of images. It is designed for feature matching in aerial imagery. Feature matching is a fundamental step in a number of important image processing operations: calibrating the cameras in a camera array, stabilizing images in aerial movies, geo-registration of images, and generating high-fidelity surface maps from aerial movies. The method uses a Shi-Tomasi corner detector and normalized cross-correlation. This process is likely to result in the production of some mismatches. The feature set is cleaned up using the assumption that there is a large planar patch visible in both images. At high altitude, this assumption is often reasonable. A mathematical transformation, called an homography, is developed that allows us to predict the position in image 2 of any point on the plane in image 1. Any feature pair that is inconsistent with the homography is thrown out. The output of the process is a set of feature pairs, and the homography. The algorithms in this innovation are well known, but the new implementation improves the process in several ways. It runs in real-time at 2 Hz on 64-megapixel imagery. The new Shi-Tomasi corner detector tries to produce the requested number of features by automatically adjusting the minimum distance between found features. The homography-finding code now uses an implementation of the RANSAC algorithm that adjusts the number of iterations automatically to achieve a pre-set probability of missing a set of inliers. The new interface allows the caller to pass in a set of predetermined points in one of the images. This allows the ability to track the same set of points through multiple frames.
A visual-tracker computer program controls an articulated mast on a Mars rover to keep a designated feature (a target) in view while the rover drives toward the target, avoiding obstacles. Several prior visual-tracker programs have been tested on rover platforms; most require very small and well-estimated motion between consecutive image frames a requirement that is not realistic for a rover on rough terrain. The present visual-tracker program is designed to handle large image motions that lead to significant changes in feature geometry and photometry between frames. When a point is selected in one of the images acquired from stereoscopic cameras on the mast, a stereo triangulation algorithm computes a three-dimensional (3D) location for the target. As the rover moves, its body-mounted cameras feed images to a visual-odometry algorithm, which tracks two-dimensional (2D) corner features and computes their old and new 3D locations. The algorithm rejects points, the 3D motions of which are inconsistent with a rigid-world constraint, and then computes the apparent change in the rover pose (i.e., translation and rotation). The mast pan and tilt angles needed to keep the target centered in the field-of-view of the cameras (thereby minimizing the area over which the 2D-tracking algorithm must operate) are computed from the estimated change in the rover pose, the 3D position of the target feature, and a model of kinematics of the mast. If the motion between the consecutive frames is still large (i.e., 3D tracking was unsuccessful), an adaptive view-based matching technique is applied to the new image. This technique uses correlation-based template matching, in which a feature template is scaled by the ratio between the depth in the original template and the depth of pixels in the new image. This is repeated over the entire search window and the best correlation results indicate the appropriate match. The program could be a core for building application programs for systems that require coordination of vision and robotic motion.
A novel computational framework was developed of a 2D affine invariant matching exploiting a parameter space. Named as affine invariant parameter space (AIPS), the technique can be applied to many image-processing and computer-vision problems, including image registration, template matching, and object tracking from image sequence. The AIPS is formed by the parameters in an affine combination of a set of feature points in the image plane. In cases where the entire image can be assumed to have undergone a single affine transformation, the new AIPS match metric and matching framework becomes very effective (compared with the state-of-the-art methods at the time of this reporting). No knowledge about scaling or any other transformation parameters need to be known a priori to apply the AIPS framework. An automated suite of software tools has been created to provide accurate image segmentation (for data cleaning) and high-quality 2D image and 3D surface registration (for fusing multi-resolution terrain, image, and map data). These tools are capable of supporting existing GIS toolkits already in the marketplace, and will also be usable in a stand-alone fashion. The toolkit applies novel algorithmic approaches for image segmentation, feature extraction, and registration of 2D imagery and 3D surface data, which supports first-pass, batched, fully automatic feature extraction (for segmentation), and registration. A hierarchical and adaptive approach is taken for achieving automatic feature extraction, segmentation, and registration. Surface registration is the process of aligning two (or more) data sets to a common coordinate system, during which the transformation between their different coordinate systems is determined. Also developed here are a novel, volumetric surface modeling and compression technique that provide both quality-guaranteed mesh surface approximations and compaction of the model sizes by efficiently coding the geometry and connectivity/topology components of the generated models. The highly efficient triangular mesh compression compacts the connectivity information at the rate of 1.5-4 bits per vertex (on average for triangle meshes), while reducing the 3D geometry by 40-50 percent. Finally, taking into consideration the characteristics of 3D terrain data, and using the innovative, regularized binary decomposition mesh modeling, a multistage, pattern-drive modeling, and compression technique has been developed to provide an effective framework for compressing digital elevation model (DEM) surfaces, high-resolution aerial imagery, and other types of NASA data.
In this paper, we describe a novel approach in fusing optical flow with inertial cues (3D acceleration and 3D angular velocities) in order to navigate a Micro Aerial Vehicle (MAV) drift free in 4DoF and metric velocity. Our approach only requires two consecutive images with a minimum of three feature matches. It does not require any (point) map nor any type of feature history. Thus it is an inherently failsafe approach that is immune to map and feature-track failures. With these minimal requirements we show in real experiments that the system is able to navigate drift free in all angles including yaw, in one metric position axis, and in 3D metric velocity. Furthermore, it is a power-on-and-go system able to online self-calibrate the inertial biases, the visual scale and the full 6DoF extrinsic transformation parameters between camera and IMU.
The NASA Stardust spacecraft exposed an aerogel collector to the interstellar dust passing through the solar system. We performed X-ray fluorescence element mapping and abundance measurements, for elements 19 < or = Z < or = 30, on six "interstellar candidates," potential interstellar impacts identified by Stardust@Home and extracted for analyses in picokeystones. One, I1044,3,33, showed no element hot-spots within the designated search area. However, we identified a nearby surface feature, consistent with the impact of a weak, high-speed particle having an approximately chondritic (CI) element abundance pattern, except for factor-of-ten enrichments in K and Zn and an S depletion. This hot-spot, containing approximately 10 fg of Fe, corresponds to an approximately 350 nm chondritic particle, small enough to be missed by Stardust@Home, indicating that other techniques may be necessary to identify all interstellar candidates. Only one interstellar candidate, I1004,1,2, showed a track. The terminal particle has large enrichments in S, Ti, Cr, Mn, Ni, Cu, and Zn relative to Fe-normalized CI values. It has high Al/Fe, but does not match the Ni/Fe range measured for samples of Al-deck material from the Stardust sample return capsule, which was within the field-of-view of the interstellar collector. A third interstellar candidate, I1075,1,25, showed an Al-rich surface feature that has a composition generally consistent with the Al-deck material, suggesting that it is a secondary particle. The other three interstellar candidates, I1001,1,16, I1001,2,17, and I1044,2,32, showed no impact features or tracks, but allowed assessment of submicron contamination in this aerogel, including Fe hot-spots having CI-like Ni/Fe ratios, complicating the search for CI-like interstellar/interplanetary dust.
Future lander missions will travel to ambitious, scientifically interesting locations near rough and dangerous terrain. They will need to operate with limited prior information about the terrain, and under varying lighting conditions. Landing safely and precisely in the face of these challenges is difficult for existing vision-based landing systems, which require detailed orbital reconnaissance, a priori hazard maps, and impose time-of-day restrictions on landing to ensure similar lighting conditions in orbital and descent imagery. Advanced 3D imaging LiDAR systems currently under development, and originally intended for single-scan hazard detection, have the potential to be operated continuously from altitudes of up to 5 km. Used together with existing inertial measurement units (IMUs), these sensors open a path-to-flight for a full navigation and mapping system, which could replace or augment a traditional landing sensor suite. A landing system based around these sensors can perform accurate altimetry, map-relative localization (MRL), LiDAR-inertial odometry, and map refinement in an illumination-insensitive manner, over unknown or partially known terrain. This paper outlines preliminary work on a LiDAR-inertial landing system that: estimates the spacecraft trajectory during entry, descent, and landing (EDL); and maps the topography of the terrain below, for future use in hazard detection and avoidance. An incremental, factor graph based, smoothing approach is used to solve for the maximum a posteriori trajectory of spacecraft states. Integrated IMU measurements and features tracked in adjacent range and intensity images are used to estimate motion (LiDAR-inertial odometry). LiDAR scans are binned into motion-corrected digital elevation models (DEMs), which are matched to an existing orbital topographic map to provide absolute position information (MRL). The estimated trajectory is then used to project the LiDAR scans into the map frame, creating a variable-resolution quadtree topographic map suitable for hazard detection and avoidance. Existing topographic maps from throughout the solar system (i.e., Earth, the Moon, Mars, Ceres, Vesta, Europa, Enceladus, and Eros) are upsampled for use in EDL simulations. The Mars 2020 Lander Vision System Simulator (LVSS) is extended to simulate LiDAR-inertial data for realistic EDL trajectories. Results of the algorithm operating on the simulated data are presented. Estimated spacecraft trajectory and refined map are compared to ground truth to assess estimation accuracy.
This paper describes a machine vision system for relative spacecraft navigation during the terminal phase of approach to docking that: 1) matches high contrast image features of the target vehicle, as seen by a camera that is bore-sighted to the docking adapter on the chase vehicle, to the corresponding features in a 3d model of the docking adapter on the target vehicle and 2) is robust to on-orbit lighting. An implementation is provided for the case of the Space Shuttle Orbiter docking to the International Space Station (ISS) with quantitative test results using a full scale, medium fidelity mock-up of the ISS docking adapter mounted on a 6-DOF motion platform at the NASA Marshall Spaceflight Center Flight Robotics Laboratory and qualitative test results using recorded video from the Orbiter Docking System Camera (ODSC) during multiple orbiter to ISS docking missions. The Natural Feature Image Registration (NFIR) system consists of two modules: 1) Tracking which tracks the target object from image to image and estimates the position and orientation (pose) of the docking camera relative to the target object and 2) Acquisition which recognizes the target object if it is in the docking camera Field-of-View and provides an approximate pose that is used to initialize tracking. Detected image edges are matched to the 3d model edges whose predicted location, based on the pose estimate and its first time derivative from the previous frame, is closest to the detected edge1 . Mismatches are eliminated using a rigid motion constraint. The remaining 2d image to 3d model matches are used to make a least squares estimate of the change in relative pose from the previous image to the current image. The changes in position and in attitude are used as data for two Kalman filters whose outputs are smoothed estimate of position and velocity plus attitude and attitude rate that are then used to predict the location of the 3d model features in the next image.
An operational system for extracting ice-motion information from synthetic aperture radar (SAR) imagery is being developed as part of the Alaska SAR Facility. This geophysical processing system (GPS) will derive ice-motion information by automated analysis of image sequences acquired by radars on the European ERS-1, Japanese ERS-1, and Canadian RADARSAT remote sensing satellites. The algorithm consists of a novel combination of feature-based and area-based techniques for the tracking of ice floes that undergo translation and rotation between imaging passes. The system performs automatic selection of the image pairs for input to the matching routines using an ice-motion estimator. It is designed to have a daily throughput of ten image pairs. A description is given of the GPS system, including an overview of the ice-motion-tracking algorithm, the system architecture, and the ice-motion products that will be available for distribution to geophysical data users.