Abstract: An aerial vehicle is navigated using hierarchical vision-aided navigation that classifies regions of acquired still image frames as featureless or feature-rich, and thereby avoids expending time and computational resources attempting to extract and match false features from the featureless regions. Pattern recognition registers an acquired image to a general area of a map database before performing feature matching to a finer map region. This hierarchical position determination is more efficient than attempting to ascertain a fine-resolution position without knowledge of coarse-resolution position. Resultant matched feature observations can be data-fused with other sensor data to correct a navigation solution based on GPS and/or IMU data.

Abstract: An aerial vehicle is navigated using hierarchical vision-aided navigation that classifies regions of acquired still image frames as featureless or feature-rich, and thereby avoids expending time and computational resources attempting to extract and match false features from the featureless regions. Pattern recognition registers an acquired image to a general area of a map database before performing feature matching to a finer map region. This hierarchical position determination is more efficient than attempting to ascertain a fine-resolution position without knowledge of coarse-resolution position. Resultant matched feature observations can be data-fused with other sensor data to correct a navigation solution based on GPS and/or IMU data.

Abstract: An aerial vehicle is navigated using vision-aided navigation that classifies regions of acquired still image frames as featureless or feature-rich, and thereby avoids expending time and computational resources attempting to extract and match false features from the featureless regions. The classification may be performed by computing a texture metric as by testing widths of peaks of the autocorrelation function of a region against a threshold, which may be an adaptive threshold, or by using a model that has been trained using a machine learning method applied to a training dataset comprising training images of featureless regions and feature-rich regions. Such machine learning method can use a support vector machine. The resultant matched feature observations can be data-fused with other sensor data to correct a navigation solution based on GPS and/or IMU data.

Abstract: An aerial vehicle is navigated using vision-aided navigation that classifies regions of acquired still image frames as featureless or feature-rich, and thereby avoids expending time and computational resources attempting to extract and match false features from the featureless regions. The classification may be performed by computing a texture metric as by testing widths of peaks of the autocorrelation function of a region against a threshold, which may be an adaptive threshold, or by using a model that has been trained using a machine learning method applied to a training dataset comprising training images of featureless regions and feature-rich regions. Such machine learning method can use a support vector machine. The resultant matched feature observations can be data-fused with other sensor data to correct a navigation solution based on GPS and/or IMU data.

Abstract: A system includes a sensor to generate a first image having a first two-dimensional image pixel data set. A database provides a second image having a second two-dimensional image pixel data set that includes a three-dimensional positional data set describing a navigational position of each pixel in the second two-dimensional image pixel data set. A vision module includes an edge extractor to extract image edge features from the first two-dimensional pixel data set and image edge features from the second two-dimensional image pixel data set. The vision module includes a feature correlator to determine a navigational position for each pixel in the first two-dimensional data set based on an image edge feature comparison of the extracted edge features from the first and second two-dimensional image pixel data sets.

Abstract: A system includes a sensor to generate a first image having a first two-dimensional image pixel data set. A database provides a second image having a second two-dimensional image pixel data set that includes a three-dimensional positional data set describing a navigational position of each pixel in the second two-dimensional image pixel data set. A vision module includes an edge extractor to extract image edge features from the first two-dimensional pixel data set and image edge features from the second two-dimensional image pixel data set. The vision module includes a feature correlator to determine a navigational position for each pixel in the first two-dimensional data set based on an image edge feature comparison of the extracted edge features from the first and second two-dimensional image pixel data sets.

Abstract: In one embodiment, a method for optimizing a set of matched features is provided. The method includes matching features between an optical image and a geo-referenced orthoimage to produce an initial set of matched features. An initial position solution is then determined for the optical image using the initial set of N matched features. The initial set of N matched features are then optimized based on a set of N sub-solutions and the initial position solution, wherein each of the N sub-solutions is a position solution using a different set of (N?1) matched features. A refined position solution is then calculated for the optical image using the optimized set of matched features.

Abstract: A system is provided for correlating evidence grids. In certain embodiments, the system includes a sensor that generates signals describing a current section of an environment; a memory configured to store measurements of historical sections of the environment; and a processor coupled to the sensor and configured to calculate navigation parameters based on signals received from the sensor. Further, the processor converts the signals received from the sensor into a current evidence grid and removes data from the current evidence grid to form a reduced evidence grid; converts the measurements of historical sections into a historical evidence grid; and correlates the reduced evidence grid with the historical evidence grid by adjusting position and orientation of the reduced evidence grid and the historical evidence grid in relation to one another and calculating correlative values, and searching for a highest correlative value.

Abstract: In one embodiment, a method for optimizing a set of matched features is provided. The method includes matching features between an optical image and a geo-referenced orthoimage to produce an initial set of matched features. An initial position solution is then determined for the optical image using the initial set of N matched features. The initial set of N matched features are then optimized based on a set of N sub-solutions and the initial position solution, wherein each of the N sub-solutions is a position solution using a different set of (N?1) matched features. A refined position solution is then calculated for the optical image using the optimized set of matched features.

Abstract: Systems and methods for 3D data based navigation using descriptor vectors are provided. In at least one embodiment, a method for identifying corresponding segments in different frames of data comprises identifying a first segment set in a first frame in multiple frames acquired by at least one sensor, and identifying a second segment set in a second frame in the multiple frames. The method also comprises calculating a first and second set of descriptor vectors, wherein the first and second sets of descriptor vectors comprise a descriptor vector for each segment in the respective first and second segment set, wherein a descriptor vector describes an indexed plurality of characteristics; and identifying corresponding segments by comparing the first set of descriptor vectors against the second set of descriptor vectors, wherein the corresponding segments describe characteristics of the same feature in the environment.

Abstract: Systems and methods for 3D data based navigation using a watershed method are provided. In at least one embodiment, a method for segmenting three-dimensional frames of data comprises acquiring at least one frame from at least one sensor, wherein the at least one frame provides a three-dimensional description of an environment containing the at least one sensor; and identifying a surface in the at least one frame. The method further comprises computing at least one residual map for the at least one frame based on the orthogonal distance from data points on the surface to at least one polynomial surface fitted to the surface; and segmenting the at least one residual map by performing a watershed algorithm on the residual map.

Abstract: Systems and methods for using image warping to improve geo-registration feature matching in vision-aided positioning is provided. In at least one embodiment, the method comprises capturing an oblique optical image of an area of interest using an image capturing device. Furthermore, digital elevation data and at least one geo-referenced orthoimage of an area that includes the area of interest are provided. The area of interest in the oblique optical image is then correlated with the digital elevation data to create an image warping matrix. The at least one geo-referenced orthoimage is then warped to the perspective of the oblique optical image using the image warping matrix. And, features in the oblique optical image are matched with features in the at least one warped geo-referenced orthoimage.

Abstract: Systems and methods for feature selection and matching are provided. In certain embodiments, a method for matching features comprises extracting a first plurality of features from current image data acquired from at least one sensor and extracting a second plurality of features from a prior map, wherein the prior map represents an environment containing the navigation system independently of data currently acquired by the at least one sensor. The method also comprises identifying at least one first feature in the first plurality of features and at least one second feature in the second plurality of features that have associated two-dimensional representations; and identifying at least one corresponding pair of features by comparing a three-dimensional representations of the at least one first feature to a three-dimensional representation of the at least one second feature.

Abstract: Systems and methods are provided for selecting landmarks for navigation. In one embodiment, a system comprises an IMU that provides inertial measurements for a vehicle and at least one image sensor that acquires measurements of the vehicle's environment. The system also comprises a processing unit that calculates a navigation solution for the vehicle based on the inertial measurements, identifies a plurality of landmarks in the acquired measurements, and identifies a plurality of usable landmarks from the plurality of landmarks. The processing unit also selects a subset of useable landmarks from the plurality of useable landmarks such that the subset of landmarks has a smaller dilution of precision (DOP) than other possible subsets of landmarks from the plurality of useable landmarks, and calculates an updated navigation solution from the subset of landmarks. The DOP is an amplification factor of measurement errors derived from the geometry of the subset of useable landmarks.

Abstract: A method for autonomous landing of an unmanned aerial vehicle (UAV) comprising: obtaining sensor data corresponding to one or more objects outside of the aircraft using at least one onboard sensor; using the sensor data to create a three dimensional evidence grid, wherein a three dimensional evidence grid is a three dimensional world model based on the sensor data; combining a priori data with the three dimensional evidence grid; locating a landing zone based on the combined three dimensional evidence grid and a priori data; validating an open spots in the landing zone, wherein validating includes performing surface condition assessment of a surface of the open spots; generating landing zone motion characterization, wherein landing zone motion characterization includes characterizing real time landing zone pitching, heaving, rolling or forward motion; processing the three dimensional evidence grid data to generate flight controls to land the aircraft in one of the open spots; and controlling the aircraft acco

Abstract: Systems and methods for comparing range data with evidence grids are provided. In certain embodiments, a system comprises an inertial measurement unit configured to provide inertial measurements; and a sensor configured to provide range detections based on scans of an environment containing the navigation system. The system further comprises a navigation processor configured to provide a navigation solution, wherein the navigation processor is coupled to receive the inertial measurements from the inertial measurement unit and the range measurements from the sensor, wherein computer readable instructions direct the navigation processor to identify a portion of an evidence grid based on the navigation solution; compare the range detections with the portion of the evidence grid; and calculate adjustments to the navigation solution based on the comparison of the range detections with the portion of the evidence grid to compensate for errors in the inertial measurement unit.

Abstract: A method for autonomous landing of an unmanned aerial vehicle (UAV) comprising: obtaining sensor data corresponding to one or more objects outside of the aircraft using at least one onboard sensor; using the sensor data to create a three dimensional evidence grid, wherein a three dimensional evidence grid is a three dimensional world model based on the sensor data; combining a priori data with the three dimensional evidence grid; locating a landing zone based on the combined three dimensional evidence grid and a priori data; validating an open spots in the landing zone, wherein validating includes performing surface condition assessment of a surface of the open spots; generating landing zone motion characterization, wherein landing zone motion characterization includes characterizing real time landing zone pitching, heaving, rolling or forward motion; processing the three dimensional evidence grid data to generate flight controls to land the aircraft in one of the open spots; and controlling the aircraft acco

Abstract: A method for collaborative navigation between two or more platforms is provided. The method comprises establishing a communication link between a first platform and a second platform, making a sensor measurement from the first platform, updating state and covariance elements of the first platform, and transmitting the updated state and covariance elements from the first platform to the second platform. A conditional update is performed on the second platform to compute a new estimate of state and covariance elements on the second platform, which takes into account the measurement from the first platform. The method further comprises making a sensor measurement from the second platform, updating state and covariance elements of the second platform, and transmitting the updated state and covariance elements from the second platform to the first platform.

Abstract: Systems and methods for navigation using cross correlation on evidence grids are provided. In one embodiment, a system for using cross-correlated evidence grids to acquire navigation information comprises: a navigation processor coupled to an inertial measurement unit, the navigation processor configured to generate a navigation solution; a sensor configured to scan an environment; an evidence grid creator coupled to the sensor and the navigation processor, wherein the evidence grid creator is configured to generate a current evidence grid based on data received from the sensor and the navigation solution; a correlator configured to correlate the current evidence grid against a historical evidence grid stored in a memory to produce displacement information; and where the navigation processor receives correction data derived from correlation of evidence grids and adjusts the navigation solution based on the correction data.

Abstract: Method and system for building a support vector machine binary tree for fast object search. An appearance model can be generated for objects in a database and computed on regions detected in an image frame. A covariance matrix can be utilized for representing the appearance model of the detected regions. The covariance matrix appearance model can be preprocessed and/or transferred into a vector-based format. The data in the vector-based format can be added with a class label to form labeled data. A support vector machine (SVM) can be utilized on the labeled data to generate a classifier with an optimal hyperplane and a margin area in order to hierarchically build a balanced SVM binary tree. A query appearance model can be searched rapidly utilizing the SVM binary tree during a search phase.