Galina Okouneva

Fault-Tolerant Position/Attitude Estimation of Free-Floating Space Objects Using a Laser Range Sensor

This paper presents a fault-tolerant method for pose estimation of space objects using 3-D vision data by integration of a Kalman filter (KF) and an iterative closest point (ICP) algorithm in a closed-loop configuration. The initial guess for the internal ICP iteration is provided by state estimate propagation of the KF. The KF is capable of not only estimating the targets states, but also its inertial parameters. This allows the motion of the target to be predictable as soon as the filter converges. Consequently, the ICP can maintain pose tracking over a wider range of velocity due to increased precision of ICP initialization. Furthermore, incorporation of the targets dynamics model in the estimation process allows the estimator to continuously provide pose estimation even when the sensor temporally loses its signal namely due to obstruction. The capabilities of the pose estimation methodology is demonstrated by a ground testbed for automated rendezvous and docking (AR&D). In this experiment, Neptecs Laser Camera System (LCS) is used for real-time scanning of a satellite model attached to a manipulator arm, which is driven by a simulator according to orbital and attitude dynamics. The results showed that robust tracking of the free-floating tumbling satellite can be achieved only if the KF and ICP are in a closed-loop configuration.

Intelligent LIDAR scanning region selection for satellite pose estimation

This paper proposes a method to select a near-optimal laser scanning area on a target body that will result in the best registration accuracy. The method is based on constraint analysis and employs a sensitivity index which is used as a registration accuracy predictor. It is shown that point cloud configurations with higher values of this index return more accurate pose estimates than unstable configurations with lower index values. Iterative Closest Point (ICP) registration tests are conducted on four satellite geometries using synthetic range data. The proposed method can be used to increase the accuracy of ICP registration and to reduce registration processing time.

Robust vision-based pose estimation of moving objects for Automated Rendezvous & Docking

This paper presents a fault-tolerant method for pose estimation of space objects using 3-D vision data by integration of a Kalman filter (KF) and an Iterative Closest Point (ICP) algorithm in a closed-loop configuration. The initial guess for the internal ICP iteration is provided by state estimate propagation of the Kalman filer. The Kalman filter is capable of not only estimating the targets states, but also its inertial parameters. This allows the motion of target to be predictable as soon as the filter converges. Consequently, the ICP can maintain pose tracking over a wider range of velocity due to increased precision of ICP initialization. Furthermore, incorporation of the targets dynamics model in the estimation process allows the estimator continuously provide pose estimation even when the sensor temporally loses its signal namely due to obstruction. The capabilities of the pose estimation methodology is demonstrated by a ground testbed for Automated Rendezvous & Docking (AR&D). In this experiment, Neptecs Laser Camera System (LCS) is used for real-time scanning of a satellite model attached to a manipulator arm, which is driven by a simulator according to orbital and attitude dynamics. The results showed that robust tracking of the free-floating tumbling satellite can be achieved only if the Kalman filter and ICP are in closed-loop configuration.

Fault-tolerant pose estimation of space objects

This paper presents a fault-tolerant method for pose estimation of space objects using 3-D vision data by integration of a Kalman filter (KF) and an Iterative Closest Point (ICP) algorithm in a closed-loop configuration. The initial guess for the internal ICP iteration is provided by state estimate propagation of the Kalman filer. The Kalman filter is capable of not only estimating the targets states, but also its inertial parameters. This allows the motion of target to be predictable as soon as the filter converges. Consequently, the ICP can maintain pose tracking over a wider range of velocity due to increased precision of ICP initialization. Furthermore, incorporation of the targets dynamics model in the estimation process allows the estimator continuously provide pose estimation even when the sensor temporally loses its signal namely due to obstruction. The capabilities of the pose estimation methodology is demonstrated by a ground testbed for Automated Rendezvous & Docking (AR&D).

Stereo vision algorithm for robotic assembly operations

A stereo vision Linear Triangulation (LT) algorithm can be utilized in space robotics assembly operations. The LT algorithm recovers the relative orientation and translation (pose) of objects marked with high contrast targets using two or more pinhole charge-coupled device (CCD) cameras. The cameras view a set (including a disjoint set) of targets measured with respect to the same point in space. This study evaluates the theoretical accuracy of the LT algorithm, its benefits and performance. The introduction of a third camera into the vision system envelope is also examined and discussed. Experiments indicated that most accuracy in pose estimation is gained by the first 15-25 degrees of camera separation, and then the decrease in values of the covariance matrix elements stabilizes. We compare the numerical data for singe, two-camera and threecamera cases using extensive experiments on simulated images.

Stability Improvement of Vision Algorithms

This paper presents and demonstrates an automated generic approach to improving the accuracy and stability of iterative pose estimation in computer vision applications. The class of problem involves the use of calibrated CCD camera video imagery to compute the pose of a slowly moving object based on an arrangement of visual targets on the surface of the object. The basis of stereo-vision algorithms is to minimize a re-projection error cost function. The proposed method estimates the optimal target locations within the area of interest. The optimal target configuration delivers the minimal condition number of the linear system associated with the iterative algorithm. The method is demonstrated for the case when targets are located within a 3D domain. Two pose estimation algorithms are compared: single camera and two-camera algorithms. A better accuracy in pose estimation can be achieved with a single camera algorithm with optimized target locations. Also, this method can be applied to perform optimization of target locations attached to a 2D surface.

Shape-based pose estimation evaluation using expectivity index artifacts

This paper recommends the use of three distinct shape artifacts to evaluate shape-based pose estimation systems, and provide their rationale. These artifacts are the Reduced Pose Ambiguity Cuboctohedron (RPAC), a cube, and an 80-triangle tessellated sphere. The rationale for these shapes derives from the range of Expectivity Index (EI) values and ambiguity intervals. The EI varies inversely with expected pose estimation error for a given shape and view, and the ambiguity interval describes a distance between symmetries where a shape fits with the incorrect pose as precisely as with the correct one. These concepts are discussed in detail and used to define the proposed shapes as good for covering a range of circumstances for performance evaluation of shape-based post estimation systems, and are proposed for inclusion in the ASTM E57.02 standards for pose estimation evaluation.

Localization of door handle using a single camera on a door opening mobile manipulator

This paper presents a novel approach to localizing a door handle of unknown geometry to assist in autonomous door opening. The localization is performed using data from a single CCD camera that is mounted at the end-effector of a mobile manipulator. The proposed algorithm extracts a 3D point cloud using optical flow and known camera motion provided by the manipulator. Segmentation of the point cloud is then performed, enabling the separation of the door and the handle points, which is then followed by fitting a boundary box to the door handle data. The fitted box can then be used to guide robotic grasping. The proposed algorithm has been validated using a 3D virtual scene, and the results have demonstrated the effectiveness of the proposed method to localize a door handle in an unknown environment.

CONTINUUM SHAPE CONSTRAINT ANALYSIS: A CLASS OF INTEGRAL SHAPE PROPERTIES APPLICABLE TO LIDAR/ICP-BASED POSE ESTIMATION

Surface registration involving the estimation of a rigid transformation (pose) which aligns a model provided as a triangulated mesh with a set of discrete points (range data) sampled from the actual object is a core task in computer vision. This paper refines and explores the previously introduced notion of Continuum Shape Constraint Analysis (CSCA) which allows the assessment of object shape towards predicting the performance of surface registration algorithms. Conceived for computer-vision assisted spacecraft rendezvous analysis, the approach was developed for blanket or localized scanning by LIDAR or similar range-finding scanner that samples non-specific points from the object across an area. Based on the use of Iterative Closest-Point Algorithm (ICP) for pose estimation, CSCA is applied to a surface-based self-registration cost function which takes into account the direction from which the surface is scanned. The continuum nature of the CSCA formulation generates a registration cost matrix and any derived metrics as pure shape properties of the object. For the context of directional scanning as considered in the paper, these properties also become functions of viewing direction and is directly applicable to the best view problem for LIDAR/ICP pose estimation. This paper introduces the Expectivity Index and uses it to illustrate the ability of the CSCA approach to identify productive views via the expected stability of the global minimum solution. Also demonstrated through the examples, CSCA can be used to produce visual maps of geometric constraint that facilitate human interpretation of the information about the shape. Like the ICP algorithm it supports, the CSCA approach processes shape information without the need for specific feature identification and is applicable to any type of object.

Development of Continuum Shape Constraint Analysis (CSCA) for Computer Vision Applications Using Range Data

This paper further presents continuum shape constraint analysis (CSCA) of surfaces. CSCA is a generalization of discrete-point based constraint analysis which can be used to predict performance of registration algorithms. A surface-based self-registration cost function to which constraint analysis can be applied is introduced. This cost function takes into account a direction the object is viewed at. A sample study is provided to illustrate this approach applied to the problem of pose estimation using range-data taken from a scanning instrument such as LIDAR. Specifically, CSCA is used to assess an object feature for suitability for local LIDAR scanning and subsequent application of the ICP (iterative closest-point) algorithm to determine pose. In this study, the constraint analysis uses noise amplification index (NAI) as an output measure. The continuum nature of the CSCA approach renders the registration cost matrix and the derived NAI as pure shape properties of the feature with a dependence on viewpoint.