DETERMINATION

CROSS REFERENCE TO RELATED APPLICATIONS

[0001] This application claim priority from United States Patent Application No. 61/319,390 filed on March 31 , 2010, the content of which are hereby incorporated by references in it's entirety.

FIELD OF INVENTION

[0002] The present invention relates to the field of positioning, navigation, surveying and mobile mapping. In particular, to a method for improving the quality of a Global navigation satellite system (GNSS) / inertial navigation system (INS) (and other auxiliary sensors, e.g. distance measuring instrument, odometer) trajectory and attitude estimate.

BACKGROUND

[0003] Mobile mapping is a method of spatial data collection by which multiple sensors are attached to a moving platform (airborne, vehicle, ship or pedestrian). The use of the sensors is two-fold: some sensors are used to determine the platforms position and orientation, and some are used to capture information on the scene surrounding the mobile platform, using both passive and active imaging techniques.

[0004] In most applications, the acquired mapping data needs to be geo-coded or geo-referenced by using an estimate of the position and trajectory of the mobile platform during data acquisition. The position and attitude of the platform is normally determined using a combination of GNSS and INS (along with other auxiliary sensors) technologies.

[0005] In some applications, GNSS signal availability can be significantly limited. GNSS observations require a line of sight between mobile mapping platform and at least 4 satellites to provide a unique position solution. In urban areas, underground and under vegetation canopy it is often difficult or impossible to maintain line of sight to 4 satellites. As a result, the position and attitude accuracy of the GNSS/INS navigation system is degraded. Without clear view of satellites for an extended period of time, the accuracy of the platform navigation solution will quickly degrade past the level acceptable for mapping applications. [0006] Existing commercial solutions to the integration of GNSS/INS and other auxiliary sensors and observations for the determination of vehicle position, velocity and attitude rely upon the framework of Kalman Filtering (KF), also referred to as the State-Space Approach (SSA). These are sequential solutions in which all information from previous measurements is contained within the system state and its associated covariance. In this type of solution, there is no possibility to take advantage of the time-correlated crossover point observations that can be gathered from the by utilizing Light Detection And Ranging (LIDAR) and/or image observations using anything other than an ad-hoc approach. These so-called ad-hoc approaches currently integrate the optical measurements within a bundle adjustment. However, the integration is not rigorous in that only GNSS/INS position and orientation estimates, and not their raw observables, are utilized. This type of integration does not allow the estimation of systematic errors at the sensor level. Therefore, this type of modeling of GNSS/INS errors is simplified to shifts or linear drifts: over some portions of the trajectory and is not valid under orientation changes or vehicle acceleration.

SUMMARY

[0007] In accordance with the disclosure there is provided a method of estimating positional trajectories of a mobile mapping platform. The method is implemented in a computer by the execution of instructions stored in memory by a processor. The method comprises receiving positional data captured by the mobile mapping platform at various times; computing positional trajectory estimates from the received positional data; geo-coding image data related to surroundings of the mobile mapping platform captured at, and associated with, respective times using the computed positional trajectory estimates; generating tie-features based on common features in the geo-coded image data captured at different times; and generating a positional trajectory at a plurality of respective times using the received positional data and the generated tie-features. [0008] In accordance with the disclosure there is further provided a system for estimating positional trajectories of a mobile mapping platform. The system comprises a positional data store storing positional data received from the mobile mapping platform captured at various times; an imaging data store storing image data related to a surrounding of the mobile mapping platform captured at various times;

[0009] a memory storing instructions; and a processor for executing the instructions stored in the memory. When the instructions are executed the configure the system to provide a trajectory estimating computing for computing positional trajectory estimates from the received positional data; a tie-feature calculation component for geo-coding image data captured at, and associated with, respective times using the computed positional trajectory estimates and generating tie-features based on common features in geo-coded image data captured at different times; and a trajectory computation component for generating a positional trajectory at a plurality of respective times using the received positional data and the tie-features.

BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Illustrative embodiments of the invention will now be described with reference to the following drawings in which:

Figure 1 depicts an illustrative mobile data capture scenario; Figure 2 depicts another illustrative mobile data capture scenario;

Figure 3 depicts an illustrative processing flow of a data capture and processing system in accordance with the present disclosure;

Figure 4 depicts in a flow chart a method of estimating a trajectory in accordance with the present disclosure; and Figure 5 depicts in a block diagram components of a system for estimating a trajectory in accordance with the present disclosure.

DETAILED DESCRIPTION

[0011] It will be appreciated that for simplicity and clarity of illustration, where considered appropriate, reference numerals may be repeated among the figures to indicate corresponding or analogous elements. In addition, numerous specific details are set forth in order to provide a thorough understanding of the embodiments described herein. However, it will be understood by those of ordinary skill in the art that the embodiments described herein may be practiced without these specific details. In other instances, well-known methods, procedures and components have not been described in detail so as not to obscure the embodiments described herein. Also, the description is not to be considered as limiting the scope of the embodiments described herein. [0012] A mobile mapping platform may comprise a positional component for detecting or determining information used to determine the position of the mobile mapping platform at various times. The mobile mapping platform includes an imaging platform that can capture image data using passive and/or active imagers/scanner to capture image data at various times. The image data captured by passive and active imaging sensor provides a geometric representation of the surroundings of the mobile mapping platform. The passive and active imaging sensors mounted on most mobile mapping platforms acquire information about the geometry and other physical properties surrounding the mobile mapping platform. Traditionally these sensors have been used exclusively to map the vehicle surroundings as input into geographic information systems (GIS) and base mapping initiatives. However, because these sensors collect precise geometric information, and because the information collected is normally redundant, that is the same objects are imaged at different times and different viewing angles, it can be used as additional information to the trajectory estimation to improve positioning accuracy in times where GNSS satellite availability is limited.

[0013] A system and method are described that allow the use of time correlated observations in the positioning data stream which allows the optimal estimation of Inertial Navigation System (INS) errors during periods of Global Navigation Satellite System (GNSS) outages and overcomes the shortcomings of Kalman Filtering (KF) or the State-Space Approach (SSSA) are described further herein.

[0014] Figure 1 depicts an illustrative mobile data capture scenario. As depicted, an automobile 102 captures image data of its surroundings as it moves in various directions. The automobile may capture a feature, such as a building 104, using the imaging platform from different points of view at different epochs (e1 , e2, e3). As the automobile is traveling along a capture corridor, or corridors, 106, the positional component detects information, such as GNSS data or INS data, used to determine the position of the mobile mapping platform at different times. All of the captured images and positional information are associated with a time the image or information was captured. The time can be provided by an accurate clock to allow subsequent synchronization of the different information.

[0015] Figure 2 depicts another illustrative typical mobile data capture scenario. As depicted, an automobile 102 captures image data of its surroundings at two different times (t1 , t2) as it moves in a direction (d1 ). Similarly to as described above, the automobile also captures positional information that can be used to determined the position, and attitude or orientation of the automobile, and so the mobile mapping platform. As depicted, features along the data collection corridor 202 are scanned multiple times from different directions at different points in time.

[0016] Figure 3 depicts an illustrative processing flow of a data capture and processing system in accordance with the present disclosure. Raw positional data is captured and stored in a data storage 310. The positional data may include information from GNSS mobile observations 302, INS 304, auxiliary navigation sensors 306 and GNSS reference receiver or network 308. Each piece of positional data is captured at, and associated with a particular time. The time may be provided by a clock of the mobile mapping platform used to capture the positional data. As the positional data is being captured, image data 312 is also captured and associated with the time at which it was captured. The image data may comprise LIDAR image data, digital imagery data such as from a digital camera, or other types of imaging devices capable of capturing images of their surroundings. The image data 312 is stored in an image data store 314. Each piece of data captured is associated with a time that can be synchronized to a common time base to allow subsequent synchronization of the different captured data. [0017] Once the positional data and the imaging data is stored, it is processed to determine a positional trajectory of the mobile mapping platform. The positional data is first processed to provide an initial estimate of the positional trajectory and attitude of the mobile mapping platform at different times using a state space approach, such as a tightly integrated Kalman filter 316. The initial estimate of the positional trajectories and attitudes provided by the Kalman filter solution 318 can be used first to provide initial position and attitude information for the imaging platform, which are necessary for LIDAR observations, but may not be necessary for line scanner or frame sensors. The image data is then geo-coded or geo-referenced using the estimate of the positional trajectories and attitudes and used to find objects, features or structures that can be identified in the imaging data that have been observed at different points in time in order to provide common features which can tie the vehicle trajectory together at different times 320. The common features are used to define common correlated overlap features, referred to as tie-features, in the overall mobile mapping platform navigation trajectory. The tie-features are time correlated 322 according to the time at which the image data were captured that the tie-features are determined from. The tie-features are then combined with the original raw positional data such as GNSS, INS and auxiliary navigation sensor data in a dynamic network adjustment 324 which computes a best estimate of platform trajectory, attitude, and navigation sensor errors 326. As described further below, the dynamic network adjustment may require initial values to base the computations from in order to provide a fast convergence. These initial values may be provided by the initial estimate of the positional trajectory and attitude 318.

[0018] Figure 4 depicts in a flow chart a method of estimating a trajectory of a mobile mapping platform in accordance with the present disclosure. The mobile mapping platform captures and stores the positional data and image data in real-time. The positional data and image data are associated with a capture time that can be referenced to a common time base to allow the positional data and image data to be synchronized. Once captured, the stored positional data and image data may be utilized in a post processing environment which considers all observations collected in one adjustment. The method 400 receives the positional data (402) and computes positional trajectory estimates (404) using the positional data. The positional trajectory estimates may be computed for different times or epochs. The positional trajectory estimates may be determined utilizing a state space approach, such as Kalman filtering, to process the GNSS, INS and auxiliary navigation sensor data to obtain an initial estimate of positional trajectory. A smoothing algorithm, such as Rauch-Tung-Striebel, may be applied to the initial positional trajectory estimates to obtain a smoothed best estimate of the positional trajectory.

[0019] Once the positional trajectory estimates are computed, they are used to geo- code image data (406). Using the best estimate of positional trajectory, along with the precise timing information associated with the image data, the position and attitude of the mobile mapping platform at the moment of data capture for all of the captured image data are determined. The position and attitude of the mobile mapping platform at the time of image data capture is used to precisely geo-code or reference all of the image data acquired from the imaging platform of the mobile mapping platform.

[0020] Once the image data is geo-coded, tie-features are generated (408). The tie- features anchor together the mobile platform trajectory at different points in time. The tie-features may be generated by examining the imaging data to locate common features that have been imaged from different locations and/or from different points in time. The common features may be immoveable or permanent features, objects or structures. The identification of the common features in the different image data can be accomplished by various methods, including automatic feature recognition, semi-automatic feature recognition, manual examination and identification of common features, or combinations there of. Once the tie-features are determined, their accuracy may be determined through application of the laws of variance and covariance propagation. The tie-features are provided in a format which can be utilized as common observations in a dynamic network adjustment.

[0021] Once the tie-features are generated, positional trajectories are generated (410). The positional trajectories of the mobile mapping platform may be generated from the original positional data, such as GNSS, INS and auxiliary navigation sensor data, along with the tie-features and their associated estimated accuracy. The positional trajectories and the tie-functions may be combined to provide the positional trajectories using dynamic network adjustment. The dynamic network adjustment uses all available navigation sensor data, and the cross over and tie point conditions to compute the positional trajectories.

[0022] The application of dynamic network adjustment stems from the traditional static network adjustment (NA). NA refers to the linking together instruments, observations and parameters by functional mathematical models. If the models are non-linear they are linearized and the network is formulated as a linear equation system which is generally solved by minimizing the weighted squared residuals of the observations. This is different from a State Space Approach because in a NA approach all observations can be considered at once, which allows the introduction of time correlation between observations - a feature the State Space Approach lacks. However, the traditional formulations of NA only consider static observation models. Therefore, a new formulation is required which considers both static and dynamic observation models, this is the so-called dynamic network adjustment.

[0023] The dynamic network adjustment is an extension of NA that allows the simultaneous handling of both static and dynamic mathematical models. The dynamic network adjustment allows use of time-correlated observations in order to recover time-correlated errors in the raw positional data. The dynamic models, normally defined by differential equations (e.g. GNSS and INS integration) are converted to difference equations which are approximations of the differential equations. The difference equations can then be rearranged to be treated as normal observations in a NA.

[0024] Using, for example a three point stencil method of equation (1) below, all epochs can be referenced by time in one long chain of parameters, which can be considered former epoch states in the state-space approach. Although depicted as using a three point stencil method, other forms of numerical differentiation approximation may also be utilized or deemed more appropriate for this purpose.

Where:

X k = system state at epoch k;

X k =time derivative of system state at epoch k; and

=time interval of state vector updates. [0025] Using the time based system state from above, the dynamic model of positional observations or measurements, can be stated as difference equations according to the form: f _d (i,,z _j . +w _i ,x _i ,x _i )=f _i (t _i ,z _i + w _i ,x _i _ _] ,x _i ,x _i+] )=0 (2) Where:

= observations, for example INS observables; W/ observation residuals; t/c =time at epoch k; and

%k =the adjustment parameters at epoch k. [0026] The Static observation model which allows introduction of observations that are not modeled by differential equations such as GPS observations can be stated as:

Where:

all values have been previously defined.

[0027] The tie-features, which add observations or introduce conditions between parameters at different times can be stated as: f _c (t _i ,t _k ,x _i ,x _k ,r _c ) = f _c (t _i ,t _k ,x _i ,x _k )-r _c = 0 ₍₄₎ Where:

r _c =residual vector for a certain cross-over or tie point.

[0028] The equations (2) to (4) can be linearized and rearranged into a Gauss- Helmert equation system of the form:

A-<5x + B- v + u = 0 (5) Where:

A = design matrix of derivatives of observation models w.r.t. the parameters; innovation vector of the parameters;

B = design matrix of derivatives of observation models w.r.t. the observations;

v = residuals of the observations; and

u = misclosure vector.

[0029] The above Gauss Helmert System formulation above can then be solved using traditional network adjustment techniques. Due to the above formulation, the dynamic network adjustment requires an initial approximation for the various parameters. These can be provided by the previously computed estimate of the vehicle trajectory used to geo-code the image data.

[0030] Figure 5 depicts in a block diagram components of a system for estimating a trajectory of a mobile mapping platform in accordance with the present disclosure. The system 500 comprises a computing system 502 for estimating the positional trajectories, a positional data store 504 for storing positional data, such as GSNN, INS and Auxiliary navigation data and an imaging data store 506 for storing image data, such as LIDAR information or digital imagery. As depicted in Figure 5, all of the data stored in both the positional data store 504 and the imaging data store 506 is associated with a time corresponding to when the data was captured by the mobile mapping platform (not shown). [0031] The computing system 502 comprises a processor 508 for executing instructions stored in memory 510. The memory 510 may comprise both volatile memory and non-volatile memory 512. The memory 510 stores instructions 514, that when executed by the processor 508, configure the computing system to provide positional trajectory determination functionality 516. [0032] The trajectory determination functionality 516 determines positional trajectories for the mobile mapping platform as described above. The trajectory determination functionality 516 comprises a trajectory estimation component 518 that receives positional data and associated time 520 from the positional data store 504 and computes positional trajectory estimates 522 from the positional data. The computed positional trajectory estimates 522 are provided to a tie-feature computation component 524 that computes tie-features 528 using the positional W trajectory estimates 522 and the image data 526 from the imaging data store 506. The tie-features 528 are provided to a positional trajectory computation component 530 that computes the positional trajectories 532 using the positional data 520. As described above, the positional trajectory computation component 530 may compute the positional trajectories 532 using dynamic network adjustments, which may require initial values be provided as an initial approximation for numerical stability and fast convergence of the Dynamic Network solution. These initial values may be provided from the trajectory estimation component 518.

[0033] Although the above discloses example method and system for providing estimating positional trajectories of a mobile mapping platform providing positional data related to a position of the mobile mapping platform captured at various times and image data related to a surrounding of the mobile mapping platform captured at various times, it should be noted that such apparatus and method are merely illustrative and should not be considered as limiting. For example, it is contemplated that any or all of these hardware and software components could be embodied exclusively in hardware, exclusively in software, exclusively in firmware, or in any combination of hardware, software, and/or firmware. Accordingly, while the following describes example methods and system, persons having ordinary skill in the art will readily appreciate that the examples provided are not the only way to implement such methods and apparatus.

[0034] It will be apparent to one skilled in the art that numerous modifications and departures from the specific embodiments described herein may be made without departing from the spirit and scope of the present invention.