Automatic correction of past position errors for location and inventory tracking

• US 8,838,374 B2
• Filed: 09/07/2010
• Issued: 09/16/2014
• Est. Priority Date: 07/05/2005
• Status: Active Grant

• Current Assignee: Mi-Jack Products Incorporated (Lanco International, Inc.)
• Original Assignee: Mi-Jack Products Incorporated (Lanco International, Inc.)
• Inventors: Tan, Han-Shue, Huang, Jihua
• Primary Examiner(s): Nguyen, Cuong H
• Application Number:
  US12/876,918
• Publication Number:
  US 20100332125A1
• Time in Patent Office: 1,470 Days
• Field of Search: 701518-521, 701/469, 701/472, 701/501, 701/507, 701518- 5, 340/995.25, 340/995.28, 340/989, 340/539.13, 340/988, 340/995.1, 700213-214, 700228-229
• US Class Current: 701/408
• CPC Class Codes: G01S 19/39 the satellite radio beacon positioning system transmitting time-stamped messages, e.g. GPS [Global Positioning System], GLONASS [Global Orbiting Navigation Satellite System] or GALILEO
• CPC Class Codes: G01S 19/393 Trajectory determination or predictive tracking, e.g. Kalman filtering
• CPC Class Codes: G01S 19/396 Determining accuracy or reliability of position or pseudorange measurements
• CPC Class Codes: G01S 19/485 whereby the further system is an optical system or imaging system
• CPC Class Codes: G01S 19/49 whereby the further system is an inertial position system, e.g. loosely-coupled
• CPC Class Codes: G01S 5/02585 at least one of the measurements being a non-radio measurement

Abstract

A method is provided for calibrating past position estimates from a positioning system that provides real-time position estimates of a mobile object. The method first stores the real-time position estimates, which as time goes by become past position estimates and naturally form a first past trajectory depicting the past movement of the mobile object. Subsequently, a calibrated past trajectory is determined, which includes calibrated past position estimates that correspond to the same time instances as the past positions in the first past trajectory. When real-time positions have low qualities, this method calibrates them at a later time by using (higher-quality) real-time positions both before and after them. Errors in the past positions are then corrected based on the calibrated past trajectory. When used with event detectors that indicate inventory transactions, this method can correct position errors associated with inventory events so as to improve the performance of inventory tracking.

Claims

1. A method for calibrating past position estimates from a positioning system that provides real-time position estimates of a mobile object, comprising

storing the real-time position estimates as first past positions and aggregating the first past positions to form a first past trajectory, wherein the first past trajectory comprises at least three first past positions;

generating a calibrated past trajectory based on the first past trajectory, wherein the calibrated past trajectory comprises second past positions and at least one of the second past positions is determined by using at least one of the first past positions that corresponds to an earlier time instance;

whereby the second past positions are closer to actual past positions of the mobile object than the first past positions and the calibrated past trajectory is closer to the actual past trajectory of the mobile object than the first past trajectory.

2. The method in claim 1, wherein the positioning system provides the real-time position estimates by computing at least a pair of position values corresponding to the mobile object'"'"'s position in a two-degree-of-freedom coordinate system, a heading angle corresponding to traveling direction of the mobile object, and a traveling speed of the mobile object.

3. The method in claim 1, wherein the generation of the calibrated past trajectory further comprises computing at least one temporary past trajectory based on mathematical models that depict movement of the mobile object and combining the temporary past trajectories with the first past trajectory to generate the calibrated past trajectory.

4. The method in claim 1, wherein the generation of the calibrated past trajectory involves at least one of the following:

employing models that depict movement of the mobile object, wherein at least one model takes the first past positions as input;

applying estimation formulas that include mathematic terms accounting for accumulated effects of noise and errors in the first past positions;

estimating and reducing noise and errors by modeling the noise and errors as well as propagation of the noise and errors in trajectory calibration process; and

fitting the calibrated past trajectory to selected segments of the first past trajectory by calibrating parameters.

5. The method in claim 1, wherein the generation of the calibrated past trajectory further comprises iterations of computing temporary past trajectories until the temporary past trajectories satisfy pre-determined criteria, wherein

the first iteration determines a first temporary past trajectory based on the first past trajectory;

each subsequent iteration uses the temporary past trajectory determined in the preceding iteration as an input to determine a subsequent temporary past trajectory; and

the temporary past trajectory determined in the last iteration is the calibrated past trajectory.

6. The method in claim 1, wherein the generation of the calibrated past trajectory further comprises

selecting a segment of the first past trajectory, wherein the segment has a length smaller or equal to length of the first part trajectory;

computing a calibrated past trajectory based on the selected segment of the first past trajectory, wherein the calibrated past trajectory corresponds to same time instances as the segment of the first past trajectory.

7. The method in claim 1, wherein the positioning system further provides a confidence level indicating accuracy and trustworthiness for each of the first past positions;

and the generation of the calibrated past trajectory further incorporates the confidence levels of the first past positions, whereby the first past positions that have higher confidence levels have more effects on the generation of the calibrated past trajectory.

8. The method in claim 2, wherein

each first past position further comprises an angular velocity of the mobile object;

and

the generation of the calibrated past trajectory further comprises calibrating the heading angle of the first past positions based on the angular velocity and using the calibrated heading angle to compute the calibrated past trajectory.

9. The method in claim 8, wherein the calibration of the heading angle further comprises estimating and approximately removing noise and errors in at least one of the following:

the heading angle and the angular velocity of the mobile object.

10. The method in claim 9, wherein

the noise and errors comprise a bias and a gain offset in the angular velocity, as well as an error in the heading angle of at least one of the most recent first past position and the earliest first past position, and

the noise and errors are estimated by formulating an optimization problem and employing optimization techniques to determine estimates of the noise and errors.

11. The method in claim 8, wherein the calibration of the heading angles further comprises

determining whether the mobile object was moving along a straight road or a curvy road based on the angular velocity in the first past positions and

fitting the calibrated past heading angles to the heading angles in the first past positions, wherein

when the mobile object is determined to be moving along a straight road, the calibrated heading angles are determined based on the heading angles in both the most recent first past position and the earliest first past position in the selected segment,

when the mobile object is determined to be moving along a curvy road, the calibrated heading angles are determined based on calibrated angular velocities, wherein the calibration of the angular velocity comprises determining a gain offset in the angular velocity computing a change in heading angle based on bias-removed angular velocities, and comparing the change with a difference between the heading angle of the most recent first past position and the heading angle of the earliest first past position.

12. The method in claim 1, wherein the generation of the calibrated past trajectory further comprises an iterative process including:

selecting a segment of the first past trajectory;

computing a calibrated past trajectory based on the selected segment of the first past trajectory, wherein the calibrated past trajectory corresponds to same time instances as the segment of the first past trajectory and at least one of the second past positions is determined by using at least one of the first past positions that corresponds to an earlier time instance;

evaluating the trustworthiness of the calibrated past trajectory; and

outputting the calibrated past trajectory if the calibrated past trajectory is determined to be trustworthy in the evaluating step, or extending the segment of the first past trajectory to include more first past positions if the evaluating step determines the calibrated past trajectory is not trustworthy and continuing to next iteration to re-compute the calibrated past trajectory until one of the following criteria is met;

the calibrated past trajectory is trustworthy or a pre-defined number of iterations has been executed,

whereby when one of the criteria is met, a last one of the iterations is output as the calibrated past trajectory.

13. The method in claim 12, wherein the trustworthiness of the calibrated past trajectory is determined based on at least one of the following indexes:

a deviation index that indicates differences between the first past trajectory and the calibrated past trajectory;

a smoothness index that indicates smoothness of the calibrated past trajectory;

a first consistency index that indicates consistency between a first travel distance based on the first past trajectory and a second travel distance based on the calibrated past trajectory; and

a second consistency index that indicates consistency between a first heading angle change based on the first past trajectory and a second heading angle change based on the calibrated past trajectory.

14. The method in claim 12, wherein the generated calibrated past trajectory is further integrated with calibrated past trajectories generated at earlier time instances to form an integrated past trajectory.