Position and velocity estimation system for adaptive weighting of GPS and dead-reckoning information

• US 5,416,712 A
• Filed: 05/28/1993
• Issued: 05/16/1995
• Est. Priority Date: 05/28/1993
• Status: Expired due to Fees

• Current Assignee: Trimble Navigation Limited (Trimble Inc.)
• Original Assignee: Trimble Navigation Limited (Trimble Inc.)
• Inventors: Geier, George J., Heshmati, Ardalan, Johnson, Kelly G., McLain, Patricia W.
• Primary Examiner(s): Teska, Kevin J.
• Assistant Examiner(s): NGUYEN, TAN QUANG
• Application Number:
  US08/068,983
• Time in Patent Office: 718 Days
• Field of Search: 364/450, 364/459, 364/449, 364/424.02, 340/988, 340/991, 340/992, 340/870.06, 342/357, 342/358, 342/352, 342/356, 342/457, 73/178 R
• US Class Current: 701/216
• CPC Class Codes: G01C 21/165 combined with non-inertial navigation instruments
• CPC Class Codes: G01S 19/49 whereby the further system is an inertial position system, e.g. loosely-coupled
• CPC Class Codes: G01S 19/52 Determining velocity

Abstract

An embodiment of the present invention is a combined GPS and dead-reckoning (DR) navigation sensor for a vehicle in which a pair of modifications are made to an otherwise conventional Kalman filter. Process noise is adapted to cope with scale factor errors associated with odometer and turning rate sensors, and correlated measurement error processing is added. When only two Doppler measurements (PRRs), or three with an awkward three-satellite geometry, are available, DR error growth can nevertheless be controlled. The measurement error correlations in the conventional Kalman filter covariance propagation and update equations are explicitly accounted for. Errors induced by selective availability periods are minimized by these two modifications.

Claims

1. A terrestrial vehicle position and velocity estimation system, comprising:

input interface means for accepting raw global position system (GPS) pseudo range (PR) and pseudo range rate (PRR) measurement data, GPS position estimate data, and GPS velocity estimate data from a GPS navigation receiver mounted to a terrestrial vehicle;

first computational means for mathematically modeling an error effect on said raw measurement data, position estimate data, and velocity estimate data due to periods of selective availability (SA) wherein range errors are deliberately introduced into GPS satellite transmission data to reduce the position accuracy of unauthorized GPS users;

second computational means for mathematically modeling a heading scale factor error effect on an accumulated heading error of a dead-reckoning turning rate sensor connected to said terrestrial vehicle;

third computational means for mathematically modeling an odometer scale factor error effect of a dead-reckoning odometer sensor connected to said terrestrial vehicle;

Kalman filter means connected to a respective output of each of the first through third computational means for blending data output from each of the first through third computational means and to produce and output an estimate of a heading error and the odometer scale factor error from at least one of said raw GPS PRR measurements and said GPS velocity estimate data; and

computer means connected to the Kalman filter means for computing a position and a velocity of said terrestrial vehicle from data obtained from said dead-reckoning turning-rate sensor and said dead-reckoning odometer sensor as individually corrected and weighted in the Kalman filter means for respective scale factor errors by partial PR and PRR measurement data periodically obtained from said GPS navigation receiver.

2. The system of claim 1, further comprising:

adaptive blending means for combining GPS position and velocity estimate data with information from said dead-reckoning sensors, wherein in said combining;

GPS velocity estimate data is used in a first single-state speed filter to estimate said odometer scale factor error;

said GPS velocity estimate data is used in a second single-state heading filter to remove a heading drift of said dead-reckoning turning rate sensor and errors caused by dead-reckoning turning rate sensor scale factor error, and said GPS position estimate data is used in one of a first two-state and a third and a fourth single-state filters connected such that their gains are coupled to said first single-state speed filter and said second single-state heading filter to correct an accumulated error of a dead-reckoned position;

each of said first and second single-state speed and heading filters add an extension to the Kalman filter means for correlated measurement error modeling and for including odometer and turning-rate sensor scale factor errors as process noise.

3. The system of claim 2, further comprising:

means connected to the Kalman filter means and for adapting said heading filter to a turning sensor anomalous behavior and including a dead-reckoning turning rate sensor heading drift monitor for operating on an output of a calibration filter to adjust a process noise model of said heading filter.

4. The system of claim 2, further comprising:

calibration means connected to the Kalman filter means and for calibrating the scale factor error associated with said turning rate sensor, including a digital filter for outputs of said heading filter for refining a set of initial estimates of sensor scale factors and separate clockwise and counterclockwise rotation filters.

5. The system of claim 2, further comprising:

isolation means connected to the Kalman filter means for rejecting GPS position and velocity estimate when a statistical rejection test indicates a substantial GPS error.

6. The system of claim 1, further comprising:

adaptive blending means connected to the input interface means, said dead-reckoning turning rate sensor and said dead-reckoning odometer sensor for combining individual GPS PR and PRR measurements with information from said dead-reckoning turning rate and odometer sensors, wherein in said combining;

individual GPS PRR measurements are processed by a four-state speed-and-heading filter which estimates heading error, odometer scale factor error, terrain slope, and receiver clock frequency error;

individual GPS PR measurements are processed by a four-state position filter for estimating three components of position error and receiver clock phase error, wherein said position filter gain calculation coupled to said speed and heading filter; and

said four-state speed-and-heading filter and said four-state position-and-receiver-clock-phase error filter each further include a modified basic Kalman filter measurement noise model for representing the effects of correlated process and measurement errors;

said position and speed-and-heading four-state filters replace a full eight-state Kalman filter to yield a factor of four improved computational efficiency; and

said position and speed-and-heading four-state filters reduce dead reckoning error growth when less than three GPS satellites are in view.

7. The system of claim 6, further comprising:

means connected to the Kalman filter means and for adapting said speed and heading filter to a turning rate sensor anomalous behavior and including a dead-reckoning turning rate sensor heading drift monitor for operating on an output of a calibration filter to adjust a process noise model of said speed and heading filter.

8. The system of claim 6, further comprising:

calibration means connected to the Kalman filter means and for calibrating said turning rate sensor scale factor error, including a digital filter for outputs of said speed and heading filter for refining a set of initial estimates of sensor scale factors and separate clockwise and counterclockwise rotation filters.

9. The system of claim 6, further comprising:

isolation means connected to the Kalman filter means and for rejecting GPS PR and PRR when a statistical rejection test indicates substantial GPS error.

10. The system of claim 1, further comprising:

adaptive blending means connected to the Kalman filter means and for combining individual GPS PR and PRR measurements with information from said dead-reckoning sensors and including two coupled four-state filters, wherein in said combining;

position and receiver clock phase error changes are estimated using an over-determined weighted least squares (WLS) solution based on dead reckoning position changes and available PR measurements from said GPS receiver; and

heading error, odometer scale factor error, terrain slope, and receiver clock frequency error are estimated using an over-determined WLS solution based on a dead reckoning velocity and available PRR measurements from said GPS receiver.

11. The system of claim 10, further comprising:

means connected to the Kalman filter means and for adapting said WLS speed and heading filter to a turning rate sensor anomalous behavior and including a dead-reckoning turning rate sensor heading drift monitor for operating on an output of a calibration filter to adjust the process noise model of said WLS speed and heading filter.

12. The system of claim 10, further comprising:

calibration means connected to the Kalman filter means and for calibrating a scale factor error associated with said turning rate sensor, including a digital filter for said WLS speed and heading filter for refining a set of initial estimates of sensor scale factors and separate clockwise and counterclockwise rotation filters.

13. The system of claim 1, further comprising:

continuous calibration means connected to the Kalman filter means and for correcting a turning rate sensor heading drift based upon one of a stationary measurement or a low-angular rate measurement using one of a fixed gain or Kalman filter.

14. The system of claim 1, further comprising:

gross-error recovery means for a system reinitialization to a next available GPS position and velocity estimate when a residual test of a WLS filter indicates a failure in dead reckoning.