Tightly-coupled GNSS/IMU integration filter having calibration features

• US 8,473,207 B2
• Filed: 10/21/2009
• Issued: 06/25/2013
• Est. Priority Date: 10/21/2008
• Status: Active Grant

• Current Assignee: Texas Instruments, Inc.
• Original Assignee: Texas Instruments, Inc.
• Inventors: Roh, June Chul
• Primary Examiner(s): Nguyen, Kim T
• Application Number:
  US12/582,846
• Publication Number:
  US 20100097268A1
• Time in Patent Office: 1,343 Days
• Field of Search: 701/500, 701/505, 701468-470, 701472-474, 701/479, 701/489, 701/510, 701/511
• US Class Current: 701/500
• CPC Class Codes: G01S 19/47 the supplementary measurement being an inertial measurement, e.g. tightly coupled inertial

Abstract

Embodiments of the invention provide a blending filter based on extended Kalman filter (EKF), which optimally integrates the IMU navigation data with all other satellite measurements (tightly-coupled integration filter). Two more states in the EKF for estimating/compensating the speed bias and the heading bias in the INS measurement are added. The integration filter has no feedback loop for INS calibration, and can estimate/compensate the navigation error in the INS measurement within the integration filter.

Claims

1. An apparatus comprising:

an integration filter for a sensor-assisted global navigation satellite system (GNSS) receiver

a GNSS measurement engine for providing GNSS measurement data to said integration filter;

an inertial measurement unit (IMU); and

an inertial navigation system (INS) block for calculating INS navigation information using a plurality of inertial sensor outputs,

wherein said integration filter performs a blending operation for combining said GNSS measurement data and said INS navigation information, and for estimating and compensating a speed bias and a heading bias in said INS measurement, wherein at least one of said speed bias and said heading bias is included in state variables of said integration filter and is compensated in calculating said blended GNSS measurement data and INS navigation information, and wherein said integration filter processes a plurality of INS user velocity data from said INS block in a measurement equation of said integration filter using a method comprising;

including a plurality of INS measurements in a local navigation coordinate in said measurement equation in a way that said INS measurements are a function of velocity variables and speed and/or heading bias variables of an integration filter state with a plurality of measurement noises.

2. The apparatus of claim 1, wherein a vertical INS measurement from said plurality of INS measurements is set to zero.

3. The apparatus of claim 1, wherein said plurality of INS measurements includes at least one speed measurement.

4. The apparatus of claim 1, wherein said plurality of INS measurements includes at least one heading measurement.

5. The apparatus of claim 1, wherein said plurality of INS measurements includes both at least one speed measurement and at least one heading measurement.

6. The apparatus of claim 1, wherein said integration filter does not integrate said INS measurements when GNSS measurements are reliable when GNSS signal condition is good.

7. The apparatus of claim 1, said integration filter does not update said speed bias and/or said heading bias when the GNSS measurement is unreliable.

8. The apparatus of claim 1, wherein a plurality of process noise variances for said speed bias and/or said heading bias are set to small numbers when a quality of the GNSS measurement is not good, so that state variables for said speed bias and/or said heading bias are effectively not changing over time.

9. The apparatus of claim 8, wherein a time instance for changing the process noise variances for a plurality of dead reckoning DR bias states are determined based on the quality of the GNSS measurement.