Vision-aided inertial navigation

• US 20090248304A1
• Filed: 03/23/2009
• Published: 10/01/2009
• Est. Priority Date: 03/28/2008
• Status: Active Grant

• Current Assignee: Regents of The University of Minnesota (University of Minnesota)
• Original Assignee: Regents of The University of Minnesota (University of Minnesota)
• Inventors: Mourikis, Anastasios I., Roumeliotis, Stergios I.
• Granted Patent: US 9,766,074 B2
• Time in Patent Office: Days
• US Class Current: 701/220
• CPC Class Codes: G01C 21/16 by integrating acceleration or speed, i.e. inertial navigation
• CPC Class Codes: G01C 21/165 combined with non-inertial navigation instruments
• CPC Class Codes: G01C 21/1656 with passive imaging devices, e.g. cameras
• CPC Class Codes: G05D 1/00 Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots drive control systems specially adapted for autonomous road vehicles B60W60/00
• CPC Class Codes: H04W 4/027 using movement velocity, acceleration information

Abstract

This document discloses, among other things, a system and method for implementing an algorithm to determine pose, velocity, acceleration or other navigation information using feature tracking data. The algorithm has computational complexity that is linear with the number of features tracked.

Claims

1. A system comprising:

a feature tracker configured to provide feature data for a plurality of features relative to a frame of reference in an environment for a period of time;

a motion sensor configured to provide motion data for navigation of the frame of reference in the environment for the period of time; and

a processor communicatively coupled to the feature tracker and communicatively coupled to the motion sensor, the processor configured to generate an estimate of navigation information for the frame of reference and further wherein the processor is configured to generate the estimate at a computational complexity that is linear with the number of tracked features.

2. The system of claim 1 wherein the processor is configured to generate the estimate such that the feature data is constrained by manipulating a residual of a feature measurement.

3. The system of claim 1 wherein each feature has a pose, and further wherein the processor is configured to generate the estimate simultaneously using each feature'"'"'s measurements to impose a constraint between the poses for a particular feature.

4. The system of claim 1 further wherein the processor is configured to manipulate a residual of the feature measurements to reduce an effect of a feature estimate error.

5. The system of claim 1 wherein the feature tracker includes a camera.

6. The system of claim 1 wherein the motion sensor includes an inertial measurement unit (IMU).

7. The system of claim 1 wherein the processor is configured to implement an extended Kalman filter algorithm.

8. The system of claim 1 further including an output device coupled to the processor, the output device including at least one of a memory, a transmitter, a display, a printer, an actuator, and a controller.

9. A method comprising:

receiving first data corresponding to time sampled navigation of a frame of reference relative to an environment, the environment having a number of features, the number greater than zero, the first data relative to the features, wherein the first data includes more than two samples corresponding to at least one feature;

receiving second data corresponding to the time sampled navigation of the frame of reference relative to the environment, the second data independent of the features; and

executing a filter algorithm to generate navigation information for the frame of reference relative to the environment using the first data and the second data, the filter algorithm having complexity linear in a number of features.

10. The method of claim 9 wherein the filter algorithm includes simultaneously using each feature'"'"'s measurements to impose constraints between the poses from which the feature was observed.

11. The method of claim 9 wherein the filter algorithm includes manipulating the residual of the feature measurements to reduce an effect of a feature estimate error.

12. The method of claim 9 wherein executing the filter algorithm to generate navigation information includes estimating at least one of position, attitude, velocity, or acceleration.

13. The method of claim 9 wherein executing the filter algorithm to generate navigation information includes estimating a pose.

14. The method of claim 9 wherein receiving the first data includes receiving data from a motion-inferring sensor.

15. The method of claim 14 wherein receiving data from the motion-inferring sensor includes receiving data from at least one of a camera-based sensor, a laser-based sensor, a sonar-based sensor, and a radar-based sensor.

16. The method of claim 9 wherein receiving the second data includes receiving data from at least one of a motion sensor, a motion-inferring sensor, and a motion tracking model.

17. The method of claim 16 wherein receiving data from the motion sensor includes receiving data from at least one of a wheel encoder, a velocity sensor, a Doppler radar-based sensor, a gyroscope, an accelerometer, an airspeed sensor, and a global positioning system (GPS) sensor.

18. The method of claim 16 wherein receiving data from a motion tracking model includes receiving data from a statistical model.

19. The method of claim 9 wherein executing the filter algorithm includes executing an Extended Kalman Filter (EKF).

20. A method comprising:

tracking a plurality of features relative to a frame of reference in an environment for a period of time;

receiving data for navigation of the frame of reference in the environment for the period of time; and

executing an algorithm to generate a state vector for the frame of reference relative to the plurality of tracked features, the algorithm having complexity that is linear in a number of features.

21. The method of claim 20 wherein executing the algorithm includes simultaneously using each feature'"'"'s measurements to impose constraints between the poses from which the feature was observed.

22. The method of claim 20 wherein executing the algorithm includes manipulating a residual of a feature measurement to reduce an effect of a feature estimate error.

23. The method of claim 20 wherein using the state vector to generate an estimate of the position of the particular feature includes estimating a position in three dimensional space.

24. The method of claim 20 wherein using the state vector to generate an estimate of a navigation information for the frame of reference includes using the state vector to generate an estimate of at least one of a pose and a higher order derivative of position and attitude with respect to time.

25. The method of claim 20 wherein tracking the plurality of features includes receiving data from at least one of a motion sensor and a motion-inferring sensor.

26. The method of claim 20 wherein receiving data for navigation includes receiving data from at least one of a motion sensor and a motion-inferring sensor.