USING IMAGE SENSOR AND TRACKING FILTER TIME-TO-GO TO AVOID MID-AIR COLLISIONS

• US 20110210872A1
• Filed: 08/12/2009
• Published: 09/01/2011
• Est. Priority Date: 08/27/2008
• Status: Active Grant

• Current Assignee: Saab AB
• Original Assignee: Saab AB
• Inventors: Molander, Sören
• Granted Patent: US 8,970,401 B2
• Time in Patent Office: Days
• US Class Current: 340/961
• CPC Class Codes: G01S 11/12 using electromagnetic waves other than radio waves
• CPC Class Codes: G01S 3/7864 T.V. type tracking systems
• CPC Class Codes: G01S 5/16 using electromagnetic waves other than radio waves
• CPC Class Codes: G08G 5/045 Navigation or guidance aids, e.g. determination of anti-collision manoeuvers radar or analogous systems specially adapted for anti-collision between aircraft G01S13/933

Abstract

A collision avoidance system for deciding whether an autonomous avoidance manoeuvre should be performed in order to avoid a mid-air collision between a host aerial vehicle equipped with the system and an intruding aerial vehicle. At least one electro-optical sensor captures consecutive images of an intruding vehicle such that the vehicle manifests itself as a target point in the images. An image processor estimates the azimuth angle, elevation angle and a first time-to-collision estimate of the time to collision between the host vehicle and the intruding vehicle. The first time-to-collision estimate is estimated based on scale change in the target point between at least two of said consecutive images. A tracking filter is arranged to estimate a second time-to-collision estimate using the azimuth angle, the elevation angle and the first time-to-collision estimate estimated by the image processor as input parameters. A collision avoidance module is arranged to decide whether or not an avoidance manoeuvre should be performed based on any of at least one parameter, of which at least one is indicative of said second time-to-collision estimate.

Claims

1. A collision avoidance system for deciding whether an autonomous avoidance manoeuvre should be performed in order to avoid a mid-air collision between a host aerial vehicle equipped with the system and an intruding aerial vehicle, said system comprising:

at least one electro optical sensor arranged to capture consecutive images of said intruding vehicle, the vehicle of which manifests itself as a target point in said images,

an image processor arranged to estimate at least the azimuth and elevation angle from the host vehicle to the intruding vehicle based on the 2D position of the target point in said images, and a first time-to-collision estimate of the time to collision between the host vehicle and the intruding vehicle based on scale change in the target point between at least two of said consecutive images,

a tracking filter arranged to estimate a second time-to-collision estimate of the time to collision between the host vehicle and the intruding vehicle using said estimated azimuth angle, said estimated elevation angle and said first time-to-collision estimate as input parameters, and

a collision avoidance module arranged to decide whether or not the avoidance manoeuvre should be performed based on any of at least one decision parameter, of which at least one is indicative of said second time-to-collision estimate.

2. The collision avoidance system according to claim 1, wherein said tracking filter further is arranged to estimate the uncertainty in said second time-to-collision estimate, and wherein said collision avoidance module is arranged to make the decision based on said second time-to-collision estimate and said uncertainty.

3. The collision avoidance system according to claim 1, wherein said collision avoidance module is arranged to make the decision based on either or both of a parameter indicative of said second time-to-collision estimate and a parameter indicative of said first time-to-collision estimate.

4. The collision avoidance system according to claim 3, further comprising:

an uncertainty determining module configured to determine the uncertainty in the first time-to-collision estimate, said collision avoidance module being arranged to make the decision based on any of the second time-to-collision estimate, the second time-to-collision estimate uncertainty, the first time-to-collision estimate, the first time-to-collision estimate uncertainty, or any combination of these.

5. The collision avoidance system according to claim 1, wherein the collision avoidance module further comprises a selection module configured to analyze at least one parameter available from sensor measurements and/or estimates in the collision avoidance system, and select on which parameter or parameters the decision as to whether or not the avoidance manoeuvre should be performed should be based, based on the value(s) of said analyzed parameter(s).

6. The collision avoidance system according to claim 1, wherein said tracking filter is arranged to, in each filter iteration, use both the first time-to-collision estimate estimated by the image processor from the latest captured image and the second time-to-collision estimate estimated by the filter in the preceding filter iteration as input parameters when estimating said second time-to-collision estimate.

7. The collision avoidance system according to claim 1, further comprising:

an uncertainty determining module configured to determine the uncertainty in the first time-to-collision estimate, and

a timing module configured to determine when to start using said first time-to-collision estimate as input parameter to said tracking filter based on said uncertainty.

8. The collision avoidance system according to claim 1, wherein said tracking filter is an Extended Kalman Filter

9. The collision avoidance system according to claim 1, wherein said tracking filter is an Unscented Kalman Filter.

10. The collision avoidance system according to claim 1, wherein said tracking filter is a particle filter.

11. A method for deciding whether an avoidance manoeuvre should be performed in order to avoid a mid-air collision between a host aerial vehicle and an intruding aerial vehicle, said method comprising:

capturing consecutive images of said intruding vehicle, the vehicle of which manifests itself as a target point in said images,

estimating at least the azimuth and elevation angle from the host vehicle to the intruding vehicle based on the 2D position of the target point in said images, and a first time-to-collision estimate of the time to collision between the host vehicle and the intruding vehicle based on scale change in the target point between at least two of said consecutive images,

estimating, utilizing a tracking filter, a second time-to-collision estimate of the time to collision between the host vehicle and the intruding vehicle by using said estimated azimuth angle, said estimated elevation angle and said first time-to-collision estimate as input parameters to said filter, and

deciding whether or not the avoidance manoeuvre should be performed based on any of at least one decision parameter, of which at least one is indicative of said second time-to-collision estimate.

12. The method according to claim 11, further comprising:

estimating the uncertainty in said second time-to-collision estimate, and

deciding whether or not the avoidance manoeuvre should be performed being based on said second time-to-collision estimate and said uncertainty.

13. The method according to claim 11, wherein deciding whether or not the avoidance manoeuvre should be performed is based on either or both of a parameter indicative of said second time-to-collision estimate and a parameter indicative of said first time-to-collision estimate.

14. The method according to claim 13, further comprising:

estimating the uncertainty in said first time-to-collision estimate,

deciding whether or not the avoidance manoeuvre should be performed being based on any of the second time-to-collision estimate, the second time-to-collision estimate uncertainty, the first time-to-collision estimate, the first time-to-collision estimate uncertainty, or any combination of these.

15. The method according to claim 11, further comprising:

estimating the uncertainty in said first time-to-collision estimate, said uncertainty being used as indicator of when to start using the first time-to-collision estimates as input parameter to the tracking filter.