Cross traffic collision alert system

• US 8,232,872 B2
• Filed: 12/03/2009
• Issued: 07/31/2012
• Est. Priority Date: 12/03/2009
• Status: Active Grant

• Current Assignee: GM Global Technology Operations LLC (General Motors Company)
• Original Assignee: GM Global Technology Operations LLC (General Motors Company)
• Inventors: Zeng, Shuqing
• Primary Examiner(s): Hunnings, Travis
• Application Number:
  US12/629,962
• Publication Number:
  US 20110133917A1
• Time in Patent Office: 971 Days
• Field of Search: 340/425.5, 340/436, 340/435, 340/426.24
• US Class Current: 340/436
• CPC Class Codes: B62D 15/029 Steering assistants using warnings or proposing actions to the driver without influencing the steering system
• CPC Class Codes: G06F 18/256 of results relating to different input data, e.g. multimodal recognition
• CPC Class Codes: G06T 2207/20164 Salient point detection; Corner detection
• CPC Class Codes: G06T 2207/30261 Obstacle
• CPC Class Codes: G06T 7/246 using feature-based methods, e.g. the tracking of corners or segments
• CPC Class Codes: G06V 10/811 the classifiers operating on different input data, e.g. multi-modal recognition
• CPC Class Codes: G06V 20/58 Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads

Abstract

A cross traffic collision alert system is provided that includes an image-based detection device for capturing motion data of objects in a region of a direction of travel. A motion analysis module monitors the captured motion data. A salient region detection module identifies regions of interest within the captured motion data. A predictive object detection module is responsive to the outputs of the motion analysis module and the salient region detection module for generating at least one object candidate. Non-imaging object detection sensors sense a region for detecting the at least one object candidate. An object tracking module tracks the at least one object candidate and fuses the at least one object candidate transitioning between the regions. A threat assessment module determines a threat assessment of a potential collision between a tracked object candidate and the driven vehicle and determines whether a warning should be provided.

Claims

1. A cross traffic collision alert system comprising:

an image-based detection device for capturing motion data of objects in a region of a direction of travel of a driven vehicle;

a motion analysis module for monitoring the captured motion data, the motion analysis module identifying potential moving objects in the captured motion data;

a salient region detection module for identifying regions of interest within the captured motion data that indicates at least portions of potential moving objects, the salient region module detecting regions having varying characteristics within the captured motion data;

a predictive object detection module that is responsive to the outputs of the motion analysis module and the salient region detection module for generating at least one object candidate based on the potential moving objects identified by the motion analysis module and the salient region detection module;

non-imaging object detection sensors for detecting the at least one object candidate in a region sensed by the non-imaging object detection sensors;

an object tracking module for tracking the at least one object candidate detected by the predictive object detection module and by the non-imaging object detection sensors, the object tracking module fusing the at least one object candidate transitioning between the region sensed by the non-imaging object detection sensors and the region captured by the image-based sensing device, the object tracking module tracking a position and velocity of the at least one object candidate in relation to the driven vehicle, wherein fusing the at least one object candidate transitioning between the regions includes merging object data representing the at least one object candidate in the motion data with potential objects in the region sensed by the non-imaging object detection sensors; and

a threat assessment module for determining a threat assessment of a potential collision between a tracked object candidate and the driven vehicle, the threat assessment module determining whether a warning should be provided in response to the threat assessment.

2. The system of claim 1 further comprising in-vehicle sensors for sensing steering and movement of the driven vehicle, wherein the object tracking module predicts a driving path of the driven vehicle in response to sensed input signals from the in-vehicle sensors, the object tracking module providing feedback of the projected path of the driven vehicle to the predictive object detection module for comparison between a current time frame and a previous time frame of the potential moving object for assessing whether the at least one object candidate is potentially on a collision path with the driven vehicle.

3. The system of claim 1 wherein a plant model is used to determine the projected path of the driven vehicle.

4. The system of claim 1 wherein the object tracking module generates fused object data that includes position and velocity data.

5. The system of claim 1 wherein the predictive object detection module uses motion analysis and scene analysis for identifying salient regions of interest.

6. The system of claim 5 further comprising an image patch extraction module for extracting image patches that identifies potential objects.

7. The system of claim 6 further comprising a feature extraction module for determining a distribution of edges and corners of an image patch, wherein the determined distribution is mapped to a feature description vector.

8. The system of claim 7 further comprising an object detector module for classifying whether the feature descriptor vector represents an object.

9. The system of claim 1 wherein the object detection sensors include short range radar sensors.

10. The system of claim 1 wherein the object detection sensors include side blind zone warning sensors.

11. The system of claim 1 wherein the image capture device includes a video camera.

12. The system of claim 1 wherein the object tracking module uses a particle filter to fuse candidate objects.

13. The system of claim 1 wherein the region sensed by the non-imaging detection sensors and the region captured by the image based detection device are non-overlapping regions.

14. The system of claim 1 wherein the region sensed by the non-imaging detection sensors and the region captured by the image based detection device include an overlapping region, the object tracking module tracking the at least one candidate object by jointly using the non-imaging detection sensors and the image based detection device while the at least one candidate object is in the overlapping region.

15. A method for detecting a potential collision between a driven vehicle and an object crossing a path of the driven vehicle, the method comprising the steps of:

detecting a potential moving object using non-imaging object detection sensors in a region sensed by the non-imaging object detection sensors;

capturing motion data of objects in a region of a direction of travel of the driven vehicle by an image-based detection device;

identifying a potential moving objects from the captured motion data in the region captured by the image-based detection device;

identifying regions of interest within the captured image data that include potential moving objects;

applying a predictive object detection analysis that is responsive to the identified potential moving object and the identified regions of interest within the captured motion data for tracking the identified potential moving object potentially on a collision path with the driven vehicle;

fusing the potential moving objects transitioning between a region sensed by the non-imaging object detection sensors and the region captured by the image-based sensing device, the fusing of the potential moving objects transitioning between the regions of interest includes merging object data representing the potential moving objects in the motion data with potential moving objects in the region sensed by the non-imaging object detection sensors, wherein a position and velocity of the potential moving objects is tracked in relation to the driven vehicle;

determining a threat assessment of a potential collision between a tracked object and the driven vehicle; and

issuing a warning of a potential collision between an object and the driven vehicle in response to the threat assessment.

16. The method of claim 15 wherein a previous jointly tracked data for the objects is applied to the prediction object detection analysis.

17. The method of claim 15 wherein applying predictive object detection analysis uses motion analysis and scene analysis for identifying salient regions of interest.

18. The method of claim 17 wherein applying predictive object detection analysis includes extracting image patches for identifying potential objects.

19. The method of claim 18 wherein applying predictive object detection analysis includes determining a distribution of edges and corners in the image patch.

20. The method of claim 19 wherein the determined distribution is mapped to a feature description vector.

21. The method of claim 20 wherein applying predictive object detection analysis includes classifying whether the feature descriptor represents an object.

22. The method of claim 15 wherein the region sensed by the non-imaging detection sensors and the region captured by the image-based detection device are non-overlapping, wherein the object tracking module fuses the potential moving objects as it transitions between the non-overlapping regions.

23. The method of claim 15 wherein the region sensed by the non-imaging detection sensors and the region captured by the image based detection device are an overlapping region, wherein the object tracking module tracks the potential moving objects by jointly using the non-imaging detection sensors and the image based detection device while the potential moving objects is in the overlapping region.