Infrastructure-free NLoS obstacle detection for autonomous cars

1. A method of non-line-of-sight (NLoS) obstacle detection for an ego vehicle, comprising:

capturing a sequence of images over a period with an image capture device;

storing the sequence of images in a cyclic buffer;

registering each image in the cyclic buffer to a projected image by estimating a homography H for each frame of the sequence of images to project to a view point of a first frame in the sequence of images and remove motion of the ego vehicle in the projected image;

enhancing the projected image through dynamic color amplification during daytime operation of the ego vehicle;

classifying an enhanced, projected image based on a scene determination; and

issuing a control signal to the vehicle upon classifying the projected image.

2. The method of claim 1 , wherein classifying the projected image comprises classifying the image as a dynamic scene.

3. The method of claim 1 , wherein classifying the projected image comprises determining a probability of a scene as one of static or dynamic.

4. The method of claim 1 , further comprising determining a region of interest (ROI) from the projected image.

5. The method of claim 1 , wherein the control signal comprises a command to halt a trajectory of the vehicle.

6. The method of claim 1 , wherein the projected image comprises a plurality of pixel values and classifying the projected image further comprises:

summing the pixel values of the projected image and comparing the summing to a threshold value, the classifying based on the threshold value.

7. The method of claim 6 , wherein the plurality of pixel values of the projected image are indicative of a change in illumination in the sequence of images.

8. The method of claim 1 , wherein classifying the projected image comprises determining a speed of an unseen object.

9. The method of claim 1 , wherein classifying the projected image comprises determining a direction of an unseen object.

10. The method of claim 1 , wherein the classifying the projected image comprises determining a size of an unseen object.

11. A non-transitory computer-readable medium having program code recorded thereon for non-line-of-sight (NLoS) obstacle detection of an ego vehicle, the program code being executed by a processor and comprising:

program code to capture a sequence of images over a period with an image capture device;

program code to store the sequence of images in a cyclic buffer;

program code to register each image in the cyclic buffer to a projected image by estimating a homography H for each frame of the sequence of images to project to a view point of a first frame in the sequence of images and remove motion of the ego vehicle in the projected image;

program code to enhance the projected image through dynamic color amplification during daytime operation of the ego vehicle;

program code to classify an enhanced, projected image based on a scene determination; and

program code to issue a control signal to the vehicle upon classifying the projected image.

12. The non-transitory computer-readable medium of claim 11 , wherein the control signal comprises a command to halt a trajectory of the vehicle.

13. The non-transitory computer-readable medium of claim 11 , wherein the projected image comprises a plurality of pixel values and the program code to classify the projected image further comprises program code to sum the pixel values of the projected image and comparing the sum to a threshold value, the classifying based on the threshold value.

14. The non-transitory computer-readable medium of claim 13 , wherein the plurality of pixel values of the projected image are indicative of a change in illumination in the sequence of images.

15. The non-transitory computer-readable medium of claim 11 , wherein the program code to classify the projected image comprises program code to determine a speed of an unseen object.

16. The non-transitory computer-readable medium of claim 11 , the program code to classify the projected image comprises program code to determine a direction of an unseen object.

17. The non-transitory computer-readable medium of claim 11 , wherein the program code to classify the projected image comprises program code to determine a size of an unseen object.

18. A system for non-line-of-sight (NLoS) obstacle detection in an ego vehicle, the system comprising:

an image capture device configured to capture a sequence of images over a period;

a cyclic buffer configured to store the sequence of images;

a registered buffer configured to register each image in the cyclic buffer to a projected image in the registered buffer by an image rectifier configured to estimate a homography H for each frame of the sequence of images to project to a view point of a first frame in the sequence of images and remove motion of the ego vehicle in the projected image;

an image processor configured to enhance the projected image through dynamic color amplification during daytime operation of the ego vehicle;

an image classifier configured to classify an enhanced, projected image based on a scene determination; and

a vehicle planner in communication with a vehicle controller, the vehicle planner configured to issue a control signal to the vehicle controller upon classifying the projected image.

19. The system of claim 18 , wherein the control signal comprises a command to halt a trajectory of the ego vehicle.

20. The system of claim 19 , wherein a shadow object is detected to trigger the command to halt the trajectory of the ego vehicle.