THREE-DIMENSIONAL LIDAR-BASED CLEAR PATH DETECTION

• US 20100121577A1
• Filed: 10/19/2009
• Published: 05/13/2010
• Est. Priority Date: 04/24/2008
• Status: Active Grant

• Current Assignee: Carnegie Mellon University, GM Global Technology Operations LLC (General Motors Company)
• Original Assignee: Carnegie Mellon University, GM Global Technology Operations Incorporated (Motors Liquidation Co. GUC Trust)
• Inventors: Peterson, Kevin, Zhang, Wende
• Granted Patent: US 8,332,134 B2
• Time in Patent Office: Days
• US Class Current: 701/301
• CPC Class Codes: B60W 2420/403 Image sensing, e.g. optical camera
• CPC Class Codes: G01S 17/86 Combinations of lidar systems with systems other than lidar, radar or sonar, e.g. with direction finders
• CPC Class Codes: G01S 17/89 for mapping or imaging
• CPC Class Codes: G01S 17/931 of land vehicles
• CPC Class Codes: G06F 18/256 of results relating to different input data, e.g. multimodal recognition
• 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
• CPC Class Codes: G06V 20/588 Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road

Abstract

A method for detecting a clear path of travel for a vehicle includes generating a datastream corresponding to a three-dimensional scan of a target area surrounding the vehicle from a vehicle LIDAR system, estimating a ground plane for a present vehicle location using the datastream corresponding to the three-dimensional scan of the target area surrounding the vehicle, and comparing the datastream corresponding to the three-dimensional scan of the target area surrounding the vehicle with the estimated ground plane to detect a clear path of vehicle travel.

Claims

1. Method for detecting a clear path of travel for a vehicle, comprising

generating a datastream corresponding to a three-dimensional scan of a target area surrounding the vehicle from a vehicle LIDAR system;

estimating a ground plane for a present vehicle location using the datastream corresponding to the three-dimensional scan of the target area surrounding the vehicle; and

comparing the datastream corresponding to the three-dimensional scan of the target area surrounding the vehicle with the estimated ground plane to detect a clear path of vehicle travel.

2. The method of claim 1, wherein estimating the ground plane for the present vehicle location comprises executing a principal component analysis on the datastream corresponding to the three-dimensional scan of the target area surrounding the vehicle.

3. The method of claim 2, wherein executing the principal component analysis on the datastream comprises executing the principal component analysis on a dataset sampled from the datastream corresponding to the three-dimensional scan of the target area surrounding the vehicle.

4. The method of claim 3, wherein the principal component analysis on the sampled dataset is iteratively executed.

5. The method of claim 4, comprising comparing the sampled dataset with the estimated ground plane to detect the clear path of vehicle travel.

6. The method of claim 3:

wherein executing the principal component analysis on the sampled dataset comprises transforming the sampled dataset into first, second, and third eigenvectors, the third eigenvector having a least variance associated therewith compared to the first and second eigenvectors;

wherein estimating the ground plane comprises estimating the ground plane to correspond to datapoints from the transformed sampled dataset having projection coefficients corresponding to the third eigenvector that are less than a threshold value; and

further comprising identifying obstacles comprising datapoints of the transformed sampled dataset having projection coefficients corresponding to the third eigenvector that exceed a second threshold value.

7. The method of claim 1, further comprising:

generating data from a vehicle camera imaging system; and

wherein comparing the datastream further comprises comparing the datastream corresponding to the three-dimensional scan of the target area surrounding the vehicle with the data generated by the vehicle camera imaging system to detect the clear path of vehicle travel.

8. Method for detecting a clear path of travel for a vehicle, comprising:

generating a datastream corresponding to a scan of a target area surrounding the vehicle from a three-dimensional vehicle LIDAR system;

estimating a ground plane for a present vehicle location using the datastream corresponding to the scan of the target area surrounding the vehicle; and

detecting a clear path of vehicle travel by comparing the datastream corresponding to the scan of the target area surrounding the vehicle with the estimated ground plane.

9. The method of claim 8, wherein estimating the ground plane for the present vehicle location comprises executing a principal component analysis on the datastream corresponding to the three-dimensional scan of the target area surrounding the vehicle.

10. The method of claim 9, wherein executing the principal component analysis on the datastream comprises executing the principal component analysis on a dataset sampled from the datastream corresponding to the scan of the target area surrounding the vehicle.

11. The method of claim 10, wherein the principal component analysis on the sampled dataset is iteratively executed.

12. The method of claim 11, comprising comparing the sampled dataset with the estimated ground plane to detect the clear path of vehicle travel.

13. The method of claim 10, further comprising:

wherein executing the principal component analysis on the sampled dataset comprises transforming the sampled dataset into first, second, and third eigenvectors, the third eigenvector having a least variance associated therewith compared to the first and second eigenvectors;

wherein estimating the ground plane comprises estimating the ground plane to correspond to datapoints from the transformed sampled dataset having projection coefficients corresponding to the third eigenvector that are less than a threshold value; and

further comprising identifying obstacles comprising datapoints of the transformed sampled dataset having projection coefficients corresponding to the third eigenvector that exceed a second threshold value.

14. The method of claim 8, further comprising:

generating data from a vehicle camera imaging system; and

wherein comparing the datastream further comprises comparing the datastream corresponding to the scan of the target area surrounding the vehicle with the data generated by the vehicle camera imaging system to detect the clear path of vehicle travel.

15. Method for detecting a clear path of travel for a vehicle equipped with a LIDAR system configured to generate a datastream corresponding to a three-dimensional scan of a target area surrounding the vehicle, comprising:

executing a principal component analysis on the datastream corresponding to the three-dimensional scan of the target area surrounding the vehicle;

estimating a ground plane for a present vehicle location using the principal component analysis;

identifying obstacles using the principal component analysis and the estimated ground plane for the present vehicle location; and

detecting a clear path of vehicle travel associated with the estimated ground plane and the identified obstacles.

16. The method of claim 15:

wherein executing a principal component analysis on the datastream comprises executing the principal component analysis on a sampled dataset of the datastream, including transforming the sampled dataset into first, second, and third eigenvectors, the third eigenvector having a least variance associated therewith compared to the first and second eigenvectors; and

further comprising

associating the estimated ground plane with datapoints from the transformed dataset having projection coefficients corresponding to the third eigenvector that are less than a threshold value; and

identifying obstacles comprising datapoints of the transformed sampled dataset having projection coefficients corresponding to the third eigenvector that exceed a second threshold value.