System and method for generating a terrain model for autonomous navigation in vegetation

• US 7,995,837 B2
• Filed: 10/05/2009
• Issued: 08/09/2011
• Est. Priority Date: 06/02/2006
• Status: Active Grant

• Current Assignee: Carnegie Mellon University
• Original Assignee: Carnegie Mellon University
• Inventors: Stentz, Anthony J., Courville, Aaron C., Wellington, Carl K.
• Primary Examiner(s): Bella; Matt
• Assistant Examiner(s): RAHMJOO, MANUCHER
• Application Number:
  US12/573,459
• Publication Number:
  US 20100021052A1
• Time in Patent Office: 673 Days
• Field of Search: 382/154, 382/156, 382/157, 382/159, 701/90, 701/82, 701/38, 701/71, 701/41, 701/42
• US Class Current: 382/154
• CPC Class Codes: G05D 1/024 in combination with a laser lasers per se H01S
• CPC Class Codes: G05D 1/0242 using non-visible light signals, e.g. IR or UV signals
• CPC Class Codes: G05D 1/0251 extracting 3D information from a plurality of images taken from different locations, e.g. stereo vision stereoscopic image analysis H04N13/00; depth recovery from images G06T7/593
• CPC Class Codes: G05D 1/0274 using mapping information stored in a memory device navigation using map-matching G01C21/30
• CPC Class Codes: G06T 17/05 Geographic models
• CPC Class Codes: G06V 10/457 by analysing connectivity, e.g. edge linking, connected component analysis or slices
• CPC Class Codes: G06V 20/10 Terrestrial scenes scenes under surveillance with static cameras G06V20/52; scenes perceived from the exterior of a vehicle G06V20/56; scenes perceived from the interior of a vehicle G06V20/59
• CPC Class Codes: G06V 20/188 Vegetation
• CPC Class Codes: G06V 20/38 Outdoor scenes
• CPC Class Codes: G06V 20/64 Three-dimensional objects

Abstract

The disclosed terrain model is a generative, probabilistic approach to modeling terrain that exploits the 3D spatial structure inherent in outdoor domains and an array of noisy but abundant sensor data to simultaneously estimate ground height, vegetation height and classify obstacles and other areas of interest, even in dense non-penetrable vegetation. Joint inference of ground height, class height and class identity over the whole model results in more accurate estimation of each quantity. Vertical spatial constraints are imposed on voxels within a column via a hidden semi-Markov model. Horizontal spatial constraints are enforced on neighboring columns of voxels via two interacting Markov random fields and a latent variable. Because of the rules governing abstracts, this abstract should not be used to construe the claims.

Claims

1. A system for determining surface height, comprising:

a sensor array that produces data about a volume in which surface height is determined, said volume being divided into a plurality of voxels;

a plurality of surface modeling rules that provide domain information for interpreting data from said sensor array, for enforcing spatial constraints vertically among voxels, and for enforcing spatial constraints horizontally among columns of voxels; and

an inference engine responsive to said sensor array and said surface modeling rules so to classify each voxel, at least certain of said voxels being classified as surface;

wherein the classification of a voxel is based on the classification of neighboring voxels, wherein said surface modeling rules for enforcing spatial constraints horizontally include two interacting Markov random fields and a latent variable that interact across columns of voxels.

2. The system of claim 1 wherein said domain information is captured in a plurality of sensor models, one for each type of sensor in said sensor array.

3. The system of claim 1 wherein said sensor array includes a camera, an infrared sensor, and a range finder.

4. The system of claim 1 wherein said surface modeling rules for enforcing spatial constraints vertically comprises a hidden semi-Markov model imposed on vertically positioned voxels.

5. The system of claim 1 wherein said horizontal spatial constraints enforced by said surface modeling rules comprise assumptions of class continuity, surface smoothness and similarity of vegetation height.

6. The system of claim 1 wherein said classifying is carried out iteratively by said inference engine.

7. The system of claim 1 wherein said enforcing spatial constraints horizontally includes enforcing priors horizontally among columns of voxels.

8. A computer-implemented method for determining surface height, comprising:

producing, using a computer, sensor data about a volume in which surface height is determined, said volume being divided into a plurality of voxels; and

classifying each voxel using the computer, with at least certain of said voxels being classified as surface, by applying a plurality of surface modeling rules, said surface modeling rules providing domain information for interpreting sensor data, for enforcing spatial constraints vertically among voxels, and for enforcing spatial constraints horizontally among columns of voxels;

wherein the classification of a voxel is based on the classification of neighboring voxels, wherein said surface modeling rules for enforcing spatial constraints horizontally are implemented using two interacting Markov random fields and a latent variable that interact across columns of voxels.

9. The method of claim 8 wherein said domain information is captured in a plurality of sensor models, one for each type of sensor used to produce sensor data.

10. The method of claim 9 wherein said sensor models include a camera, an infrared sensor, and a range finder sensor models.

11. The method of claim 8 wherein said surface modeling rules for enforcing spatial constraints vertically are implemented using a hidden semi-Markov model on vertically positioned voxels.

12. The method of claim 8 wherein said horizontal spatial constraints enforced by said surface modeling rules comprise assumptions of class continuity, surface smoothness and similarity of vegetation height.

13. The method of claim 8 wherein said classifying is carried out iteratively.

14. The method of claim 8 wherein said enforcing spatial constraints horizontally includes enforcing priors horizontally among columns of voxels.