Road profile prediction

• US 6,807,287 B1
• Filed: 08/25/2000
• Issued: 10/19/2004
• Est. Priority Date: 02/27/1998
• Status: Expired due to Term

• Current Assignee: TRW Automotive USLLC (Zeppelin - Stiftung der Stadt Friedrichshafen)
• Original Assignee: Lucas Industries Limited (Zeppelin - Stiftung der Stadt Friedrichshafen)
• Inventors: Hermans, Filip Jozef Johan
• Primary Examiner(s): Au, Amelia M.
• Assistant Examiner(s): Kim, Charles
• Application Number:
  US09/622,984
• Time in Patent Office: 1,516 Days
• Field of Search: 382/104, 382/106-107, 382/291, 382/293, 701/23, 701/28, 348/118
• US Class Current: 382/104
• CPC Class Codes: G06T 7/73 using feature-based methods
• 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

An apparatus for road profile prediction comprises digitization and storage means operable to digitize and store video image data relating to the road ahead of the vehicle and processor means operable to process the video image data to identify lane markings in the road, characterized in that the processor means includes estimator means operable to simultaneously analyze pairs of lines considered to constitute lane markings by a single recursive estimation procedure by finding the best fit to an expression involving left and right offsets, heading angle, lane curvature and horizon offset. A system embodying the invention has the purpose of overriding normal vehicle cruise control when another vehicle is detected in front of the vehicle.

Claims

1. Apparatus for road profile prediction ahead of a vehicle comprising:

a digitizer operable to digitize successive frames of video image data each containing feature points, including a screen image horizon, relating to the road profile ahead of the vehicle;

an inverse perspective transform operable to process a frame of digitized video image data with reference to a horizon position on the screen to generate representations of said feature points in real coordinates;

a mathematical lane marking model characterizing real geometrical constraints of the road by relating said real coordinates to a horizon error and road profile parameters including left and right lane offset values, lane heading, and lane curvature in terms of radius of curvature, said horizon error representing the difference between a position of the horizon on the screen obtained during calibration and the horizon position on the screen when processing said digitized video image data; and

an estimator, taking into account current and previous video frames, operable for estimating the best fit of said real coordinates to said mathematical lane marking model to derive said horizon error, and correct estimates for lane parameters including left and right lane offset values, heading angle, and lane curvature in terms of radius of curvature.

2. An autonomous cruise control system comprising an apparatus as claimed in claim 1.

3. A method for road profile prediction ahead of a vehicle comprising:

digitizing successive frames of video image data each containing feature points each including a screen image horizon, said frames relating to the road profile ahead of the vehicle in screen coordinates;

processing by way of an inverse perspective transform a frame of digitized video image data with reference to said screen image horizon position to generate a representation of said feature points in real coordinates;

storing a mathematical lane marking model representing the real geometrical constraints of the road, which model relates said real coordinates to a horizon error and to road profile parameters including left and right lane offset values, lane heading, and lane curvature in terms of radius of curvature, which horizon error represents the difference between a position of the horizon on the screen obtained during calibration and the horizon position on the screen when processing said digitized video image data;

finding the best fit of said second frame real coordinates to said mathematical lane marking model taking into account said horizon error; and

estimating, taking into account current and previous video frames, from said best fit correct estimates for said left and right lane offset values, heading angle, and lane curvature in terms of radius of curvature thereby to derive said road profile prediction ahead of said vehicle.

4. A method according to claim 3, comprising the step of correcting said real coordinates in dependence on said horizon error in said processing step of said inverse perspective transform, thereby taking into account said horizon error.

5. A method according to claim 3, comprising the step of correcting said real coordinates in dependence on said horizon error in said finding the best fit step, thereby taking into account said horizon error.

6. A method according to claim 3, wherein the estimating is performed by way of a single recursive estimation process.

7. A method according to claim 6, wherein said single recursive estimation process is a least squares parameter estimation process.

8. An autonomous cruise control system using a method as claimed in claim 3.