ROAD-TERRAIN DETECTION METHOD AND SYSTEM FOR DRIVER ASSISTANCE SYSTEMS

• US 20130079990A1
• Filed: 07/26/2012
• Published: 03/28/2013
• Est. Priority Date: 09/28/2011
• Status: Active Grant

• Current Assignee: Honda Research Institute Europe GmbH (Honda Motor Company)
• Original Assignee: Honda Research Institute Europe GmbH (Honda Motor Company)
• Inventors: KHNL, Tobias, FRITSCH, Jannik
• Granted Patent: US 9,435,885 B2
• Time in Patent Office: Days
• US Class Current: 701/41
• CPC Class Codes: B62D 6/007 adjustable by the driver, e.g. sport mode
• CPC Class Codes: G01S 13/89 for mapping or imaging
• CPC Class Codes: G01S 13/93 for anti-collision purposes
• CPC Class Codes: G01S 15/89 for mapping or imaging
• CPC Class Codes: G01S 15/93 for anti-collision purposes
• CPC Class Codes: G01S 17/89 for mapping or imaging
• CPC Class Codes: G01S 17/93 for anti-collision purposes
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/30256 Lane; Road marking
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/143 involving probabilistic approaches, e.g. Markov random field [MRF] modelling
• 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

The present invention describes a road terrain detection system that comprises a method for classifying selected locations in the environment of a vehicle based on sensory input signals such as pixel values of a camera image. The method comprises a high level spatial feature generation for selected locations in the environment called base points. The spatial feature generation of the base points is based on a value-continuous confidence representation that captures visual and physical properties of the environment, generated by so called base classifiers operating on raw sensory data. Consequently, the road terrain detection incorporates both local properties of sensor data and their spatial relationship in a two-step feature extraction process.

Claims

1. A road-terrain detection method for driver assistance systems, wherein the road-terrain detection method comprises

sensing the environment of a vehicle with at least one sensor;

transforming a sensor signal from the at least one sensor into at least one confidence map of local properties of the environment by using at least one base classifier;

generating spatial features for the local properties based on the at least one confidence map;

classifying locations in the environment of the vehicle to a certain category of road terrain based on the generated spatial features.

2. The method according to claim 1, wherein the at least one sensor is one or more cameras, radars, laser scanners or GPS\navigation systems.

3. The method according to claim 1, wherein based on the classified locations in the environment a visual or acoustic signal is generated or an effector of the vehicle, such as e.g. a steering wheel, an accelerator, or a brake, is operated.

4. The method according to claim 1, wherein the method applied for a specific road terrain such as “

road-like area”

, “

drivable road”

, “

ego-lane”

, “

non-ego-lane”

, “

non-drivable road”

, “

sidewalk”

, “

traffic island”

, or “

off-limits terrain”

is automatically parameterized by using positive and negative samples, which are given by training regions such as polygons.

5. The method according to claim 1 comprising a step of sensory preprocessing for all sensor signals to obtain a common representation of the environment,

wherein the sensory preprocessing is preferably an inverse perspective mapping of a camera image to get a metric image representation.

6. The method according to claim 1, wherein the generation of a spatial feature for a specific location in the environment of the vehicle comprises

extracting a ray, wherein a ray is defined as a directed line with a certain angular orientation starting from the location in the environment.

analyzing confidence values along the ray to extract spatial feature.

7. The method according to claim 6, wherein the analysis of the confidence values along the ray is performed by

integrating the confidence values along the ray;

extracting the ray length, at which the integral exceeds an absorption threshold, which is a certain numeric value.

8. The method according to claim 7, wherein for a specific location in the environment at least one ray and at least one absorption threshold is used to generate a feature vector that encodes the relative position to a local property given by a confidence map.

9. The method according to claim 1 comprising a base classifier for visual boundary classification, in order to find visual features that discriminate a road boundary, such as curbstones, from a road-like area, wherein the method

includes feature generation such as color, texture, appearance, flow, and depth

uses a positive training set with polygonal training regions for the road boundary

uses a negative training set with polygonal training regions for road-like area excluding areas with lane-markings.

10. The method according to claim 1 comprising a base classifier for road classification that generates a confidence map, wherein locations in the environment that correspond to a road-like area have high confidence values.

11. The method according to claim 1 comprising a base classifier for lane marking detection that generates a confidence map, wherein locations in the environment that correspond to lane markings have high confidence values.

12. A computer program product, executing a method according to claim 1 when run on a computing device in a land vehicle.

13. A road-terrain detection system for driver assistance systems, the road-terrain detection system being adapted to perform a method according to claim 1.

14. A road-terrain detection system for driver assistance systems, the road-terrain detection system being adapted to

sense with at least one sensor the environment of a vehicle;

transform at least one sensor signal from at least one sensor into at least one confidence map of local properties of the environment by using at least one base classifier;

generate spatial features for local properties based on the at least one confidence map;

classify locations in the environment of the vehicle to a certain category of road terrain based on the generated spatial features.

15. The system according to claim 14, wherein the road terrain detection system can select one out of multiple road terrain classification outputs as a system output, in order to internally have different subsystems for coping with

different weather conditions resulting in different visual properties;

different road types resulting in different physical and visual properties.

16. A land vehicle, being provided with a system according to claim 14.