DRIVER ASSISTANCE SYSTEM OR ROBOT WITH DYNAMIC ATTENTION MODULE

• US 20100305755A1
• Filed: 04/20/2010
• Published: 12/02/2010
• Est. Priority Date: 05/28/2009
• Status: Active Grant

• Current Assignee: Honda Research Institute Europe GmbH (Honda Motor Company)
• Original Assignee: Honda Research Institute Europe GmbH (Honda Motor Company)
• Inventors: HERACLES, Martin
• Granted Patent: US 8,364,311 B2
• Time in Patent Office: Days
• US Class Current: 700/253
• CPC Class Codes: G06V 10/462 Salient features, e.g. scale invariant feature transforms [SIFT]
• 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: G08G 1/16 Anti-collision systems road vehicle drive control systems for predicting or avoiding probable or impending collision otherwise than by control of a particular sub-unit B60W30/08
• CPC Class Codes: G08G 1/165 for passive traffic, e.g. including static obstacles, trees
• CPC Class Codes: G08G 1/166 for active traffic, e.g. moving vehicles, pedestrians, bikes

Abstract

The invention relates to a vision-based attention system, comprising:

• • at least one vision sensor,
  • at least one image processing module processing an output signal of the vision sensor in order to generate at least one two-dimensional feature map,
  • a dorsal attention subsystem generating a first saliency map on the basis of the at least one feature map, the saliency map indicating a first focus of attention for the driver assistance system,
  • a ventral attention subsystem, independent to the dorsal attention subsystem, for generating a second saliency map on the basis of at least one feature map, which can be the same as the one used for the dorsal attention system or a different one,
    the second saliency map indicating unexpected visual stimuli.

Claims

1. A vision-based attention system, comprising:

at least one vision sensor, —

at least one image processing module processing an output signal of the vision sensor in order to generate at least one two-dimensional feature map,

a dorsal attention subsystem generating a first saliency map on the basis of the at least one feature map, the saliency map indicating a focus of attention for the driver assistance system,

an expectation generation subsystem generating an expectation propagated to the dorsal attention subsystem influencing the focus of attention,

a ventral attention subsystem, independent to the dorsal attention subsystem, for generating a second saliency map on the basis of the at least one feature map, which is the same or a different one as the one used for the dorsal attention system, the second saliency map indicating an unexpected visual stimulus contradicting the expectation of the system, generating an interrupt event for the dorsal attention subsystem thereby enabling the system to reorient the focus of attention to the unexpected visual stimulus.

2. The system according to claim 1, wherein the unexpected visual stimuli are visual stimuli in the input field of the vision sensor which remain after deleting from the input field of the vision sensor

visual stimuli which are predicted from a previous visual input and the sensed ego-motion of the system, and —

pre-stored models of visual stimuli, which can be e.g. learned to be expected.

3. The system according to claim 1, wherein the feature map of the ventral attention subsystem represents the optical flow of the vision sensor, the optical flow being a two dimensional map in which each pixel is assigned a velocity vector representing a motion between two different sampled images.

4. The system according to claim 1, wherein the ventral attention subsystem has a priority over the attention produced in by the dorsal attention subsystem in case an unexpected stimulus is detected.

5. The system according to claim 1, wherein the unexpected visual stimuli in the second saliency map are detected based on the difference between the expected direction and the sensed direction of the optical flow.

6. The system according to claim 5, wherein the expected direction of the optical flow is a radial flow around a centre point in the input field of the vision sensor.

7. The system according to claim 1, wherein the dorsal attention subsystem comprises a bottom-up saliency module.

8. The system according to claim 1, wherein the dorsal attention subsystem comprises a top-down saliency module.

9. The system according to claim 1, wherein the output of the dorsal attention subsystem is supplied to a classifier module.

10. A driver assistance system, being provided with a attention system according to claim 1.

11. A vehicle, being provided with a driver assistance system according to claim 10, wherein the driver assistance system generates an output signal for an actuator of the vehicle, such as e.g. the steering, braking means and/or accelerating means.

12. An autonomous robot, being provided with an attention system according to claim 1.