Method and device for activating personal protection means

• US 8,374,752 B2
• Filed: 07/10/2007
• Issued: 02/12/2013
• Est. Priority Date: 08/16/2006
• Status: Active Grant

• Current Assignee: Robert Bosch GmbH
• Original Assignee: Robert Bosch GmbH
• Inventors: Breuninger, Joerg, Kolatschek, Josef
• Primary Examiner(s): Mawari, Redhwan K
• Application Number:
  US12/308,182
• Publication Number:
  US 20090306858A1
• Time in Patent Office: 2,044 Days
• Field of Search: 701/45, 701/408, 701/46, 701/47, 707/0.117, 707/45, 707/999.1, 280/735, 280/739, 280/734
• US Class Current: 701/45
• CPC Class Codes: B60R 21/0132 responsive to vehicle motion parameters , e.g. to vehicle longitudinal or transversal deceleration or speed value
• CPC Class Codes: B60R 21/01338 using vector analysis
• CPC Class Codes: B60R 21/0156 by deceleration

Abstract

A device and a method for activating a personal protection device are provided, in which a feature vector with at least two features is formed by an evaluation circuit from at least one signal of an accident sensor system. The evaluation circuit classifies the feature vector in the corresponding dimension using at least one class boundary. The activation circuit generates an activation signal which activates the personal protection device.

Claims

1. A method for activating a personal protection device in a vehicle, comprising:

forming, by a processor of a control unit in the vehicle, a feature vector with at least two features from at least one signal of an accident sensor system;

comparing, by the processor, the feature vector to each of at least one class boundary;

assigning, by the processor, the feature vector, in a corresponding dimension, to one of a plurality of classes based on where the feature vector is positioned with respect to the class boundaries as determined in the comparing step; and

activating the personal protection device as a function of the assigned classification;

wherein;

the at least one class boundary is non-linear and determined using at least one training vector and a kernel; and

the kernel is applied to reduce each at least one class boundary to a corresponding linear class boundary in the feature space.

2. The method as recited in claim 1, wherein the at least one class boundary is obtained from a data memory.

3. The method as recited in claim 1, wherein the classification is carried out in binary form.

4. The method as recited in claim 3, wherein, for the binary classification, at least a first decision tree is used, and wherein a respective binary classification is carried out in each branching point of the decision tree.

5. The method as recited in claim 4, wherein the personal protection device is assigned to the first decision tree.

6. The method as recited in claim 4, wherein a crash severity is assigned to a second decision tree and a crash type is assigned to a third decision tree.

7. The method as recited in claim 1, wherein, for the classification, a regression is used to generate a continuous value.

8. The method as recited in claim 1, wherein the at least two features are obtained from a time block.

9. The method as recited in claim 1, wherein the at least two features are formed from signals of different accident sensors of the accident sensor system.

10. A device for activating a personal protection device in a vehicle, comprising:

at least one interface configured to provide at least one signal of an accident sensor system;

a hardware evaluation circuit configured to;

form a feature vector with at least two features from the at least one signal;

compare the feature vector to each of at least one class boundary;

assign the feature vector, in a corresponding dimension, to one of a plurality of classes based on where the feature vector is positioned with respect to the class boundaries as determined in the comparing step; and

generate an activation signal as a function of the assigned classification; and

a hardware activation circuit configured to activate the personal protection device as a function of the activation signal;

wherein;

the at least one class boundary is non-linear and determined using at least one training vector and a kernel; and

the kernel is applied to reduce each at least one class boundary to a corresponding linear class boundary in the feature space.

11. A non-volatile hardware computer-readable storage medium storing a computer program having a plurality of codes executable by a computer processor and which, when executed by the computer processor, cause the computer processor to control a method for activating a personal protection device in a vehicle, the method comprising:

forming a feature vector with at least two features from at least one signal of an accident sensor system;

comparing the feature vector to each of at least one class boundary;

assigning the feature vector, in a corresponding dimension, to one of a plurality of classes based on where the feature vector is positioned with respect to the class boundaries as determined in the comparing step; and

activating the personal protection device as a function of the assigned classification;

wherein;

the at least one class boundary is non-linear and determined using at least one training vector and a kernel; and

the kernel is applied to reduce each at least one class boundary to a corresponding linear class boundary in the feature space.