Methods and devices for identifying the type of occupancy of a supporting surface

• US 7,200,475 B2
• Filed: 02/28/2005
• Issued: 04/03/2007
• Est. Priority Date: 08/29/2002
• Status: Expired due to Fees

• Current Assignee: Sartorius Lab Instruments GmbH & Co. KG (Sartorius AG)
• Original Assignee: Sartorius AG
• Inventors: Bettwieser, Erich, Maaz, Guenther
• Primary Examiner(s): Tran; Dalena
• Application Number:
  US11/066,479
• Publication Number:
  US 20050197754A1
• Time in Patent Office: 764 Days
• Field of Search: 701/45, 701/46, 180/271, 180/273, 180/268, 180/282, 280/735, 280/730.01, 280/730.02, 340/436, 340/438
• US Class Current: 701/45
• CPC Class Codes: B60R 21/01516 using force or pressure sensing means
• CPC Class Codes: B60R 21/01556 Child-seat detection systems

Abstract

A method and device for identifying the type of occupancy of a supporting surface, particularly of a motor vehicle seat (2), with the aid of force sensor-assisted signals. The sensor signals s_i, which are recorded at predetermined instants t_i, or quantities derived therefrom, are continuously stored in a memory (4) in such a manner that, the sensor signals from the recent past, or the quantities derived therefrom are available for analysis at any time, and that the type of occupancy is derived from these stored values using at least two independent calculation methods or using a neural network.

Claims

1. A method for identifying the type of occupancy of a supporting surface, using sensor signals, comprising:

storing the sensor signals recorded at predefined instants t_i, or quantities derived from the sensor signals, in a memory as stored values, st_i,

making available for evaluation the stored values st_i, from a recent past,

evaluating further sensor signals at a selected predefined instant t_i in conjunction with the stored values st_i, and

deriving the type of occupancy from the evaluation,

wherein the recording of the stored values st_i is started via an external start signal.

2. The method as claimed in claim 1, wherein the supporting surface is a motor vehicle seat and the sensor signals are emitted from sensors that sense forces on the motor vehicle seat.

3. The method as claimed in claim 2, wherein said storing of the sensor signals is performed continuously.

4. The method as claimed in claim 3, wherein said deriving of the type of occupancy comprises using at least two independent calculation methods or a neural network.

5. The method as claimed in claim 4, wherein the occupancy of the motor vehicle seat by an adult is distinguished from the occupancy by a child seat and a child.

6. The method as claimed in claim 4, wherein the type of occupancy is continuously monitored by continuously recording and evaluating the stored values st_i.

7. The method as claimed in claim 4, wherein the stored values st_i are stored in a shift register.

8. The method as claimed in claim 4, wherein an airbag system associated with the motor vehicle seat is deactivated when the type of occupancy is determined to be a child seat, and wherein the airbag system is activated when the type of occupancy is determined to be a person.

9. The method as claimed in claim 2, wherein the external start signal is supplied by an opening of a door of a motor vehicle containing the motor vehicle seat, and wherein the stored values st_i are continuously recorded at least until the seating process is complete.

10. A method for identifying the type of occupancy of a supporting surface, using sensor signals, comprising:

storing the sensor signals recorded at predefined instants t_i, or quantities derived from the sensor signals, in a memory as stored values, st_i,

making available for evaluation the stored values st_i, from a recent past, and

deriving the type of occupancy from these stored values,

wherein the supporting surface is a motor vehicle seat and the sensor signals are emitted from sensors that sense forces on the motor vehicle seat,

wherein said storing of the sensor signals is performed continuously,

wherein said deriving of the type of occupancy comprises using at least two independent calculation methods or a neural network,

wherein the stored values st_i are stored in a shift register, and

wherein at least the following steps are executed;

(a) reading out the stored values st_i at the instants t_i, i=1 . . . n, with a shift register number n from the shift register, forming measured quantities for an occupancy mass m(t_i) and a location of the center of gravity (x,y)(t_i), storing the measured quantities in the shift register, deleting the first value and shifting the shift register by one step,

(b) classifying the seat occupancy based on a plurality of classification states by means of occupancy probabilities W_k(t_i) determined from the content of the shift register using the at least two calculation methods or the neural network,

(c) calculating occupancy classes k=p for occupancy by a person and k=cs for occupancy by a child seat, which satisfy an occupancy probability condition W_k(t_i)>

W_lim, with a probability limit value W_lim, and storing the values in a classification window,

(d) counting up the classification window of the last n values starting from n toward the past and entering the values in a confidence algorithm,

(e) executing the confidence algorithm,

(f) repeating the steps (a) to (e) until a confidence coefficient Ck>

C_lim, where C_lim;

confidence limit, has been determined, or aborting when an abortion condition is satisfied,

(g) establishing the occupancy class k by means of the calculated confidence coefficient C_k for this class,

(h) providing a control signal as a function of the established occupancy class k.

11. The method as claimed in claim 10, wherein the classification occurs based on at least the following four classification states:

(A) CONSTANT,

(B) UNOCCUPIED,

(C) PERSON,

(D) CHILD SEAT.

12. The method as claimed in claim 10, wherein the confidence algorithm is executed by calculating the confidence coefficient according to the equation C_k=N_k/n for the classes k=p and k=cs from the corresponding number N_k of successive occupancy classes p and cs which satisfy the condition W_k(t_i)>

W_lim, taking into consideration the boundary conditions;

(a) reducing the confidence coefficient according to the equation C_k=(N_k−

1)/n, if a reclassification occurs at an instant t_i or if a classification “

NON-CLASSIFIABLE”

occurs,

(b) deleting a previous allocation of the classification window and setting C_k=0 if the classification state “

UNOCCUPIED”

is determined, and

outputting C_k if the classification state “

CONSTANT”

is determined.

13. The method as claimed in claim 10, wherein the shift register memory number is n=64 or greater.

14. The method as claimed in claim 10, wherein the probability limit W_lim=0.9 or greater.

15. The method as claimed in claim 10, wherein the confidence limit C_lim=0.75 or greater.

16. The method as claimed in claim 10, wherein an airbag system associated with the motor vehicle seat is deactivated via the control signal when the occupancy class k=cs is determined, and wherein the airbag system is activated when the occupancy class k=p is determined.

17. The method as claimed in claim 16, wherein the airbag system associated with the motor vehicle seat is deactivated via the control signal such that the airbag system, when triggered, is inflated with less pressure force when the occupancy class k=cs is determined, and the airbag system, when triggered, is inflated with full pressure force when the occupancy class k=p is determined.

18. The method as claimed in claim 10, wherein the occupancy mass m(t_i) measured as the occupancy of the motor vehicle seat is compared with a predefined minimum occupancy mass m_{p min}, and that the comparison result is taken into account in the classification.

19. The method as claimed in claim 18, wherein the minimum occupancy mass m_{p min}=32.5 kg.

20. The method as claimed in claim 10, wherein, during operation of the motor vehicle, acceleration forces caused by driving effects and acting on the vehicle seat are recorded via the sensor signals and are taken into account in the classification.

21. The method as claimed in claim 10, wherein influences by objects stowed under the motor vehicle seat and applying forces to the motor vehicle seat and influencing the sensor signals are taken into account in the classification.

22. The method as claimed in claim 10, wherein changes in the measured quantities δ

m, δ

x, δ

y as a result of long-term effects are taken into account in the classification.

23. The method as claimed in claim 10, wherein the occupancy detection of the motor vehicle seat is based on the detection of a seat occupancy pattern characteristic for the type of the seat occupancy by evaluating the memory contents using the neural network.

24. The method as claimed in claim 23, wherein the neural network comprises at least the following steps:

(a) entering mass characteristics over time m(t) using a normalized mass scale [0,1], step functions and typical child seat patterns, and generating a learning record,

(b) supplying the current n values from the memory,

(c) calculating output neurons for probabilities of the occupancy based on at least the following states;

CONSTANT

PERSON

CHILD SEAT

NON-CLASS CLASSIFIABLE and/or

REMOVAL.

25. The method as claimed in claim 10, wherein at least two independent calculation methods are provided.

26. The method as claimed in claim 25, wherein the occupancy probability W_k(t_i) is determined as a total occupancy probability from the employed individual calculation methods based on the equation

$W_k$

( t i ) = ∑

j = 1 u ⁢

c j ·

W k j ⁡

( t i ) ,

with c_j;

confidence factor of the method j, and u;

number of the calculation methods.

27. The method as claimed in claim 25, wherein the first calculation method is based on a structural feature detection of the seat occupancy by evaluating the memory contents using a regression analysis, and the second calculation method is based on the structural feature detection by evaluating the memory contents using an occupancy entropy analysis.

28. The method as claimed in claim 27, wherein the regression analysis comprises at least the following steps:

(a) determining a correlation coefficient and a step point t_step, STEP UP or STEP DOWN, from the analysis of the structural features of the seat occupancy by a person over n values,

(b) determining a correlation coefficient, a slope in mass per time and a step in mass by analyzing a STEP UP in proximity of t_step over a predefined number of v<

n values,

(c) counting up the number N_step of the STEPS UP, which are above the predefined minimum mass m_{p min} for an adult,

(d) calculating the occupancy probability for a person from the height of the step relative to a norm value Norm1 and the amount of the slope relative to a norm value Norm2 in a member function according to the relation;

W_p=⅓

*(step/Norm1)+⅓

*(slope / Norm2)+⅓

*δ

_N1,

with δ

_N1=1(for N_step=1) and δ

_N1=0(for N_step≠

1), and

(e) checking the structural features for constant occupancy.

29. The method as claimed in claim 27, wherein the occupancy entropy analysis comprises at least the following steps:

(a) normalizing the mass scale to [0,1],

(b) dividing the normalized mass scale into z<

n cells,

(c) counting up the number of the values w_i for each cell across the classification window,

(d) calculating the occupancy entropy based on the equation

$S=-\underset{i=1}{\overset{z}{\sum <br><br>}}<br><br>c_i<br><br>w_i<br><br>\ln <br><br><br><br>w_i,$

with c_i;

weighting, negative in proximity of the scale ends [0,1], higher weighting of the cells between the scale ends, normalizing to a total entropy S_total,

(e) calculating the probabilities for the seat occupancies according to the relations W_cs=S/S_total−

1, with k=cs for S/S_total>

1 and W_p=1−

S/S_total, with k=p for S/S_total <

1.

30. The method as claimed in claim 25, wherein an additional calculation is performed using a fast Fourier transform analysis of the memory contents to detect the structural feature of the seat occupancy.

31. The method as claimed in claim 25, wherein an additional calculation is performed using an autocorrelation analysis of the memory contents to detect the structural feature of the seat occupancy.

32. A system for identifying the type of occupancy of a supporting surface, comprising:

at least three force sensors or at least one axis of rotation in conjunction with at least one force sensor arranged on the supporting surface;

measurement electronics via which the at least three force sensors or the at least one axis of rotation in conjunction with the at least one force sensor measure forces acting on the supporting surface; and

a control unit that controls the system as a function of measured values as determined by the measurement electronics;

a memory in communication with the measurement electronics such that sensor signals recorded at predefined instants t_i, or quantities derived from the sensor signals, are continuously stored as stored values st_i, with the stored values st_i from the recent past available for evaluation at any time; and

a data processing unit that determines the type of occupancy by evaluating further sensor signals at a selected predefined instant t_i, in conjunction with the stored values st_i using at least two independent calculation methods or using a neural network,

wherein the recording of the stored values st_i is started via an external start signal.

33. The system as claimed in claim 32, wherein the memory is configured as a shift register.

34. The system as claimed in claim 32, wherein the system controlled is an airbag system for a motor vehicle.

35. The system as claimed in claim 32, wherein a result memory independent of a vehicle power supply is provided for storing data evaluated by the data processing unit.

36. The system as claimed in claim 32, wherein the external start signal is supplied by an opening of a door of a motor vehicle, and wherein the stored values st_i are continuously recorded at least until the occupancy process is complete.