Method for demonstrating reliability of occupant classification systems
First Claim
1. A method for estimating reliability of a system which classifies an occupant of a vehicle seat as an adult or a child based on output signals of at least one sensor that measures a parameter that characterizes said occupant, by comparing measured parameter values to a selected threshold value, said method comprising:
- providing a first test object having a measurable parameter value that represents a lower boundary limit for an adult;
providing a second test object having a measurable parameter value that represents an upper boundary limit for a child, said upper boundary limit being smaller than said lower boundary limit for an adult;
testing each of a plurality of n said systems comprising a sample, using said first and second test objects to generate n first measured values and n second measured values, respectively;
calculating average and standard deviation of said first measured values;
calculating average and standard deviation of said second measured values;
determining for the first test object a first margin which is calculated as the surplus of average of said first values minus a constant times the standard deviation of the first values beyond said threshold value;
determining for the second test object a second margin which is calculated as the surplus of said threshold value beyond the average of the second values plus the constant times the standard deviation of the second values; and
concluding that said system has a desired minimum reliability at a desired level of confidence if both of said first and second margins are positive.
20 Assignments
0 Petitions
Accused Products
Abstract
In a method and apparatus for verifying that an occupant sensing system reliably classifies vehicle seat occupants as adults or children, a small number of systems is tested using test objects that have physical characteristics representative of a lower limit adult and an upper limit child. Average measured parameter values are calculated for each test object, as well as standard deviation. Margins are then calculated for each of the two test objects, using separate statistical formulas. If the margins for both objects are non-negative, it is concluded that the system demonstrates a minimum reliability, with a desired confidence level.
-
Citations
15 Claims
-
1. A method for estimating reliability of a system which classifies an occupant of a vehicle seat as an adult or a child based on output signals of at least one sensor that measures a parameter that characterizes said occupant, by comparing measured parameter values to a selected threshold value, said method comprising:
-
providing a first test object having a measurable parameter value that represents a lower boundary limit for an adult;
providing a second test object having a measurable parameter value that represents an upper boundary limit for a child, said upper boundary limit being smaller than said lower boundary limit for an adult;
testing each of a plurality of n said systems comprising a sample, using said first and second test objects to generate n first measured values and n second measured values, respectively;
calculating average and standard deviation of said first measured values;
calculating average and standard deviation of said second measured values;
determining for the first test object a first margin which is calculated as the surplus of average of said first values minus a constant times the standard deviation of the first values beyond said threshold value;
determining for the second test object a second margin which is calculated as the surplus of said threshold value beyond the average of the second values plus the constant times the standard deviation of the second values; and
concluding that said system has a desired minimum reliability at a desired level of confidence if both of said first and second margins are positive. - View Dependent Claims (2, 3, 4)
the method is performed initially with n=3, and is terminated if the desired minimum reliability is established;
if the desired minimum reliability is not established, n is incremented by 1 and the method is repeated; and
processing continues until the desired degree of reliability is established or until n reaches a predetermined maximum value.
-
-
5. A method for evaluating reliability of a system for classifying an occupant of a vehicle seat as between a first occupant category that is limited by a lower boundary value and a second occupant category that is limited by an upper boundary value smaller than the lower boundary value of the first category, based on output signals of at least one sensor for measuring at least one parameter which characterizes said occupant, said method comprising:
-
providing first and second test objects having physical properties that are representative of said lower boundary value of same first occupant category and said upper boundary value of said second occupant category, respectively, said physical properties being characterized by said at least one parameter;
providing a sample of n vehicle seats, each of said seats including an embodiment of the system whose reliability is to be estimated;
measuring a first value of said at least one parameter for said first object in each vehicle seat in said sample, and a second value of said at least one parameter for said second object in each vehicle seat in said sample;
calculating average and standard deviation of said first and second measured values;
determining for the first test object a first margin which is calculated as the surplus of the average of said first values minus a constant times the standard deviation of the first values beyond the threshold value for said classification;
determining for the second test object a second margin which is calculated as the surplus of said threshold value beyond the average of the second values plus the constant times the standard deviation of the second values; and
concluding that said system has a desired degree of reliability if both of said first and second margins are non-negative. - View Dependent Claims (6, 7, 8, 9, 10)
said first test object is an anthropomorphic test device having at least one physical property that is characteristic of a small adult; and
said second test object is an anthopomorphic test device having at least one physical property that is characteristic of a child.
-
-
9. The method according to claim 5, wherein n is less than or equal to eight, and the constant is determined as a function of n, Reliability and confidence.
-
10. The method according to claim 5, wherein:
-
the method is performed initially with n=3, and is completed if the desired degree of reliability is established;
if the desired degree of reliability is not established, n is incremented by 1 and the method is repeated; and
processing continues until the desired degree of reliability is established or until n reaches a predetermined maximum value.
-
-
11. An apparatus for evaluating reliability of a system for classifying an occupant of a vehicle seat as an adult or a child based on output signals of at least one sensor for measuring a parameter that characterizes said occupant, by comparing measured parameter values to a selected threshold value, said apparatus comprising:
-
a digital computer adapted to receive output signals from said sensors, and having signal processing means for converting said signals to a form that is readable by said computer;
a memory contained in said computer and having capacity to store measured parameter values from said sensors for each system in a sample of n systems that are to be tested, which signals characterize a first test object having a measurable parameter value that represents a lower boundary limit for an adult, and a second test object having a measurable parameter value that represents an upper boundary limit for a child, said upper boundary limit being smaller than said lower boundary limit for an adult; and
a computer readable medium contained in said computer and encoded with a program for causing said computer to perform the operations of;
calculating average and standard deviation for a set of first stored measured parameter values for said first test object, and for a set of second stored measured parameter values for said second test object;
determining for the first test object a first margin which is calculated as the surplus of the average of said first values minus a constant times the standard deviation of the first values beyond said threshold value;
determining for the second test object a second margin which is calculated as the surplus of said threshold value beyond the average of the second values plus the constant times the standard deviation of the second values; and
concluding that said system has a desired minimum reliability if both of said first and second margins are non-negative. - View Dependent Claims (12, 13, 14)
n is initially set at three, and if the desired degree of reliability is established, the operations are ended;
if the desired degree of reliability is not established, n is incremented by 1 and the computer repeats said operations; and
processing continues until the desired minimum reliability is demonstrated or until n reaches a preset maximum value, whichever comes first.
-
-
15. A method for evaluating whether a threshold value can be established for use in a system for reliably discriminating between adult and child occupants of a vehicle seat by comparing a measured weight value to said threshold value, said method comprising:
-
providing a first test object having a weight that corresponds to a lower boundary limit for an adult;
providing a second test object having a weight that represents an upper boundary limit for a child, said upper boundary value limit being smaller than said lower boundary limit for an adult;
testing each of a plurality of n said systems comprising a sample, using said first and second test objects to generate n first measured values and n second measured values, respectively;
calculating an average and a standard deviation of said first and second measured values;
calculating a first limit ease bound as the average of the first measured values minus a constant times the standard deviation of the first measured values;
calculating a second limit case bound as the average of the second measured values plus the constant times the standard deviation of the second measured values;
if the first limit case bound is greater than the second limit case bound, constructing a threshold whose value lies between the first and second limit case bounds will enable discrimination between adult and child seat occupants with a required degree of reliability.
-
Specification