Occupant Detection Method and Apparatus for Air Bag System

This invention relates to an occupant restraint system using an occupant detection device and particularly to an airbag system having seat pressure detectors in the seat.

The expanding use of supplemental inflatable restraints (SIRs) or air bags for occupant protection in vehicles increasingly involves equipment for the front outboard passenger seat. The driver side air bag has been deployed whenever an imminent crash is sensed. The position and size of the driver is fairly predictable so that such deployment can advantageously interact with the driver upon a crash. The passenger seat, however, may be occupied by a large or a small occupant including a baby in an infant seat. It can not be assumed that a passenger of any size is at an optimum position (leaning against or near the seat back). In a system designed for effective interaction with a full sized adult, an advantageous interaction with a small person may not be attained. In such cases it is preferred to disable the passenger side airbag when a small person occupies the seat or when the seat is empty.

It has been proposed in U.S. Pat. No. 5,474,327 to Schousek, entitled "VEHICLE OCCUPANT RESTRAINT WITH SEAT PRESSURE SENSOR", and in U.S. Pat. No. 5,732,375, issued Mar. 24, 1998 and assigned to the assignee of this invention, to incorporate pressure sensors in the passenger seat and monitor the response of the sensors by a microprocessor to evaluate the weight and weight distribution, and for inhibiting deployment in certain cases. These disclosures teach the use of sensors on the top surface of the seat, just under the seat cover, and algorithms especially for detecting the presence and orientation of infant seats. Both of these disclosures form a foundation for the present invention and are incorporated herein by reference. It is desirable, however to provide a system which is particularly suited for discriminating between heavy and light occupants and for robust operation under dynamic conditions such as occupant shifting or bouncing due to rough roads.

It is therefore an object of the invention to discriminate in a SIR system between large and small seat occupants for a determination of whether an airbag deployment should be permitted. Another object in such a system is to maintain reliable operation in spite of dynamic variations in sensed pressures.

A SIR system, as is well known, has an acceleration sensor to detect an impending crash, a microprocessor to process the sensor signal and to decide whether to deploy an air bag, and a deployment unit fired by the microprocessor. An occupant detection system can determine if an occupant or infant seat is positioned in a way to not benefit from deployment, and then signaling the microprocessor whether to allow or inhibit deploying the air bag.

A number of sensors, judicially located in the seat, can garner sufficient load and distribution information to allow determination of the occupant size. Each sensor is a very thin resistive device, having lower resistance as pressure increases. This information is then used to determine whether to inhibit airbag deployment. The sensors are arranged in groups in the seat. A microprocessor is programmed to sample each sensor, determine a total weight parameter by summing the forces, determine the forces on local groups of sensors, and averaging or filtering to provide several different measures of seat occupancy, each of which can be used determine whether to allow deployment.

The above and other advantages of the invention will become more apparent from the following description taken in conjunction with the accompanying drawings wherein like references refer to like parts and wherein:

FIG. 1 is a schematic diagram of a prior art SIR system incorporating a seat occupant detector;

FIG. 2 is a cross section of a seat equipped with pressure sensors, according to the invention;

FIG. 3 is a view of a seat support of FIG. 2 equipped with pressure sensors;

FIG. 4 is flow chart representing an overview of an algorithm for determining deployment consent according to the invention;

FIG. 5 is a flow chart representing a method of computing decision measures used in the algorithm of FIG. 4;

FIG. 6 is a flow chart representing a method of computing variable thresholds according to the invention;

FIG. 7 is a graphical representation of a function used in fuzzy logic for determining load ratings and a fuzzy measure;

FIG. 8 is a flow chart representing a method of computing an adult lock flag according to the invention;

FIG. 9 is a flow chart for deployment decision according to the invention; and

FIG. 10 is a flow chart representing a method of filtering allow and inhibit decisions according to the invention.

Referring to FIG. 1, a SIR system includes a SIR module 13 coupled to a seat occupant sensing system 14. The SIR module 13 includes an accelerometer 15 mounted on the vehicle body for sensing an impending crash, a microprocessor 16 for receiving a signal from the accelerometer and for deciding whether to deploy an air bag. An air bag deployment unit 18 is controlled by the microprocessor 16 and fires a pyrotechnic or compressed gas device to inflate an air bag when a deploy command is received. A fault indicator 20, also controlled by the microprocessor 16 will show a failure of the seat occupant sensing system 14.

It is the aim of the seat sensing system 14 to inhibit air bag deployment when a seat is empty or occupied by a small child, while allowing deployment when the occupant is large. For example, the system may be tuned to always inhibit deployment for occupants weighing less than 66 pounds, and always allow deployment for occupants exceeding 105 pounds. The seat occupant sensing system 14 comprises a microprocessor 22 having a 5 volt supply and an enabling line 24 periodically provided with a 5 volt enabling pulse, and a series of voltage dividers coupled between the enabling line 24 and ground. Each voltage divider has a fixed resistor 26 in series with a pressure sensor or variable resistor 28, and the junction point of each resistor 26 and variable resistor 28 is connected to an A/D port 30 of the microprocessor 22. The microprocessor 22 controls the pulse on enabling line 24 and reads each sensor 28 voltage during the pulse period. The microprocessor 22 analyzes the sensor inputs and issues a decision whether to inhibit air bag deployment and the decision is coupled to the microprocessor 16 by a line 32. The microprocessor 22 also monitors its decisions for consistency and issues a fault signal on line 34 to the microprocessor 16 if faults continue to occur over a long period.

Each fixed resistor 26 is, for example, 10 kohms and the variable resistors vary between 10 kohms at high pressure and 100 kohms at low pressure. Then the voltage applied to the ports 30 will vary with pressure. Each sensor comprises two polyester sheets each having a film of resistive ink connected to a conductive electrode, the two resistive films contacting one another such that the resistance between electrodes decreases as pressure increases. Such pressure sensors are available as ALPS pressure sensors from Alps Electric Co., Ltd., Tokyo, Japan.

FIG. 2 shows a seat cushion 36 having an upper surface 38 for holding an occupant, and a lower surface 40 seated on a rigid sheet or plastic form 42 which in turn is supported by a seat subassembly 44. The form 42, also shown in FIG. 3, holds a dozen pressure sensors 28 on its upper surface so that the sensors are pressed against the bottom surface 40 of the seat cushion 36. Automotive seat cushions assemblies do not normally have the form 42 but here it serves to hold the sensors 28 and to provide a reaction surface for the sensors, allowing each sensor to detect a force imposed by the weight of a seat occupant.

The method of operation is illustrated by a series of flowcharts wherein the functional description of each block in the chart is accompanied by a number in angle brackets <nn> which corresponds to the reference number of the block. The overall operation is shown in FIG. 4 wherein the sensor values are read by the microprocessor 22 <46> and the data is adjusted by bias correction and low pass filtering <48>. Once every 100 ms one sensor at a time is turned on and sampled. Then a bias calibrated for each sensor is subtracted from each sensor reading. Then all decision measures are computed <50> and decision algorithms are run <52>. The algorithm output is filtered to avoid the effects of transient events and ultimately a decision is made to allow or inhibit air bag deployment <54>. Then either an inhibit signal is issued <56> or an allow signal is issued <58>. The microprocessor executes the algorithm every 100 ms.

The computation of decision measures, as shown in FIG. 5, involves calculating total force and its threshold, sensor load ratings and measure, long term average of sensor readings and its threshold, the measure of each sensor group (right, left, etc.) and corresponding threshold, and a fuzzy measure of sensor readings. A fixed threshold is provided for the fuzzy measure and the load rating measure. The other thresholds are variable.

The variable threshold for a measure will slowly increase if the measure is above a selected minimum activity level (chosen for each measure) and will quickly decrease if the measure is below the level. Inhibit times are chosen for each measure to control the rate of increase or decrease; for increase the time T1 is preferably in the range of 30 to 300 seconds, and for decrease the time T2 is preferably less than 1 second. The threshold is allowed to vary between a minimum value and a maximum value. The variable threshold is calculated as shown in FIG. 6. For this and subsequent flowcharts the functional description of each block in the chart is accompanied by a number in angle brackets <nn> which corresponds to the reference number of the block. Inhibit times are selected for each measure. The inhibit times T1 and T2 for the particular measure is retrieved from memory <60>. If the measure is above the minimum activity level <62> and below the variable threshold <64>, and a timer is greater than T2 <66>, the threshold is incremented <68> and the timer is reset <70>. When the measure is less than the minimum activity level <62> and the timer exceeds T1 <72>, the threshold is decremented <74> and the timer reset <70>.

Referring again to FIG. 5, the total force is simply the sum of the sensor outputs. The load ratings are determined in the same way as in the above mentioned application Ser. No. 08/566,029 and as reflected in FIG. 7. There if a measure has a value lower than a it has a zero rating and if it has a value greater than b has a maximum rating, while intermediate values are linearly dependent on the measure. Thus each sensor is given a rating (fuzzy term) depending on its output and reflects the certainty that a load is present. The sum of the ratings gives the load rating measure. The long term average is calculated by 1) averaging all the sensor outputs in each sample period, 2) averaging all of the averages over, say, 16 sample periods, and then 3) long term filtering the result by passing the result through a low pass software filter with a 10 to 20 second time constant. The filter output is the long term average measure. Each group measure is the sum of sensor outputs for various groups of sensors such as a right group, left group, front group, rear group and central group.

The fuzzy measure is calculated by 1) applying the FIG. 7 function to the long term average measure to obtain a long term fuzzy value, 2) applying the FIG. 7 function to the load rating measure to obtain a load rating fuzzy value, and 3) calculating the product of the two fuzzy values.

FIG. 8 is a flowchart for processing an Adult Lock Flag which will be used is the main decision algorithm. The term "Adult" refers not to the age or maturity of an occupant but rather to a weight which is chosen to distinguish from a small child. When the Adult Lock Flag is set, the output decision will always be to allow deployment. The algorithm uses a lock threshold which is above the total force threshold range and an unlock threshold which represents an empty seat. It also uses a lock delay on the order of one to five minutes, and a lock timer which measures the time since vehicle ignition is turned on. If the decision filter 54 is at its maximum value, the total force is greater than the lock threshold, and the lock timer is larger than the lock delay <76>, a flag value is increased toward a maximum value <78> and the Adult Lock Flag is set <80>. If the decision at block 76 is No, it is determined whether the total force is above the unlock threshold <82> and if not, whether the total force is below the unlock threshold and the flag value is greater than zero <84>. If so, the flag value is decremented toward zero <86>, and in either case the flag value is tested <88>; if the value is above zero the Flag is set <80> and if the value is zero the Flag is cleared <90>.

The main decision algorithm 42 is shown in FIG. 9. Note that this algorithm will result in an allow or an inhibit decision, but this decision is preliminary, subject to subsequent filtering to obtain a final consent to deployment. Each measure is determined to be high or low by comparison with its variable threshold if one has been computed, or against a fixed threshold. The Adult Lock Flag is processed <92> according to FIG. 8 and if the Flag is set <94> an allow decision is made. If not, and the load rating is low <96> an inhibit decision is made. If the rating is not low the total force is tested <98, 100>. If high, an allow decision is issued and if low an inhibit decision is issued. If neither, it is determined whether the long term average measure <102> the load rating <104>, or a group measure <106> is high, and to issue an allow decision. Finally, if no decision has yet been made, an allow or inhibit decision is made on the basis of the fuzzy measure <108>.

The final judgment of whether to consent to deployment is made in the decision filter as shown in FIG. 10. An up and down counter starting at zero and having a maximum count of 255 is used. If an allow decision is made <42> the counter is incremented <110> and if an inhibit decision is made the counter is decremented <112>. When the count exceeds 133 <114> final consent to deployment is granted <116>; if consent is already present, a count over 123 is needed to maintain that state to afford hysteresis. When the count falls below 123 the consent is revoked and deployment will be inhibited. Assuming that the increment size is one count, at the 100 ms loop execution rate a minimum of 13.3 seconds will be required to issue the consent, and at least 25.5 seconds are needed to reach the maximum count needed to set the Adult Lock Flag. Similarly, once the maximum count is attained, at least 13.2 seconds are needed to revoke the consent.

It will thus be seen that process of determining whether an adult size person is occupying the seat is carried out by analyzing sensor output with several measures to insure both that deployment will be allowed with a large occupant and will not occur with a small occupant. Rapid detection of large adults is enabled by the total force and load rating measures, while dynamic sensor outputs caused by frequent occupant movement are managed by the long term average measure. The fuzzy measure helps discriminate between large and small occupants in borderline cases. The seat structure with sensors placed on the bottom surface of the seat cushion permits sensing of occupant weight without great sensitivity to localized forces on the top surface of the seat. Off center weight distributions caused by sitting on a seat edge or leaning in one direction are still detectable.