Method and apparatus for detecting a pedestrian impact

[0001] The present invention relates to pedestrian impact detection for a vehicle, and more particularly to a sensing method that provides timely and reliable detection of pedestrian impacts for which pedestrian safety devices should be deployed.

[0002] A vehicle can be equipped with deployable safety devices designed to reduce injury to a pedestrian struck by the vehicle. For example, the vehicle may be equipped with one or more pedestrian air bags and/or a device for changing the inclination angle of the hood. Since these devices are only to be deployed in the event of a pedestrian impact, the deployment system must be capable of reliably distinguishing pedestrian impacts from impacts with other objects. However, equipping a production vehicle with the required sensors can be both costly and difficult. Accordingly, what is needed is a way of detecting pedestrian impacts that is more practical and cost-effective.

[0003] The present invention provides an improved method and apparatus for detecting pedestrian impacts with a vehicle. Bend sensor segments are affixed to a vehicle body panel such as a bumper fascia and are responsive to deflection of the body panel due to impacts. The sensor data is processed to identify the location of an object impacting the body panel, and is correlated with calibration data to determine the shape and mass of the object. Impacts with pedestrians are discriminated from impacts with other objects based on the determined shape and mass.

[0010] Referring to FIG. 1, the reference numeral 10 designates a vehicle that is equipped with one or more pedestrian safety devices and a sensing system for deploying the safety devices when a pedestrian impact is detected. The pedestrian safety devices (PSDs) are designated by a single block 12, and may include one or more pedestrian air bags and a mechanism for changing the inclination angle of the vehicle hood. The PSDs 12 are selectively activated by a microprocessor-based electronic control unit (ECU) 16, which issues a deployment command to PSD 12 on line 18 when a pedestrian impact is detected. The ECU 16 detects pedestrian impacts based on inputs from a number of sensors, including a set of bend sensors 20a, 20b, 20c, 20d, 20e and a vehicle speed sensor 22 (which may be responsive to wheel speed, for example). Bend sensors 20a-20e (also known as flex sensors) are deflectable strip devices having an electrical resistance that varies in relation to the amount of their deflection. Suitable bend sensors are available from Flexpoint Sensor Systems, Inc., for example. In the illustrated embodiment, bend sensors 20a-20e are mounted on the inner surface of the front bumper fascia 24 to detect frontal pedestrian impacts. A similar set of bend sensors could additionally be mounted on the rear bumper fascia 26 or any other body panel that deflects on impact.

[0011] The block diagram of FIG. 2 illustrates functional elements of the ECU 16, including an archival memory 30 for storing calibration data, a correlation unit 32 and a discrimination unit 34. The calibration data stored in memory block 30 of FIG. 2 is obtained by collecting bend sensor data produced when the vehicle 10 collides with various test objects at various speeds. The test objects have different masses and different shapes such as round, flat and pointed. In general, the change in output signal level (i.e., the response) of the bend sensors 20a-20e increases with increasing object mass and impact speed, and the relationship among the sensor outputs varies with object shape. An impact is detected when the response of one or more of the bend sensors 20a-20e exceeds a threshold, and the correlation unit 32 determines the impact location according to the bend sensor segment 20a-20e having the highest response. The correlation unit 32 records the vehicle speed at the time of impact and characterizes segment-to-segment differences among the bend sensors 20a-20e. By correlating this data with the calibration data of memory 30, the correlation unit 32 additionally determines the object mass and shape data. The impact location, object mass and object shape are provided as inputs to discrimination unit 34, which determines if a pedestrian impact has occurred. In the event of a pedestrian impact, the discrimination unit 34 commands deployment of one or more PSDs 12 via line 18.

[0012] The calibration data stored in memory 30 is acquired during a series of controlled impacts at the various sensor locations along bumper fascia 24, with different test objects, and at different speeds. For each impact, two types of bend sensor data are recorded: the response of the bend sensor at the location of the impact (i.e., the on-location sensor), and normalized responses of the other bend sensors (i.e., the off-location sensors). The responses of off-location sensors are normalized by dividing them by the response of the on-location sensor. Finally, the response of the on-location sensor is recorded under the various speed and object shape constraints for objects differing in mass. For example, when a test object impacts the bumper fascia 24 at the location of bend sensor 20a, the highest response will occur at bend sensor 20a, and the other bend sensors 20b-20e will exhibit some change in output. All of the responses are recorded, and the off-location sensor responses are normalized with respect to the response of on-location sensor 20a. The normalized values are then stored for various combinations of vehicle speed and object shape. The mass of the object is then adjusted, and the response of the on-location sensor 20a for each object mass is recorded.

[0013] FIG. 3A represents the stored calibration data for impacts to bend sensor 20a in the form of a hierarchical look-up table. Similar data structures would exist for each of the other bend sensors 20b-20e. In the representation of FIG. 3, calibration data has been recorded at each of four different impact speeds (VS1, VS2, VS3, VS4), for objects having three different shapes (Round, Flat, Pointed) and two different masses (M1, M2). Of course, the number of speed, shape and mass variations can be different than shown. Normalized responses (NR) for off-location sensors (i.e, sensors 20b-20e) are stored for each combination of impact speed and object shape. FIG. 3B depicts a representative normalized response (NR) data set; as indicated, the responses R20b, R20c, R20d, R20e of the off-location sensors 20b, 20c, 20d, 20e are each divided by the response R20a of the on-location sensor 20a. The response (R) of the on-location sensor 20a is stored for each combination of vehicle speed, object shape and object mass.

[0014] The process of collecting the calibration data of FIG. 3A is summarized by the calibration routine 50 of FIG. 4. First, the block 52 records the bend sensor output signals and determines baseline signal values for each of the sensors 20a-20e, by calculating a moving average, for example. Then an object of specified shape and mass impacts a specified sensor location at a specified velocity (block 54) while the sensor signals are monitored (block 56). The block 58 identifies and stores the response (R) of the on-location bend sensor, and the block 60 calculates and stores a set of normalized responses (NR) for the off-location bend sensors. In each case, the response is the peak change in value of a sensor signal relative to the respective baseline signal value. The block 62 directs re-execution of the blocks 52-60 with respect to a different type of impact until the calibration process has been completed.

[0015] The flow diagram of FIG. 5 represents a software routine periodically executed by the correlation unit 32 of ECU 16 during operation of the vehicle 10. Initially, the block 70 is executed to determine baseline signal values for each of the sensors 20a-20e as described above in respect to block 52 of the calibration routine 50. The blocks 72 and 74 then monitor the sensor signals and compare the sensor responses to a predetermined threshold. If the threshold is not exceeded, the block 70 updates the moving averages used to establish the baseline signal values, and block 72 continues to monitor the sensor responses. When one or more sensor responses exceed the threshold, the blocks 76-92 are executed to determine and output the impact location, the object shape and the object mass. The block 76 sets the impact speed equal to the current value of vehicle speed VS. The block 78 identifies the on-location sensor as the sensor having the highest response, and the block 80 records the on-location sensor response (R). Optionally, the block 80 can also record the duration of the on-location response for correlation with corresponding calibration data. Then block 82 records a data set containing the normalized off-location sensor responses (NR).

[0016] The blocks 84 and 86 correlate the recorded sensor data with the stored calibration data to determine the object shape. Block 84 accesses all stored off-location calibration data for the sensor identified at block 78 and the impact speed recorded at block 76. Referring to the table representation of FIG. 3A, it will be assumed, for example, that sensor 20a has been identified as the on-location sensor and that the recorded impact speed is VS1; in this example, the correlation unit 32 accesses the normalized response (NR) data sets stored at 94, 96 and 98. Returning to the flow diagram of FIG. 5, the block 86 then correlates the off-location normalized responses recorded at block 82 with the accessed calibration data sets to determine the object shape. For example, if the recorded off-location normalized responses most nearly correlate with the normalized calibration responses stored at block 94 of FIG. 3A, the object shape is determined to be Round as signified by the table block 100.

[0017] Once the object shape has been determined, the blocks 88 and 90 correlate the recorded sensor data with the stored calibration data to determine the object mass. Block 88 accesses all stored on-location calibration data for the sensor identified at block 78, the impact speed recorded at block 76 and the object shape determined at block 86. Referring to the table representation of FIG. 3A, the correlation unit 32 accesses the response (R) data stored at table blocks 102 and 104 for the example given in the previous paragraph. Returning to the flow diagram of FIG. 5, the block 90 then correlates the on-location response recorded at block 80 with the accessed calibration data to determine the object mass. For example, if the recorded on-location sensor response most nearly correlates with the response stored at calibration table block 102 of FIG. 3A, the object mass is determined to be M1 as signified by the table block 106.

[0018] The routine of FIG. 5 concludes at block 92, which outputs the impact location, the object shape and the object mass to discrimination unit 34 of FIG. 2. The discrimination unit 34 uses predetermined rules to determine if the object shape and mass are representative of a pedestrian, or some other object such as a trash can or a bicycle. For example, a pedestrian impact can be detected if the object shape is round (possibly a pedestrian's leg) and the object mass is about 15 Kg. In addition to determining if the object is a pedestrian, the discrimination unit 34 can determine if and how PSD deployment should be activated based on the impact speed and impact location, for example.

[0019] In summary, the present invention provides a practical and cost-effective method and apparatus for detecting pedestrian impacts. While the invention has been described with respect to the illustrated embodiments, it is recognized that numerous modifications and variations in addition to those mentioned herein will occur to those skilled in the art. For example, the sensor response can be based on time rate of change or time at peak level, and so on. Accordingly, it is intended that the invention not be limited to the disclosed embodiment, but that it have the full scope permitted by the language of the following claims.