Search Results

In this paper, we describe some important aspects of crash severity sensing for “smart” airbag implementation. In particular, gray zone considerations are described in detail. Another important issue discussed is the loss of the additional front sensor used for crash severity sensing and its impact on system performance. A third issue of great importance is the second stage (assume a dual-stage inflator) firing delay. Finally, a discussion of some inherent problems with the crash severity sensor approach are discussed as well as dynamic range/resolution problems that accelerometers can encounter when used in the crush zone of a vehicle. A summary with conclusions and recommended approaches is presented.

This paper expands on an earlier paper on the use of Magnetostrictive Sensors (MsS) for vehicle crash detection sensing. Analysis of vehicle crash data has shown promising results especially for use in side impact detection. Topics concerned with sensor system implementation are discussed, as well as the use of magnetostrictive sensors for frontal crash severity measurement and pedestrian impact applications.

This paper presents a new system for crash severity detection. The proposed system design recommends the use of a mechanical Ball-In-Tube (BIT) style sensor(s) in the front of a vehicle, and an accelerometer based system in the passenger compartment. An advanced algorithm accomplishes data fusion between the sensors. Results to date have been extremely favorable.

New sensor approaches termed magnetostrictive sensor (MsS) and nonlinear harmonics (NLH) for vehicle safety applications such as crash detection and occupant seat weight sensing are described. Both sensor approaches utilize the changes in the magnetic properties of ferromagnetic materials that occur when a stress (or a strain) is applied to the material. The MsS is a passive device suitable for vehicle crash sensing. The NLH is an active device suitable for occupant seat weight sensing. Technical features of these sensors are also discussed together with preliminary results of ongoing testing.

In this paper, testing techniques for electronic single point frontal crash sensing systems are presented. Particular attention is paid to the algorithms used by these modules. First, a description of a new performance measure called the Crash Detector Operating Characteristic (CDOC) is given. Using the Monte Carlo crash testing technique, a quantitative definition of this measure is described. Two new ways of testing single point modules or algorithms are presented. All of these techniques and approaches lead to objective performance measures. Conclusions and recommendations are drawn from the discussion of the new measures.

An occupant detection system must employ advanced signal processing techniques if it is to work throughout a full range of environmental variations, and make use of all potentially relevant information. This review begins with a discussion of the kinds of signals that one expects from an occupant position sensor described in [1]. Potential noise sources are then examined. The kinds of information that can be extracted from those signals are described. Finally, an example of the application of signal processing techniques to a complex problem is presented.

In this paper, tradeoffs and testing of frontal crash sensing systems are presented. First, an overall description of distributed and single point sensor systems is given. The differences between the two system configurations is presented. The sensor design would include an algorithm for a single point sensor. Tradeoffs associated with the various configurations and algorithm designs are developed to quantify the performance of any system. Testing techniques are then discussed to help determine the effectiveness of a given sensor system design. Finally, misconceptions throughout the industry are clarified in an objective manner.

In this paper we discuss possible applications and design considerations for an Occupant Position Sensing System (OPSS). First, the minimum functional requirements and possible advanced applications for an OPSS are presented. The relation of an OPSS to current and future deployment decision systems is discussed. We then consider the implications of the functional requirements for the choice of technology. Finally, we describe a prototype occupant position sensor, and display its position and motion tracking capability.

Many Algorithms have been developed recently to fire an airbag in a crash situation. These include algorithms based on velocity, energy, power, etc. All of these algorithms are based on physical principles but manipulated to perform the task of prediction. They have one or more of the following problems: 1) fire on low MPH crashes, 2) Do Not fire on time for pole, offset, etc. crashes, 3) Do Not fire on time for high MPH crashes, 4) Fire on various types of rough road scenarios, 5) Do Not reduce the effects of noise or EMI, 6) Do Not operate as effectively on deployment scenarios outside of the initial training set, 7) Are Not robust to the concatenation of the events listed above. Therefore, an algorithm is needed that can satisfy the above list. Since prediction is necessary to achieve these goals, an algorithm based on prediction should provide better performance than those currently in use. This paper presents the outline of such an algorithm.

The lack of a large number of crash data waveforms can limit the reliability of electronic Crash Detection Algorithms (CDAs). This paper discusses how statistics and the Monte Carlo (MC) method can be used to generate a large number of crash waveforms, and therefore increase CDA reliability. The MC method is used to model a crash waveform into two parts: 1) an underlying crash waveform, and 2) noise superimposed on the crash. The noise statistics are then varied and recombined with the underlying crash waveform to generate a large number of new crash waveforms. In addition Rough Road models were developed and concatenated with crash waveforms to better simulate real life. Finally a comparison between two CDAs was performed. The results show that one CDA is more robust than the other.