Neural network radar processor

Details Neural network radar processor Neural network radar processor

2000-08-11

2002-04-02

Blum, Theodore M. (Department: 3662)

Communications: directive radio wave systems and devices (e.g.,

With particular circuit

Digital processing

C342S118000, C342S200000

Reexamination Certificate

active

06366236

ABSTRACT:

TECHNICAL ART
The instant invention generally relates to systems and methods for detecting targets from radar signals and more particularly to systems and methods utilizing neural network processing to detect targets from a Frequency Modulated Continuous Wave (FMCW) automotive radar.
BACKGROUND OF THE INVENTION
A vehicle may contain automatic safety restraint actuators that are activated responsive to a vehicle crash for purposes of mitigating occupant injury. Examples of such automatic safety restraint actuators include air bags, seat belt pretensioners, and deployable knee bolsters. One objective of an automatic restraint system is to mitigate occupant injury, thereby not causing more injury with the automatic restraint system than would be caused by the crash had the automatic restraint system not been activated. Generally, it is desirable to only activate automatic safety restraint actuators when needed to mitigate injury because of the expense of replacing the associated components of the safety restraint system, and because of the potential for such activations to harm occupants. This is particularly true of air bag restraint systems, wherein occupants too close to the air bag at the time of deployment—i.e. out-of-position occupants—are vulnerable to injury or death from the deploying air bag even when the associated vehicle crash is relatively mild. Moreover, occupants who are of small stature, or with weak constitution, such as children, small adults or people with frail bones, are particularly vulnerable to injury induced by the air bag inflator. Furthermore, infants properly secured in a normally positioned rear facing infant seat (RFIS) in proximity to a front seat passenger-side air bag are also vulnerable to injury or death from the deploying air bag because of the close proximity of the infant seat's rear surface to the air bag inflator module.
Air bag inflators are designed with a given restraint capacity, as for example, the capacity to protect an unbelted normally seated fiftieth percentile occupant when subjected to a 30 MPH barrier equivalent crash, which results in associated energy and power levels which can be injurious to out-of-position occupants. While relatively infrequent, cases of injury or death caused by air bag inflators in crashes for which the occupants would have otherwise survived relatively unharmed have provided the impetus to reduce or eliminate the potential for air bag inflators to injure the occupants which they are intended to protect.
Known deployment systems for vehicle safety devices such as an air bag require the host vehicle to actually collide with an obstacle or other vehicle before the deployment decision process begins. At that point in time, the sensors detect a deceleration in the host vehicle and deploy one or more safety systems. Thus, the crash is identified based solely on the characteristic of the acceleration versus time measure. The disadvantage with existing post-crash detection systems derives from the fact that the time available to deploy an active safety device is relatively short, particularly for side impact or high speed frontal collisions where occupant restraint systems can provide significant safety benefits. These short time frames lead to rates of inflation of the air bags that are so great that injury or death are possible if the occupant is not properly situated with respect to the air bag.
Ideally, the air bag would be inflated prior to any interaction with a normally seated occupant, and at a rate which is sufficiently slow that an out of position occupant would not be injured by the inflating air bag. For a crash of sufficient severity, this requires the crash sensing system to be able to predict immanent crashes because the time required to inflate the bag at an inflation rate sufficiently slow to be safe for out-of-position occupants may be greater than either that required for the occupant to move so as to commence interaction with an inflated air bag or to safely decelerate the occupant.
Typically, predictive collision sensing systems utilize radar to sense the range and possibly velocity of a target relative to the vehicle. A radar system measures the distance and/or velocity of a target by sensing the effects of the interaction of the target with a beam of wave either continuous or pulsed energy, whereby the range to the target is determined by measuring the transit time of the radar signal, and the velocity of the target is determined by measuring the Doppler frequency shift of the received backscattered signal relative to the transmitted signal.
Frequency Modulated Continuous Wave (FMCW) radar generally operates by illuminating one or more targets with a constant amplitude transmitted signal which is frequency modulated over time. In a Linear FMCW (LFMCW) radar, the frequency modulation is linear with respect to time, or in other words, the transmitted signal is chirped—either up-chirped, down-chirped, or a combination of the two where the associated direction indicates the relative change in frequency with time. The transmitted signal is backscattered off each target back to an associated receiver, thereby undergoing a propagation delay which is proportional to the range of the target relative to the transmitter and receiver. For each point of reflection on each target, the corresponding received signal has a time varying frequency content similar to the transmitted waveform but shifted in time by a time delay proportional to the time required for the signal to travel from the transmitter to the target and then to the receiver. For a system with the transmitter and receiver relatively fixed with respect to one another, this time delay is also proportional to the range of the target thereto. Typically, the transmitter and receiver either share a common antenna, or utilize separate antennas which are in close proximity to one another. The received signal is recombined with either the transmitted signal, or a replica thereof as generated by a local oscillator, to produce a intermediate signal having a beat or intermediate frequency which is proportional to the target range. Alternately, the received signal may be separately combined with two different local oscillator signals, one of them being in-phase with the transmitted signal, the other being in phase quadrature thereto, which both effectively increases the sampling rate and enables a distinction with respect to the sign of the Doppler shift in the received backscattered signal in accordance with the associated target velocity. A sequence of received backscattered signals may be used to unambiguously resolve both target range and target velocity.
Some of the advantages of FMCW radar compared with other types of radar are 1) the modulation thereof is readily compatible with a wide range of solid-state transmitters, 2) the measurement of frequency necessary for range measurements can be performed digitally using a Fast Fourier Transform (FFT), and 3) the FMCW signals are relatively difficult to detect with conventional intercept receivers.
Traditional Linear Frequency Modulated (LFM) signal processing of a continuous wave (CW) radar signal requires a number of steps of processing including: 1) read the A/D converters and format the resulting data, 2) remove leakage, DC bias and in-phase/quadrature-phase imbalance in the received signal, 3) Fourier Transform processing, and 4) peak finding and detection and Constant False Alarm Rate (CFAR) detection.
This processing is inherently pipelined (serial) in time rather than parallel in time which leads to significant processing throughput requirements, depending on the LFM sweep update time. The only way to effectively parallelize the algorithm is to either context switch between tasks on a single processor that operates four times as fast as a single task requires or to have four dedicated processors, one for each task as seen in
FIG. 1
a
. Alternately as illustrated in
FIG. 1
b,
if a single time-slicing processor is used it requires four times the clock speed needed to execute a single t

Affiliated with

Also associated with

Rating

Neural network radar processor does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Neural network radar processor, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Neural network radar processor will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2829479