Data processing: measuring – calibrating – or testing – Measurement system – Accelerometer
Reexamination Certificate
2002-04-30
2004-09-28
Bui, Bryan (Department: 2863)
Data processing: measuring, calibrating, or testing
Measurement system
Accelerometer
C702S142000, C702S150000, C244S003100
Reexamination Certificate
active
06799138
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to trackers that track targets and to an improved system for detecting when a tracker is no longer locked on the target.
BACKGROUND OF THE INVENTION
A target tracker is used in many military systems such as for guiding anti-aircraft missiles, so-called smart bombs, torpedoes, anti-tank ordinance, and the like. In all these systems the tracker obtains target location information at timed intervals and “locks on” the target. The designers and operators of potential targets attempt to break the lock on the target in a number of ways. Missiles may deploy dummy warheads or other decoys to confuse the tracker. A tank may avoid a heat seeking tracker by driving through a field of fires. Any of these techniques, among others, can cause a tracker to breaklock and track some object other than the target. Target trackers attempt to detect these decoys and evasive maneuvers, and upon detection revert to a different method of tracking. A difficulty in this process lies in determining or even estimating where the target may have gone in the time between when the tracker started tracking the wrong target and the time when the breaklock was detected. This time period is termed the latency period.
Prior attempts at breaklock detection are highly algorithm dependent. The general method is to measure the difference between some measured target attribute and the corresponding predicted value at each measurement interval. The predicted value is based on prior measurements. For example, a correlation tracker may use as its attribute the correlation coefficient between the reference image (from the past) and the current image. The correlation coefficient is compared to a threshold to determine the accuracy of the measurement. A segmentation-based tracker may use prior target intensities to pick out objects of similar intensity in the present image and deduce the accuracy of tracking on the basis of the consistency of segmentation. While the individual breaklock detection techniques differ, the common feature is that all decisions are made independently at each observation time. If and when a breaklock is detected, the prior art simply coasts with the estimated target rate at the time of breaklock detection and hopes that the estimated rate is not too badly corrupted.
As noted, current methods of breaklock detection are based on single measurements. Because of noise associated with measurements, these breaklock detection methods are generally effective only when the disturbance is catastrophic. Current methods of breaklock detection fail to detect breaklocks that occurs gradually. The failure to detect gradual breaklocks occurs because the disturbance between adjacent observation times is small compared to the noise in the measurement.
Another problem associated with current breaklock detection methods has to do with the corruption of the target state estimator. Because breaklock detection always occurs after the fact i.e., after a latency period, there is always a corruption of the estimated target state due to updating with erroneous measurements. When breaklock is not detected promptly, there is considerable corruption of the target state estimates. The incurred corruption leads to incorrect posting and diminishes the likelihood of reacquisition of the target.
In effect current systems base their prediction of where the target is based on where the target was at the time the system discovered it was no longer tracking the target. As explained below, the present invention bases its prediction of where the target is on where the target was at the last time its position was accurately known.
SUMMARY OF THE INVENTION
In one aspect the present invention provides a system that detects a breaklock condition by detecting anomalous acceleration of the putative target over multiple observation intervals. In another aspect, the present invention provides a projection of the location of a target based on uncorrupted target information acquired and stored before a breaklock is detected.
The system acquires target information from any existing or future device that provides a stream of target data representing a target position at timed intervals or “frames”. The target data is fed through a second order filter such as a Kalman filter, to reduce random noise and so produce estimated true target position and velocity data. The filtered position and velocity data are used to generate projected target data for the next frame.
Projected target data is then compared to the acquired target data for that frame, and the difference between the two, called a residual, is an initial measure of target acceleration. The residual is passed through a low pass filter to produce a filtered residual which is essentially an average of the residuals over a selected time period, generally a fairly short time period. The filtered residual is a good approximation of the instantaneous acceleration of the target and is sent to one input of an adder to generate a signal that determines whether a breaklock has occurred.
The adder receives as its other input a signal from a buffer that determines the median filtered residual over a selected number, N, of immediately preceding frames. When N is relatively large, the median filtered residual is a good approximation of the target's long term acceleration. This median is sent to the adder where the difference between the median residual is compared to the current value of the mean residual. By selecting the number, N, of frames over which the median is calculated and the time constant of the low pass filter, the system generates a difference between a fairly stable number (the median) and a short term average acceleration that approximates the instantaneous acceleration of the target. If the difference exceeds a set threshold, a breaklock has been detected. The threshold for triggering a breaklock determination may be made by selecting the maximum acceleration of which the target is capable divided by the gain of the low pass filter.
Once a breaklock has been detected, several things occur in the system. First, target positions determined by the tracker are no longer used to update the target states in the buffer. This is done by inhibiting the output from the second order filter to the buffer. Next the system looks in the buffer for the most recent uncorrupted data. It does this by looking backward in the buffer for first filtered residual that is on the opposite side of the median as the filtered residual of the frame where the breaklock was detected. Only the positions and velocities stored in the buffer from before this first opposite-side-of-the-median frame are used to project the target location.
Thus, the present invention detects a breaklock by comparing the long term average target acceleration to the instantaneous target acceleration. When a breaklock is detected, target position is re-estimated based on target position and velocity data accumulated prior to when the break lock occurred.
These and other features of the present invention will be clear to those of ordinary skill in the art from reading the following specification in conjunction with the accompanying drawing.
REFERENCES:
patent: 4224507 (1980-09-01), Gendreu
patent: 5044195 (1991-09-01), James et al.
patent: 5171933 (1992-12-01), Eldering
patent: 6265704 (2001-07-01), Livingston
patent: 2 737 578 (1997-02-01), None
patent: 09 236400 (1997-09-01), None
Lawrence Nikki J.
Lo Thomas K.
Moellenhoff Mark S.
Whorf Joshua A.
Bui Bryan
Raytheon Company
Renner , Otto, Boisselle & Sklar, LLP
LandOfFree
Breaklock detection system and method does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Breaklock detection system and method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Breaklock detection system and method will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3200194