SPEED DETECTION DEVICE
Field of the invention
This invention relates to a detection device for detecting the (relative) speed of an object within a field of view of the device.
Background of the invention
In order to detect the speed of an object within a field of view, the position of the object essentially needs to be monitored over time. Various detection devices are available for measuring the distance to a remote object, and thereby determine the relative position of an object, and these are generally referred to as range finding devices. Such devices may be used, for example, in vehicle collision avoidance systems. Optical range finding systems are well known examples. This invention relates to the use of a range finding device architecture in order measure the speed of the object.
One common approach to measuring distance to a far object is to measure the time of flight of a pulse of light from the measuring system to the far object and back again, and then to calculate the distance to the far object based upon the speed of light. Systems employing this method commonly employ a laser to generate the light pulse and so are known generically as "laser range finders" (LRF's) or "light detection and ranging" (LiDAR) systems. Typical applications are the measurement of altitude, target range or distance for survey applications in civil engineering and metrology. LRF's may be built as stand alone hand held units or embodied in larger systems.
A known LRF is shown in Figure 1 and comprises a laser 1, an optical transmission system 2, an optical reception system 3, a light sensitive detector 4, pulse detection circuitry 5, and timing calculation and display electronics 6.
in operation, the user initiates a measurement of range using input 7, which causes a laser fire pulse to be sent to the laser 1 and the laser to emit a pulse of light at time TO as represented by the plot 10. This pulse is focussed by the transmission optics 2 and travels to the remote object 8 where it is reflected. The receiving optics 3 collects a portion of the reflected light pulse illustrated as plot 12 and focuses the energy onto the light sensitive
detector 4. The detector 4 converts the received light pulse into an electrical signal and the pulse detector 5 discriminates against any electrical noise generated by the light sensitive detector to provide a clean, logic level pulse from the incoming light detector signal at time Tl.
This pulse is passed to timing calculation and display electronics 6 which calculates and displays the range to the remote object based upon the time of flight of the laser pulse (TI¬ TO) and the speed of light (c) in the intervening medium.
Often there are multiple pulses apparent within the reflected signal captured by the detector 4 due to reflections from a number of different objects (e.g. vegetation) in the path of the light pulse, or variations in the refractive index of the intervening atmosphere. The water particles present in a foggy atmosphere do not generate strong light reflections but instead generally introduce noise and cause signal attenuation.
PCT GB00/04968 discloses an optical distance measurement apparatus using a signal known as the Maximal Length Sequence (MLS). This is a family of pseudo random noise binary signal (PRBS) which are typically generated using a digital shift register whose input is generated from appropriate feedback taps. The maximal length sequence is the pseudo random noise sequence with the longest period which can be generated with a shift register of r sections. It has a length N=2r-1 shift register clock cycles and has good auto¬ correlation properties as the auto-correlation function has only two values; either -1/N or a peak of 1.0 at the point of correlation.
Figure 2 illustrates one example of a maximal length sequence generated by a four stage shift register 20. Alternative length sequences can be generated by using longer shift registers with the appropriate feedback taps.
The system of PCT GB00/04968 provides various advantages and refinements to the basic use of an MLS sequence and provides a low cost apparatus for distance measurement and which can function over long range, hi addition, the processing power required to operate the system is kept to a minimum.
One approach to the measurement of the speed of the remote object is simply to process successive distance measurements. This has the disadvantage of requiring multiple distance measurement operations.
Summary of the invention
According to the invention there is provided a speed detection device, comprising: a signal source for supplying a modulation signal; a transmission system connected to the signal source for transmitting a transmission signal modulated by the modulation signal; a reception system for receiving a received signal which is a reflected and delayed version of the transmitted signal; a cross-correlator for determining a cross correlation function between a time delayed version of the modulation signal and the received signal, for different values of the time delay; and means for analysing the cross correlation function to detect a peak in the cross correlation function representing a remote object, and wherein the shape of the cross correlation function peak is used to determine the relative speed of the remote object.
The invention thus enables the hardware of range finding device to be used to detect the relative speed of a remote object, based on a single cross correlation function operation.
The transmitted signal is preferably an optical signal, and the invention then relates to optical range finding apparatus.
The modulation signal preferably comprises an MLS sequence, and which has a bit period which is an integer multiple a master clock bit period. The modulation signal preferably comprises a sequence which is repeated a plurality of times. By using a repeating sequence, as the remote object moves relatively to the system, the cross correlation function peak will spread, as different parts of the multiple sequence signal have maximum correlation at slightly different times. This spreading can be used to determine the relative speed.
The reception system preferably comprises an analogue to digital converter clocked at the master clock bit rate.
The shape of the peak which is analysed is preferably the width of the peak and the sequence (MLS) bit period.
The cross-correlator may comprise: a coarse cross-correlator for coarsely determining the time delay of the modulation signal needed to maximise the correlation between the time delayed modulation signal and the received signal, and a fine cross-correlator for calculating the correlation between the modulation signal and the received signal as a function of the time delay of the modulation signal with respect to the received signal in a time delay range around the time shift determined by the coarse cross-correlator.
This approach enables a reduction of processing power in order to determine the exact location of cross correlation peaks.
The invention also provides a method of detecting the speed of a remote object, comprising: supplying a modulation signal, transmitting a transmission signal modulated by the modulation signal, receiving an signal which is a reflected and delayed version of the transmitted signal, determining a cross correlation function between a time delayed version of the modulation signal and the received signal, for different values of the time delay; analysing the cross correlation function to detect a peak in the cross correlation function representing a remote object; and analysing the shape of the cross correlation function peak to determine the relative speed of the remote obj ect.
Brief Description of the Drawings
An example of the invention will now be described in detail with reference to the accompanying drawings, in which: Figure 1 shows a known laser range finding apparatus;
Figure 2 shows circuitry for generating a maximal length sequence; Figure 3 shows a schematic diagram optical distance measuring equipment using a time delay measurement technique which can be used by the system of the invention;
Figure 4 shows a signal generated by the distance measuring equipment in Figure 3;
Figure 5 shows a schematic diagram of a second embodiment of optical distance measuring equipment which can be used by the system of the invention;
Figure 6 shows the effect on movement on the cross correlation function; and Figure 7 shows how a single cross correlation function can be used to determine the relative speed of a remote object.
Detailed description of the invention
The invention uses an optical distance measurement arrangement using the cross- correlation between a time delayed version of a modulation signal and a received reflected version of the modulation signal. This cross correlation function is analysed to detect relative speed of as well as the distance to a target. The invention enables fog or other airborne particulate material to be detected.
One example of optical distance measurement equipment using an MLS technique with cross correlation will first be described, and which can be used to implement the invention. This system is described further in WO 01/55746, which is incorporated herein by reference.
The operation of a simplified system is first described with reference to Figure 3 and 4, and a more detailed system is then described with reference to Figure 5 and which can be used to provide speed measurement from a single cross correlation analysis.
In operation of the system of Figure 3, the user initiates a measurement of range at input 32 which causes an MLS generator 34 to generate an MLS signal. The MLS generator clock signal is derived from the system master clock Fmck 36 by divider 38 so that the MLS clock frequency FmIs is a known sub-multiple M of the master clock signal. In effect, the MLS is stretched in time by factor M. The "stretched" MLS signal causes the laser 1 to emit an optical stretched MLS signal starting at time TO, as represented at 40. This optical signal is focussed by the transmission optics 2 and travels to the remote object 8 where it is reflected. The receiving optics 3 collects a portion of the reflected optical signal and focuses this energy onto a light sensitive detector 4. This detector converts the collected light signal into an electrical signal which is digitised by the analogue to digital converter 42 and passed to coarse 44 and fine 46 cross-correlation calculation units.
The digital to analogue converter sample clock is set equal to the system master clock frequency. In this way, an oversampling D/A conversion is implemented, and the oversampling ratio M is used to interpret the results as will become apparent from the description below.
The coarse cross-correlation unit 44 is clocked at the MLS clock frequency FmIs and hence correlates a sub-sampled version of the digitised reflected MLS signal and original stretched MLS transmitted signal. The output from this cross correlation unit is a peak which is detected by pulse detector 48 and which indicates the coarse time delay TcI of the reflected signal. The fine cross-correlation unit 46 is clocked at the master clock frequency Fmck.
The control electronics 50 then causes the fine cross-correlator 46 to calculate the cross- correlation of the transmitted and reflected signals only in the region of time delay TcI. Typically, the fine cross-correlation function would be calculated for 2M samples before and after TcI. The output of the fine cross correlator is the cross correlation function of the transmitted and reflected signals in the region of the peak as shown in Figure 4, where M=4.
The shape of the correlation peak for a PRBS signal such as an MLS is a triangular pulse. This may be understood by viewing the MLS as the summation of a sequence of N
identical pulses, each of width T=l/Fmls and appropriately delayed and summed together.
The cross-correlation operation may be viewed as being similar to convolving the MLS with a delayed version of itself and then sampling the result at a frequency equal to the cross correlator clock frequency.
Therefore, the shape of the correlation peak output by the cross-correlation unit is given by the convolution function of two identical pulses of width T, which is a triangular pulse of width 2T = 2/Fmsl, sampled by the cross correlator clock frequency, which for the fine cross correlator is Fmck = M x FmIs. Hence the cross correlation function output by the fine cross correlator 46 takes the form shown in Figure 4.
The x-axis in Figure 4 is the master clock sample number, and this can be converted into time. The width of the peak at half height is equal to the cross correlator clock sampling period, or M times the master clock period, hi the example of Figure 5, this is 4 master clock cycles.
This signal is passed to the timing calculation and control electronics which calculates using known standard techniques the coefficients mi and k\ for equation of the best fit line through the M samples prior to the peak of the signal :
and the coefficients m
2 and k
2 for equation of the best fit line through the M/2 samples after the peak of the signal :
These lines are shown plotted on figure 6. The timing, calculation and display electronics then compute the intersection T0 of the two best fit lines from:
_ (k2 —kλ )
0 ~ ( \
where T0 is an estimate of the time of the peak of the signal which equates to the time delay between the transmitted and reflected signals.
The distance to the object is then calculated from the determined time To; it is half the speed of light multiplied by the time taken
The system described above has particular advantages which may be seen by comparison with an MLS system just using one correlator. Assume such a system is constructed using an MLS of order 10, a master clock period of 3OnS and a delay step size equal to one fifth of the MLS clock sample frequency. As described above, the total number of calculations required to compute the full cross-correlation for one MLS signal is 10232 or 1046529 operations. Thus to determine the position of the cross correlation peak to within one master clock period (or 5m) is 1046529 operations. To improve the precision further, and assuming the signal to noise ratio is sufficient, up to five iterations may be required to step the delay to find the precise position of the correlation peak so a worst case total of 5 x 10232 = 5232645 calculations is required to achieve the best precision, which is one fifth of a master clock cycle = 6nS, giving a best distance precision of Im.
Compare the proposed MLS system in figure 3 and assume that it operates with the same master clock period, the ratio of master clock frequency to MLS frequency M=8 and that the MLS is order 7. This gives a MLS duration of (27 - 1) = 127 which stretched by the factor M=8 to 1016 master clock cycles (comparable in duration to the system of figure 4).
The coarse correlator 44 is clocked at the MLS frequency and hence the total number of calculations required to compute the coarse correlation is 1272 or 16129 operations. The fine correlator only needs to compute the convolution peak in the region of the coarse peak. As M=8, we will assume that 16 full length correlation calculations are required so a total of 16 x 1016 = 16256 operations are required. At this point, the position of the correlation peak is known to within one master clock cycle but only 16129+16256 = 32385 operations have been required to fmd this, rather than 10232 or 1046529 operations for known systems. Thus it can be seen that the use of a stretched MLS has enabled a two step approach to be taken to computing the cross-correlation function of reflected pulse which yields a substantial (in this case 32 fold) reduction in computing requirements allowing the proposed invention to be implemented on much simpler and lower cost hardware.
In addition, the system uses a prior knowledge of the triangular form of the MLS cross correlation function in combination with the stretched form of the MLS to allow the time To of the peak of the cross correlation function to be estimated using the approach described above with a precision better than the duration of one master clock cycle. In practice, for an M=8 system it is found that the position of cross correlation peak can be estimated to better than one quarter of the master clock cycle giving a distance precision very similar to the known system, but without the need for transmitting additional MLS cycles.
Figure 5 shows a second embodiment of the range finding apparatus wherein a memory 52 is provided on the output of the analogue to digital converter. Where the signal to noise ratio is poor, multiple stretched MLS signals can be sent and the received signals averaged in memory 52 prior to carrying out the coarse and fine cross-correlation calculations. As detector and ambient noise is uncorrelated, this improves the signal to noise ratio of the digitised signal and enables the maximum range and precision of the system to be improved.
In this embodiment the coarse cross-correlation unit sub-samples the received and stretched MLS signals at a different frequency to the MLS clock signal. In this case the coarse cross-correlator is clocked at a frequency FCcc which is a different sub-multiple N of the master clock signal, obtained by divider 54. This can be advantageous in improving the detection of the coarse position of the cross correlation function when the signal to noise ratio is poor, hi addition, the coarse cross-correlation unit may be preceded by a low pass filter to further improve the detection of the coarse position of the cross correlation function when the signal to noise ratio is poor. For example, the low pass filter may be implemented very simply by adding together N successive samples of the received signal.
It can be seen that many different combinations of the ratios of master clock frequency to MLS clock frequency (M=Fc/Fmls) and master clock frequency to coarse cross-correlation clock frequency (N=Fc/Fcc) are possible to yield different levels of improvement in range, precision and computation time and these combinations may be selected adaptively to optimise the performance of the LRF for different measurement conditions.
The coarse correlator periodically calculates the signal stored in the memory 52 until it detects that the signal to noise ratio is sufficient for a fine measurement to be made. Then, a fine measurement is made by the fine cross-correlation unit.
The use of the system to calculate the distance to a remote object has so far been described. The invention uses the system to detect the speed of the remote object using a single cross correlation analysis. The invention makes use of the averaging of multiple sequences, as the averging of the cross correlation function for these sequences over time results in shaping of the cross correlation function which is dependent on the speed (i.e. the change in position) of the remote obj ect.
However, the range measurement technique described above in connection with Figure 5 relies on building up a cross correlation signal with a high signal-to-noise ratio by averaging over a large number of MLS cycles. If the distance between the system and the target changes significantly during this averaging period, the position of the cross correlation signal will move in time as shown in the Figure 6. The width of an individual peak is shown as ΔTtarget, and as explained above this is the oversample clock period which will be referred to ΔTo (i.e. M times the master clock period).
The effect of this moving peak is to broaden the averaged correlation signal as shown in Figure 7.
The duration of the peak due to stationary targets is ΔTtarget « AT0. The broadened peak has duration ΔT, the additional duration ΔTV being due to the motion of the target relative to the system during the averaging period, Tav:
Δrv =Δr-Δrtarget
If d is the distance moved by the target relative to the system during Tav, then the average speed of the target relative to the PDM during the averaging period is d c AT, v = - -
T av I T av
This can be written:
AT = — -T
The table below shows values of d and ΔTV for some typical values of v for Tav = 0.1s:
For a system clock frequency of 50MHz, the fundamental clock period is 20ns. The system algorithm is capable of resolving fractions of this fundamental clock period, by the extrapolation method described above, and so it can be used to measure relative speeds with accuracy of a few miles per hour.
The invention thus involves analysing the shape of the cross correlation function peak to determine the relative speed of the remote object. In particular, the width of the peak is used to determine the relative speed of the remote object, based on a knowledge of the time period over which the multiple sequences are averaged.
In order to implement the invention, the data for the cross correlation function peak is stored, for example in a memory associated with the fine cross correlator 46. The unit 50 has a processor for analysing all of the received data in the manner explained above.
The pulse broadening does not reveal whether the target is moving towards or away from the system. A simple two-point range method can then be implemented to measure the sense of the relative motion.
The invention has been described applied to one specific cross correlation system which uses an MLS sequence. However, the invention can be applied to other correlation based systems. Furthermore, the invention has been described as implemented with a system using coarse and fine cross correlators. This system has the advantage that peaks can be accurately located with low processing power, and this enables real time distance
measurement. However, the invention can be applied more simple cross correlation based distance measurement systems.
The use of the invention in a vehicle safety system has been mentioned, in which the apparatus combines distance measurement and speed detection. There are numerous other applications in which speed measurement of a remote object can be carried out. Obviously, if the system is mounted on a stationary platform, it can provide an absolute measurement of speed, which is again applicable to many applications.
The invention has been described in detail in connection with optical range finding apparatus. The techniques of the invention could also be applied to systems using other signals, such as ultrasound.
Various other modifications will be apparent to those skilled in the art.