METHOD FOR PERFORMING RADAR MEASUREMENTS AND U OF THE METHOD FOR CALIBRATING THE RADAR SYSTEM
Technical field of the invention
The present invention relates to a method for performing radar measurements with a vehicle mounted radar 5 system. The invention also relates to the use of such a method m a vehicle mounted radar system.
Technical background
P-adar systems mounted on vehicles such as cars,
10 buses, lorries and the like are currently being developed. The purpose of such systems are to provide functions assisting the driver of the vehicle with features such as cruise control, collision warning and collision avoidance, the functions being based on the detection of
15 vehicles, stationary objects, and other objects m front of the vehicle.
Before such a radar system can be used, the system has to be mounted on the vehicle and be subject :o some kind of calibration procedure, either at the vehicle
20 factory or at a service station. The requirements on a precise mounting and/or a calibration of the radar system are very high m order to ensure a correct operation of the system. Moreover, it will probably be required to perform re-calibrations of the system at regular inter-
25 vals at an appropriate service station or the like.
In prior art, calibration, or adjustment, of a radar sensor in a radar system mounted on a vehicle, such as a car, is often performed m a similar way to headlamp adjustment. The radar sensor assembly has the same design
30 as a headlamp assembly and is equipped with alignment screws. With a radar reflector positioned m front of tne car, adjustments are performed m the car factory or m a service station using the alignment screws. Such calibrations, and re-calibrations , needed m order to
obtain, and maintain, an operation which is as correct as possible, are time consuming and costly for the owner of the vehicle.
The publisned European patent application EP 0 899 581 discloses a system for automatically measuring and compensating for any angle of misalignment of a forward- looking radar sensor of a vehicle. A trajectory line of a target is estimated and the angle of misalignment is estimated from the angle between the trajectory line and the path of travel of the radar host vehicle.
The published patent US 6 670 963 discloses a radar apparatus which calculates a correction value for an error between the beam emission axis and a forward running path of the radar host vehicle.
Summary of the invention
An object with the present invention is to provide an improved method for performing radar measurements with a vehicle mounted radar system that eliminates the need for calibrating the radar system at a vehicle manufacturer or at a vehicle service station.
According to the present invention, this objects is achieved by a method and use of a method having the features as defined in the appended claims. According to a first aspect of the present invention, there is provided a method for performing radar measurements with a vehicle mounted radar system, including the steps of: assuming that said radar system is associated with parameter errors, which errors include a presumed offset of a horizontal yaw rate sensor measurement and a presumed erroneous alignment of a radar antenna; transmitting a radar microwave signal; performing first and second measurements related to a position of a stationary object; comparing the results from the first and the second measurements, while assigning that at least either of
these measurements is dependent upon the assumed parameter errors; deriving a new presumed yaw rate offset and a new presumed erroneous antenna alignment based on a result cf said comparing step; and performing radar measurements under the assumption of the new derived parameter errors .
According to a second aspect, the invention provides use of a method m a vehicle mounted radar system for electrically calibrating the radar system when performing radar measurements with the radar system.
In the present invention, the radar system includes a radar sensor for transmitting and receiving radar beams, and processing means, such as a Signal Processing Unit, for processing measurement data derived from received radar signal reflections. For the purpose of course prediction, the radar system further includes a yaw rate sensor which provides information on the present curvature of the vehicle's directional movement. One advantage is that no precise calibrations of tne radar system has to be performed at a factory, service station, or the like, after the radar system has been mounted on the vehicle, but such calibrations are performed automatically during normal operational radar measurements.
Another advantage with the method of the invention is that calibration of the radar system is performed both with respect to the alignment of the radar sensor antenna and with respect to the offset of the yaw rate sensor measurements, thereby improving the overall performance and the accuracy of the radar system.
With the radar system, at least two different measurements related to a position of a stationary object are performed. It is assumed that the antenna alignment and the yaw rate sensor effects these measurements due to an erroneous alignment and an offset of the measured yaw rate value, respectively. By comparing the different
measurements related to a position of an object, assigneα parameter errors of the kind mentioned are derived and taken into consideration during future radar measurements . It should be noted that the result of the measurements relating to a position of an object can define any parameter which is possible to measure directly or to calculate from the measurements, as long as at least one of the measurements can be derived frorr a combination of radar measurement data and assumed parameter errors of the kind mentioned above. Examples cf parameters that are based on measurement results are tne horizontal yaw rate of the vehicle hosting the radar system and the angle m the horizontal plane to the stationary object with respect to the directional axis cf the radar system antenna.
The comparison can be performed m a numerous ways. According to one embodiment, the comparison includes minimising a mean square error between the results of tne first and the second measurements relating to the position of the object. According to another embodiment, the comparison is performed as an iterative process, eacn iteration comparing the result of one of the first measurements with the result of one of the second measurements.
The method of the invention is performed with respect to one or more stationary objects, wherein each stationary object gives rise to one or more results of first measurements and one or more results of second measurements. When basing the calibration on several results of measurements of a stationary object, the system performs a number of consecutive radar measurements on the object while simultaneously tracking the obj ect . Preferably, the provision of measurement results based on normal operational radar measurements and the following calibration of the radar system, is a radar
system process that is performed continuously by the radar system when the system is m use on a moving vehicle. Thus, the invention provides a method which repeatedly controls the calibration of the radar system while performing normal operational radar measurements.
With these regular calibrations, or re-calibrations , of the radar system, the information which is based on the radar measurement data, and provided to a user of the radar system, will always be reliable and up to date. In order to achieve as reliable calibrations as possible, it is preferred to calibrate the radar system after having performed radar measurements with respect to a certain number of stationary objects, for example a couple of hundred objects. It is also preferred to cali- brate the radar system at regular time intervals.
According to the invention, the radar system is adapted to be mounted on a land based vehicle, such as a car, bus or lorry.
The radar sensor included by the radar system is preferably an FMCW-radar sensor (Frequency Modulated Continuous Wave) .
Hence, the present invention provides a method which makes calibration and subsequent re-calibrations of the vehicle mounted radar system redundant, i.e. there is no need for any initial calibration or subsequent regular service of the radar system at a service station or the like. The antenna of the radar sensor will automatically be aligned with the directional axis of the vehicle, i.e. with the optimal direction of the antenna. This will clearly lower the cost for maintenance of the radar system. Furthermore, the radar system can handle imperfections of the yaw rate sensor by calibrating the yaw rate sensor and processing measured yaw rate values which are very close to the true yaw rate values .
Brief description of the drawings
An exemplifying embodiment of the invention will now be described below with reference to the accompanying drawings, m which: Fig. 1 schematically shows a side view of a vehicle on which a radar system according to the present invention has been mounted;
Fig. 2 schematically shows a vehicle carrying a radar system m a system of co-ordinates having the vehicle as a reference; and
Fig. 3 is a flow chart of an embodiment of a method according to the invention.
Detailed description of a preferred embodiment With reference to Fig. 1, a radar system according to the present invention is schematically shown. The radar system is mounted on a vemcle 110 and includes a radar sensor 100, signal processing means 130 and a yaw rate sensor 140. The radar sensor 100 is located m the front of the vehicle 110 and transmits a radar microwave signal 105 which is reflected by objects (not shown) positioned m front of the vehicle 110. The radar sensor 100 receives the reflected signal and transfers radar measurement data derived from the reflected signal to the signal processing means 130, which is electrically connected to the radar sensor 100. The yaw rate sensor 140 is a horizontal curvature sensor electrically connected to the signal processing means 130. The yaw rate sensor 140 is generally used for course prediction and provides information on the present curvature of the vehicle's directional movement. The yaw rate of the vehicle 110 provided by the yaw rate sensor 140 is fed to the signal processing means 130. To implement an overall system functionality, a number of other means are also mounted on the vehicle 110, such as means for engine and brake control, a man-machine interface, and so on (not shown) .
Fig. 2 schematically shows a vehicle 110 carrying a radar system (not shown) m accordance with the present invention. The vehicle 110 is shown in a system of coordinates having the vehicle as a reference, i.e. the origin of co-ordinates and the x-axis are equal to the position of the vehicle and the directional axis of the vehicle, respectively. Indicated m the system of coordinates is also a stationary object 210, depicted as a dot. Ideally, the radar sensor antenna (not shown) should be aligned with the directional axis of the vehicle, i.e. with the x-axis of the system of co-ordinates.
With reference to Fig. 2 an embodiment of tne invention will be described below using the following variables :
which is a measured yaw rate of the vehicle; m(t) which is a zero drift of the yaw rate sensor; k(t) which is a scaling factor of the yaw rate sensor;
which is a deviation of an antenna direction from an antenna direction aligned with the directional axis of the vehicle;
.(0 which is a distance to the stationary object from the vehicle; r(t) which is the relative radial speed of the vehicle with respect to the stationary object; v(t) which is the angle to the stationary object with respect to the x-axis; and lit) which is the derivative of the angle to the stationary object.
How to derive the variables r^ /(r), lN ) and ^ is well Known to a person skilled m the art. The distance
to an object 210 is proportional to the time delay of an emitted radar signal which is reflected by the object 210 and received again. Difference frequencies are obtained by mixing the received signal with the emitted signal and adding frequency shifts resulting from the doppler effect. Using these difference frequencies the distance to the object 210 and the relative radial speed of the object 210 is derived. The determination of the angular position of the object 210 with respect to tne sensor axis is done by comparing the amplitudes of the radar reflections of the radar signals with a known antenna pattern .
In order to have a reference for the yaw rate of the vehicle, a reference direction axis is indicated in Fig. 2. Since the system of co-ordinates has the vehicle 110 as a reference, the stationary object 210 can be seen as a moving object m the system of co-ordinates of the vehicle. If the vehicle speed is v>--, the following equations can be derived from Fig. 2:
Eliminating gives us
σ - tan( v)
A . 2 )
In Fig. 2, we also see that
where u is the true yaw rate of the vehicle with respect to the reference axis. By eliminating θ and σ m equa
¬ tions 2.2 and 2.3, and by observing the system of Fig. 2 at a point of time t, we arrive at.
w(t) = - v(t) - tan( O) r(t) '2.4'
Thus, a theoretical value of the yaw rate of the vehicle can be calculated m accordance with equation 2.4 by tne signal processing means 130 using radar measurement data of the signals reflected by the stationary opject 220 With the yaw rate sensor 140, a yaw rate u ,Λ (t) of the vehicle 110 is also measured. However, due to imperfections of the radar system, and its included yaw rate sensor 140, the measured yaw rate value u rø (t) will not be a true yaw rate of the vehicle, but a true yaw rate value affected by at least two presumed radar system parameter errors, the scaling factor k {t) and the zero drift m(t) of the yaw rate sensor. At a given point of time, the zero drift corresponds to an offset included m the measured yaw rate value. Thus, the measured yaw rate value \\ )R (t) , measured with the yaw rate sensor 140, satisfies the following equation:
W.Λ(0 = λ( w(t) + m(t) (2.5)
where u(t) is the true yaw rate of the vehicle.
If we assume that our antenna is not properly aligned with the directional axis of the vehicle 110, i.e. with the x-axis, m Fig. 2, but has a deviation Δ (/) from a correct alignment, the system will measure a direction v (t) rather than \ (t) to the stationary object 220, where:
= -Δ,( [2.6)
By substituting v(t) m equation 2.4, using equation 2.6 and then substituting vv(t) m equation 2.5 using equation 2.4, equation 2.5 can be written:
r(t)
,( = (- 0- tan(i'(t) + Δι( )) k(t) + m(t) r(t) 2.1
If we assume that the deviation Δ, (t) is small, equation 2.7 can be approximated with:
Equation 2.8 can be rewritten as
A*! ) t) -^-tan(H ) r(t) (
Δ,(0" + e(l) >'( r(t) k(t)
This expression can be divided m to separate expressions for the yaw rate of vehicle 110:
A(.) = - v(t) + ^--tan( ) (2.9) r( j
and
t) = + ^ >n(t) [2.10' k r(t) k(t)
Thus, we now have two different calculations of the yaw rate, a first calculation given by equation 2.9 and a second calculation given by equation 2.10. The first expression 2.9 provides the yaw rate as a result of calculations based on radar measurement data v(- l> , , ' X> and X i.e. the direction to the stationary object 220 m the form of the angle, the derivative of said direction, the distance to the object and a radial relative velocity with respect to the object, respectively, which radar measurement data is received/calculated py the signal processing means 130. The second expression 2.10 provides the yaw rate as a result of calculations based on radar measurement data A/N) ' and ' X> . Included m expression 2.10 is furthermore the assumed radar system parameter errors Δ^t) , k(t) and m(t) caused by the erroneous alignment of the radar antenna and the scaling factor and zero drift of the yaw rate sensor, respectively.
The difference between the two resulting values of the two expressions 2.9 and 2.10, is a measure of how correct our k(t) , ιn(t ) and Δs(.) values are. Thus, by minimising the difference between the two resulting values of the two expressions 2.9 and 2.10 with respect to k{t) , m(t) and At (t ) , the result will be derived values of k(t) , m(t) and Δ, (/) that are best suitable to be used when calibrating the radar system with respect to these system parameters. The calibration with respect to the scaling factor k(t) and zero drift m(t) is preferably a software controlled compensation of the yaw rate values, as measured with the yaw rate sensor 140, and performed by the processing means 130. The calibration with respect to the deviation Δ, (t) from a properly aligned antenna is preferably an electrically controlled elimination of this deviation. The elimination is either performed internally of the processing means 130 by compensating radar
measurements for the deviation, or, it is initiated by the processing means 130 and performed by means of an electrically controlled, physical adjustment of the antenna direction.
Our system now takes the following form:
Θ,( =
10 k( t ) is the inversion of the scaling factor of the yaw rate sensor.
(t)
Θ:(t) = kit) is the zero drift of the yaw rate sensor normalised with the scaling factor.
1 .5_. Θ, (t) ' = Δ ,Λ(t)' is ^t,he d-,eviat,ion f ,rom an optima-l. antenna alignment .
11, (1) ! (t) H, ( t ) and e(t) are white noise having a normal distribution with tne expected value 0 and variances
20 <τ CT-. and °~ , respectively.
The expression updates now take the following form: Θ(t + 1) = Θ(t) + K(w2 (t) - , (t)) (2.12;
We can calculate m(t), k(t) and ' *- ' from the equations of
0,(0-0,(0
To calculate the amplification K it is preferred to either use an RLS (Recursive Least Square) algorithm or a simpler EMS (Least Mean Square) algorithm. Dependant on the chosen algorithm, different number of radar measurements are required to achieve good estimations of the presumed radar system parameter errors. The calculation is also dependant on different time constants, e.g., we can assume that the zero drift error varies within minutes, while the antenna deviation remains unchanged for several years .
With reference to the flow chart of Fig. 3, the embodiment described above will be further illustrated. After start-up of the radar system m step SI, the system is initialised m step S2. The initialisation includes reading the values of Θ, , Θ2 and Θ3 from an internal, non-volatile memory to obtain an initial Θ(t) . These values were previously stored m the memory when the system was last m operation. In step S3, normal radar measurements are performed by transmitting and receiving radar signals. Data s derived from radar signal reflections of a stationary object 210 located m front of the vehicle 110, the data includes an angle to the stationary object 210 with respect to the directional axis of the vehicle 110, a calculated derivative of said angle, a distance to the stationary object 210 and a relative velocity of the vehicle 110 with respect to the stationary object 210. In addition, m step S3, the yaw rate of the vehicle is measured with the yaw rate sensor 140, the measurement being performed at the same time as said data is derived.
In step S4 , two estimations of the true yaw rate of the vehicle is provided. The data derived from the signal reflections is used for calculating the first resulting yaw rate as estimation vt, m accordance with equation
2.9. The second resulting yaw rate, estimation A, , is calculated from the data of the signal reflections and a yaw rate wYR as measured with the yaw rate sensor 140, which measured yaw rate is compensated for presumed radar system parameter errors, all m accordance with equation
2.10. In the descπped embodiment, these errors are, as described above, the angular deviation Δ, from an optimal antenna direction and the scaling factor /. and zero drift m of the yaw rate sensor 140. In step S5, Θ is updated based on the difference between the first and the second estimation using equation 2.12 and, for example, an LMS algorithm for calculating the amplification K . The resulting Θ(t + 1) is stored m the non-volatile memory and will be used m the next iteration of equation 2.12. If, m step S6, radar measurements according to step S3 have been performed with respect to a minimum number of stationary objects, for example 300, the process continues to step S7, otherwise the process returns to step S3. In step S7 it is checKed whether enough time has elapsed since the system parameter errors were last updated. If this predefined time, for example two minutes, is exceeded the process continues to step S8, otherwise the process returns to step S3. In step S8 the system parameter errors Δ k and in are calculated and updated using the equations of Θ, , Θ, and Θ3. Finally, the radar system is calibrated using the values calculated m step S8. The process then returns to step S3, and, eventually, a new calibration of the radar system.
Even though the invention has been described with reference to a specific exemplifying embodiment, many different alterations, modifications and the like will become apparent for those skilled m the art. The described embodiment is therefore not intended to limit the scope of the invention, as defined by the appended claims.