Driver assistance apparatus
In accordance with an aspect of a driver assistance system includes a radar sensor installed in a vehicle, having a side sensing field of view of the vehicle, and configured to acquire side sensing data; and a controller including a processor configured to process the side sensing data; and the controller may be configured to estimate motion state of a moving object at the side of the vehicle based on applying the side sensing data to at least one of a first estimation model and a second estimation model.
1. A driver assistance system, comprising:
a radar sensor installed in a vehicle, having a side sensing field of view of the vehicle, and configured to acquire side sensing data; and
a controller including a processor configured to process the side sensing data,
wherein the controller is configured to apply the side sensing data to at least one estimation model of a first estimation model and a second estimation model and estimate a motion state of a moving object at a side of the vehicle based on the applying the side sensing data to the at least one estimation model, wherein the controller is configured to:
generate an expected value of the motion state of the moving object based on a sum of products of estimated values of previous motion states of the moving object and coupling probabilities between the first and second estimation models, and
estimate the motion state of the moving object based on the expected value of the motion state of the moving object.
2. The driver assistance system of claim 1 , wherein the controller is configured to estimate the motion state of the moving object based on at least one of the first estimation model that calculates on a premise that a lateral motion and a longitudinal motion of the moving object are an equivalent acceleration motion, and the second estimation model that calculates on the premise that the moving object is in a curved motion.
3. The driver assistance system of claim 1 , wherein the first estimation model is configured to estimate the motion state of the moving object based on a position of the moving object, a speed of the moving object, and an acceleration of the moving object.
4. The driver assistance system of claim 1 , wherein the second estimation model is configured to estimate the motion state of the moving object based on a position of the moving object, a speed of the moving object, an acceleration of the moving object, curvature of a lane on which the moving object travels, a center line of the lane, and error information of a traveling direction angle of the moving object.
5. The driver assistance system of claim 4 , wherein the second estimation model is configured to estimate the motion state of the moving object by applying the side sensing data of the moving object to a Frenet frame.
6. The driver assistance system of claim 1 , wherein the controller is configured to calculate the expected value of the motion state of the moving object based on Equation 3 below,
=
∑
i
=
1
M
(
k
-
1
❘
k
-
1
)
*
μ
i
❘
j
(
k
-
1
❘
k
-
1
)
[
Equation
3
]
here, is the expected value of the motion state of the moving object, is an estimated value of a previous motion state of the moving object, and μ i|j is a coupling probability between the first and second estimation models.
7. The driver assistance system of claim 6 , wherein the controller is configured to calculate the coupling probability based on Equation 4 below,
μ
i
❘
j
(
k
-
1
❘
k
-
1
)
=
1
c
j
_
p
ij
μ
i
(
k
-
1
)
[
Equation
4
]
here, μ i|j is the coupling probability between the first and second estimation models, c j is a normalization coefficient, c j =Σ i=1 M p ij μ i (k−1), p ij is a model transition probability, p ij =Prob{I(θ k )=j|I(θ k-1 )=i}.
8. The driver assistance system of claim 7 , wherein the controller is configured to generate a likelihood function based on equation 5 below,
Λ
j
(
k
)
=
1
2
π
❘
"\[LeftBracketingBar]"
S
j
(
k
)
❘
"\[RightBracketingBar]"
exp
(
-
E
j
(
k
)
S
j
(
k
)
-
1
E
j
(
k
)
T
2
)
[
Equation
5
]
here, Λ j (k) is the likelihood function, S j (k) is an error covariance between a measurement model and a measurement value (S j (k)=H j P j (k|k−1)H j T +R j (k)), E j (k) is an error between an estimate from the measurement model and an actual measurement (E j (k)=z(k)−{circumflex over (z)} j (k|k−1)).
9. The driver assistance system of claim 6 , wherein the controller is configured to calculate a model probability selected from the first estimation model and the second estimation model based on Equation 6 below,
μ
j
(
k
)
=
Λ
j
(
k
)
c
j
_
∑
i
=
1
M
Λ
j
(
k
)
c
j
_
[
Equation
6
]
here, c j is a normalization coefficient, c j =Σ i=1 M p ij μ i (k−1), and Λ j (k) is a likelihood function.
10. The driver assistance system of claim 7 , wherein the controller is configured to estimate the motion state of the moving object based on a model probability according to Equation 7 below,
X
^
(
k
❘
k
)
=
∑
i
=
1
M
X
^
j
(
k
❘
k
)
μ
j
(
k
)
[
Equation
7
]
here, μ j (k) is the model probability, {circumflex over (X)} j (k|k) is the expected value of the motion state of the moving object.
11. A method for driver assistance of a vehicle, comprising:
providing at least one radar sensor having a side sensing field of view to the vehicle;
acquiring, by a processor, side sensing data from the at least one radar sensor;
applying, by the processor, the side sensing data to at least one estimation model of a first estimation model and a second estimation model; and
estimating, by the processor, a motion state of a moving object at a side of the vehicle based on the applying of the side sensing data to the at least one estimation model,
wherein the estimating of the motion state of the moving object includes:
generating, by the processor, an expected value of the motion state of the moving object based on a sum of products of estimated values of previous motion states of the moving object and coupling probabilities between the first and second estimation models,
estimating, by the processor, the motion state of the moving object based on the expected value of the motion state of the moving object, and
controlling the vehicle based on a result of the estimating of the motion state of the moving object.
12. The method of claim 11 , wherein the estimating of the motion state of the moving object includes estimating the motion state of the moving object based on at least one of the first estimation model that calculates on a premise that a lateral motion and a longitudinal motion of the moving object are an equivalent acceleration motion, and the second estimation model that calculates on the premise that the moving object is in a curved motion.
13. The method of claim 11 , wherein the first estimation model is configured to estimate the motion state of the moving object based on a position of the moving object, a speed of the moving object, and an acceleration of the moving object.
14. The method of claim 11 , wherein the second estimation model is configured to estimate the motion state of the moving object based on a position of the moving object, a speed of the moving object, an acceleration of the moving object, curvature of a lane on which the moving object travels, a center line of the lane, and error information of a traveling direction angle of the moving object.
15. The method of claim 14 , wherein the second estimation model is configured to estimate the motion state of the moving object by applying the side sensing data of the moving object to a Frenet frame.
16. The method of claim 11 , wherein the estimating of the motion state of the moving object includes calculating the expected value of the motion state of the moving object based on Equation 3 below,
=
∑
i
=
1
M
(
k
-
1
❘
k
-
1
)
*
μ
i
❘
j
(
k
-
1
❘
k
-
1
)
[
Equation
3
]
here, is the expected value of the motion state of the moving object, is an estimated value of a previous motion state of the moving object, and μ i|j is a coupling probability between the first and second estimation models.
17. The method of claim 16 , wherein the estimating of the motion state of the moving object includes calculating the coupling probability based on Equation 4 below,
μ
i
❘
j
(
k
-
1
❘
k
-
1
)
=
1
c
j
_
p
ij
μ
i
(
k
-
1
)
[
Equation
4
]
here, μ i|j is the coupling probability between the first and second estimation models, c j is a normalization coefficient, c j =Σ i=1 M p ij μ i (k−1), p ij is a model transition probability, p ij =Prob{I(θ k )=j|I(θ k-1 )=i}.
18. The method of claim 17 , wherein the estimating of the motion state of the moving object includes generating a likelihood function based on equation 5 below,
Λ
j
(
k
)
=
1
2
π
❘
"\[LeftBracketingBar]"
S
j
(
k
)
❘
"\[RightBracketingBar]"
exp
(
-
E
j
(
k
)
S
j
(
k
)
-
1
E
j
(
k
)
T
2
)
[
Equation
5
]
here, Λ j (k) is the likelihood function, S j (k) is an error covariance between a measurement model and a measurement value, E j (k) is an error between an estimate from the measurement model and an actual measurement.
19. The method of claim 11 , wherein the estimating of the motion state of the moving object includes calculating a model probability selected from the first estimation model and the second estimation model based on Equation 6 below,
μ
j
(
k
)
=
Λ
j
(
k
)
c
j
_
∑
i
=
1
M
Λ
j
(
k
)
c
j
_
[
Equation
6
]
here, c j is a normalization coefficient, c j =Σ 1=1 M p ij μ i (k−1), and Λ j (k) is a likelihood function.
20. The method of claim 17 , wherein the estimating of the motion state of the moving object includes estimating the motion state of the moving object based on a model probability according to Equation 7 below,
X
^
(
k
❘
k
)
=
∑
i
=
1
M
X
^
j
(
k
❘
k
)
μ
j
(
k
)
[
Equation
7
]
here, μ j (k) is the model probability, {circumflex over (X)} j (k|k) is the expected value of the motion state of the moving object.