Path generation system for vehicular collision avoidance and mitigation
A path generation system for vehicular collision avoidance and mitigation generates a path with a lowest risk level of collision damage at the point in time at which an ego vehicle faces an imminent collision. The path generation system includes a neural network learning unit configured to output a terminal state of the ego vehicle based on information of the ego vehicle and information of surrounding vehicles. The path generation system may thus generate a path for minimizing damage to a driver and a passenger of the ego vehicle at the point in time at which the ego vehicle faces an imminent collision.
1 . A path generation system, comprising:
a plurality of sensors configured to acquire state information of an ego vehicle and state information of a surrounding environment of the ego vehicle, including information of surrounding vehicles; and
a processor operatively connected to the plurality of sensors,
wherein the processor comprises:
a neural network learning unit configured to output a terminal state of the ego vehicle based on the state information of the ego vehicle and the information of the surrounding vehicles; and
a path generation unit configured to generate a collision avoidance and mitigation path based on the terminal state to mitigate a risk level of collision damage in a situation in which the ego vehicle faces an imminent collision,
wherein the neural network learning unit is configured to perform learning using a deep reinforcement learning algorithm, and
wherein the neural network learning unit is configured to feed a collision impact index and time-to-collision (TTC)-based risk level, as a reward function, back to the deep reinforcement learning algorithm.
2 . The path generation system of claim 1 , wherein the neural network learning unit is configured to learn i) speeds along X-axis and Y-axis, a yaw value, and lane information of the ego vehicle and ii) relative positions along X-axis and Y-axis, a speed, a yaw value, and lane information of the surrounding vehicles.
3 . The path generation system of claim 2 , wherein the terminal state includes a terminal position on the collision avoidance and mitigation path, a terminal speed on the collision avoidance and mitigation path, and a time taken to reach the terminal state.
4 . The path generation system of claim 3 , wherein the path generation unit is configured to generate the collision avoidance and mitigation path using a quintic polynomial based algorithm for path planning.
5 . The path generation system of claim 4 , wherein the quintic polynomial based algorithm for path planning employs a quintic polynomial equation, initial value, and boundary value, as follows:
x
(
t
)
=
a
0
+
a
1
t
+
a
2
t
2
+
a
3
t
3
+
a
4
t
4
+
a
5
t
5
y
(
t
)
=
b
0
+
b
1
t
+
b
2
t
2
+
b
3
t
3
+
b
4
t
4
+
b
5
t
5
t
terminal
=
T
x
(
0
)
=
X
ego
[
m
]
y
(
0
)
=
Y
ego
[
m
]
v
(
0
)
=
V
ego
[
m
/
s
]
a
(
0
)
=
A
ego
[
m
/
s
2
]
ψ
(
0
)
=
ψ
ego
[
rad
]
x
(
t
terminal
)
=
X
ego
+
R
[
m
]
y
(
t
terminal
)
=
Y
ego
+
L
[
m
]
v
(
t
terminal
)
=
v
T
×
tanh
(
R
)
[
m
/
s
]
a
(
t
terminal
)
=
0
[
m
/
s
2
]
ψ
(
t
terminal
)
=
0
[
rad
]
where x and y depict positions along x-y axes, v depicts a speed, a depicts an acceleration, Ψ depicts a yaw value, t depicts a time, and a 0 , a 1 , a 2 , a 3 , a 4 , a 5 , b 0 , b 1 , b 2 , b 3 , b 4 , and b 5 depict coefficients.
6 . The path generation system of claim 5 , wherein the path generation unit is configured to determine coefficients of the quintic polynomial equation using the initial value and the boundary value.
7 . A path generation method, comprising:
acquiring state information of an ego vehicle and state information of a surrounding environment of the ego vehicle detected by a plurality of sensors;
outputting, by a processor, a terminal state of the ego vehicle based on information of the ego vehicle and information of surrounding vehicles; and
generating, by the processor, a collision avoidance and mitigation path based on the terminal state to mitigate a risk level of collision damage in a situation in which the ego vehicle faces an imminent collision,
wherein further comprising performing, by the processor, learning using a deep reinforcement learning algorithm, and
further comprising feeding, by the processor, a collision impact index and time-to-collision (TTC)-based risk level, as a reward function, back to the deep reinforcement learning algorithm.
8 . The path generation method of claim 7 , further comprising learning, by the processor, i) speeds along X-axis and Y-axis, a yaw value, and lane information of the ego vehicle and ii) relative positions along X-axis and Y-axis, a speed, a yaw value, and lane information, of the surrounding vehicles.
9 . The path generation method of claim 8 , wherein the terminal state includes a terminal position on the collision avoidance and mitigation path, a terminal speed on the collision avoidance and mitigation path, and a time taken to reach the terminal state.
10 . The path generation method of claim 9 , wherein generating the collision avoidance and mitigation path comprises generating the collision avoidance and mitigation path using a quintic polynomial based algorithm for path planning.
11 . The path generation method of claim 10 , wherein the quintic polynomial based algorithm for path planning employs a quintic polynomial equation, initial value, and boundary value, as follows:
x
(
t
)
=
a
0
+
a
1
t
+
a
2
t
2
+
a
3
t
3
+
a
4
t
4
+
a
5
t
5
y
(
t
)
=
b
0
+
b
1
t
+
b
2
t
2
+
b
3
t
3
+
b
4
t
4
+
b
5
t
5
t
terminal
=
T
x
(
0
)
=
X
ego
[
m
]
y
(
0
)
=
Y
ego
[
m
]
v
(
0
)
=
V
ego
[
m
/
s
]
a
(
0
)
=
A
ego
[
m
/
s
2
]
ψ
(
0
)
=
ψ
ego
[
rad
]
x
(
t
terminal
)
=
X
ego
+
R
[
m
]
y
(
t
terminal
)
=
Y
ego
+
L
[
m
]
v
(
t
terminal
)
=
v
T
×
tanh
(
R
)
[
m
/
s
]
a
(
t
terminal
)
=
0
[
m
/
s
2
]
ψ
(
t
terminal
)
=
0
[
rad
]
where x and y depict positions along x-y axes, v depicts a speed, a depicts an acceleration, Ψ depicts a yaw value, t depicts a time, and a 0 , a 1 , a 2 , a 3 , a 4 , a 5 , b 0 , b 1 , b 2 , b 3 , b 4 , and b 5 depict coefficients.
12 . The path generation method of claim 11 , further comprising determining, by the processor, coefficients of the quintic polynomial equation using the initial value and the boundary value.