Real-time driving danger level prediction
View Patent ↗Systems and methods are disclosed to predict driving danger by capturing vehicle dynamic parameter, driver physiological data and driver behavior feature; applying a learning algorithm to the features; and predicting driving danger.
1. A method to predict driving danger, comprising:
a. capturing vehicle dynamic parameter, driver physiological data and driver behavior feature;
b. applying a learning algorithm including a statistical model having a continuous danger level function to the features, said training algorithm being learned; and
c, predicting driving danger with the statistical model in real time including a driving danger level DL*(x n ) at time instance n with a continuous danger level function DL n =DL(x n ,Θ) with parameters Θ where
Θ
*
=
arg
min
Θ
∑
i
=
1
N
∑
n
=
1
T
n
(
DL
(
x
n
i
,
Θ
)
-
DL
*
(
x
n
i
)
)
.
2. The method of claim 1 , wherein the learning algorithm includes one of: Hidden Markov Model, Conditional Random Field and Reinforcement Learning.
3. The method of claim 1 , wherein the vehicle dynamic parameter includes one or more of: driver's lateral lane position, steering wheel angle, longitudinal acceleration, longitudinal velocity, distance between vehicles.
4. The method of claim 1 , wherein the driver's physiological data includes one or more of: respiration, heart rate, blood volume, skin temperature, skin conductance.
5. The method of claim 1 , wherein the driver behavior feature comprises a PERCLOSE feature.
6. The method of claim 1 , wherein the driver behavior feature comprises fatigue, vision, distraction.
7. The method of claim 1 , comprising performing off line cross-validation.
8. The method of claim 1 , comprising communicating a reason for a predicted driving danger to a user.
9. The system of claim 1 , wherein the learning processor predicts driving danger in real time.
10. The system of claim 1 , comprising a display to communicate a reason for a predicted driving danger to a user.
11. A system to predict driving danger, comprising:
a. a data acquisition unit to capture vehicle dynamic parameter, driver physiological data and driver behavior feature;
b. a learning processor coupled to the data acquisition unit to process the features and to predict a driving danger level DL*(x n ) at time instance n with a continuous danger level function DL n =DL(x n ,Θ) with parameters Θ where
Θ
*
=
arg
min
Θ
∑
i
=
1
N
∑
n
=
1
T
n
(
DL
(
x
n
i
,
Θ
)
-
DL
*
(
x
n
i
)
)
;
and
c. a user interface coupled to the learning processor to indicate driving danger.
12. The system of claim 11 , wherein the learning processor includes one of: Hidden Markov Model, Conditional Random Field and Reinforcement Learning.
13. The system of claim 11 , wherein the vehicle dynamic parameter includes one or more of: driver's lateral lane position, steering wheel angle, longitudinal acceleration, longitudinal velocity, distance between vehicles.
14. The system of claim 11 , wherein the driver's physiological data includes one or more of: respiration, heart rate, blood volume, skin temperature, skin conductance.
15. The system of claim 11 , wherein the driver behavior feature comprises a PERCLOSE feature.
16. The system of claim 11 , wherein the driver behavior feature comprises fatigue, vision, distraction.
17. The system of claim 11 , wherein the learning processor is trained using previously captured data.
18. The system of claim 11 , wherein the learning processor cross-validates predicted driving dangers.