Method and apparatus for generating automatic driving strategy

1. A method for generating an autonomous driving strategy, comprising:

measuring state information of an instant vehicle and ambient scene information, the ambient scene information comprising: state information of an impediment vehicle, road structure information, and traffic scene information of the instant vehicle;

determining a running track of the impediment vehicle according to the state information of the impediment vehicle within a predetermined period of time;

determining a first mapping relation based on the state information of the impediment vehicle within the predetermined period of time and the running track of the impediment vehicle, the first mapping relation being a mapping relation from the traffic scene information of the impediment vehicle to driving behavior information of the impediment vehicle; and

generating the autonomous driving strategy of the instant vehicle based on the first mapping relation, the road structure information, the state information of the instant vehicle and the traffic scene information of the instant vehicle,

wherein the determining a first mapping relation based on the state information of the impediment vehicle within the predetermined period of time and the running track of the impediment vehicle comprises:

determining a second mapping relation of the impediment vehicle according to the state information of the impediment vehicle within the predetermined period of time and the running track of the impediment vehicle, the second mapping relation being a mapping relation from the traffic scene information of the impediment vehicle to the running track of the impediment vehicle;

segmenting the running track of the impediment vehicle along a time axis to obtain track segments of the impediment vehicle;

filtering out an effective track segment of the impediment vehicle from the track segments of the impediment vehicle;

learning a third mapping relation using an unsupervised learning method, the third mapping relation being a mapping relation from the effective track segment of the impediment vehicle to the driving behavior information of the impediment vehicle; and

determining the first mapping relation using a supervised learning method based on the second mapping relation and the third mapping relation.

2. The method according to claim 1 , wherein the segmenting the running track of the impediment vehicle along a time axis to obtain track segments of the impediment vehicle comprises:

traversing all track points in the running track of the impediment vehicle and a track time length to determine the track segments.

3. The method according to claim 1 , wherein the filtering out an effective track segment of the impediment vehicle from the track segments of the impediment vehicle comprises:

evaluating the track segments using an evaluation function; and

determining the track segment as the effective track segment in response to an evaluated value being greater than a threshold value.

4. The method according to claim 1 , wherein the generating the autonomous driving strategy of the instant vehicle based on the first mapping relation, the road structure information, the state information of the instant vehicle and the traffic scene information of the instant vehicle comprises:

determining the driving behavior information of the impediment vehicle corresponding to the traffic scene information of the instant vehicle in the first mapping relation as driving behavior information of the instant vehicle;

establishing an inverse mapping relation of the third mapping relation to obtain a fourth mapping relation;

determining the running track of the impediment vehicle corresponding to the driving behavior information of the instant vehicle in the fourth mapping relation as a running track of the instant vehicle; and

optimizing the running track of the instant vehicle based on a pre-established running performance index function of an autonomous vehicle, the road structure information, the state information of the instant vehicle and the traffic scene information of the instant vehicle to obtain the driving strategy of the instant vehicle.

5. An apparatus for generating an autonomous driving strategy, comprising:

at least one processor; and

a memory storing instructions, which when executed by the at least one processor, cause the at least one processor to perform operations, the operations comprising:

measuring state information of an instant vehicle and ambient scene information, the ambient scene information comprising: state information of an impediment vehicle, road structure information, and traffic scene information of the instant vehicle;

determining a running track of the impediment vehicle according to the state information of the impediment vehicle within a predetermined period of time;

determining a first mapping relation based on the state information of the impediment vehicle within the predetermined period of time and the running track of the impediment vehicle, the first mapping relation being a mapping relation from the traffic scene information of the impediment vehicle to driving behavior information of the impediment vehicle; and

generating the autonomous driving strategy of the instant vehicle based on the first mapping relation, the road structure information, the state information of the instant vehicle and the traffic scene information of the instant vehicle,

wherein the determining a first mapping relation based on the state information of the impediment vehicle within the predetermined period of time and the running track of the impediment vehicle comprises:

determining a second mapping relation of the impediment vehicle according to the state information of the impediment vehicle within the predetermined period of time and the running track of the impediment vehicle, the second mapping relation being a mapping relation from the traffic scene information of the impediment vehicle to the running track of the impediment vehicle;

segmenting the running track of the impediment vehicle along a time axis to obtain track segments of the impediment vehicle;

filtering out an effective track segment of the impediment vehicle from the track segments of the impediment vehicle;

learning a third mapping relation using an unsupervised learning apparatus, the third mapping relation being a mapping relation from the effective track segment of the impediment vehicle to the driving behavior information of the impediment vehicle; and

determining the first mapping relation using a supervised learning apparatus based on the second mapping relation and the third mapping relation.

6. The apparatus according to claim 5 , wherein the segmenting the running track of the impediment vehicle along a time axis to obtain track segments of the impediment vehicle comprises:

traversing all track points in the running track of the impediment vehicle and a track time length to determine the track segments.

7. The apparatus according to claim 5 , wherein the filtering out an effective track segment of the impediment vehicle from the track segments of the impediment vehicle comprises:

evaluating the track segments using an evaluation function; and

determining the track segment as the effective track segment in response to an evaluated value being greater than a threshold value.

8. The apparatus according to claim 5 , wherein the generating the autonomous driving strategy of the instant vehicle based on the first mapping relation, the road structure information, the state information of the instant vehicle and the traffic scene information of the instant vehicle comprises:

determining the driving behavior information of the impediment vehicle corresponding to the traffic scene information of the instant vehicle in the first mapping relation as driving behavior information of the instant vehicle;

establishing an inverse mapping relation of the third mapping relation to obtain a fourth mapping relation;

determining the running track of the impediment vehicle corresponding to the driving behavior information of the instant vehicle in the fourth mapping relation as a running track of the instant vehicle; and

optimizing the running track of the instant vehicle based on a pre-established running performance index function of an autonomous vehicle, the road structure information, the state information of the instant vehicle and the traffic scene information of the instant vehicle to obtain the driving strategy of the instant vehicle.

9. A non-transitory computer readable storage medium, storing a computer program, which when executed by one or more processors, causes the one or more processors to perform operations, the operations comprising:

measuring state information of an instant vehicle and ambient scene information, the ambient scene information comprising: state information of an impediment vehicle, road structure information, and traffic scene information of the instant vehicle;

determining a running track of the impediment vehicle according to the state information of the impediment vehicle within a predetermined period of time;

determining a first mapping relation based on the state information of the impediment vehicle within the predetermined period of time and the running track of the impediment vehicle, the first mapping relation being a mapping relation from the traffic scene information of the impediment vehicle to driving behavior information of the impediment vehicle; and

generating the autonomous driving strategy of the instant vehicle based on the first mapping relation, the road structure information, the state information of the instant vehicle and the traffic scene information of the instant vehicle,

wherein the determining a first mapping relation based on the state information of the impediment vehicle within the predetermined period of time and the running track of the impediment vehicle comprises:

determining a second mapping relation of the impediment vehicle according to the state information of the impediment vehicle within the predetermined period of time and the running track of the impediment vehicle, the second mapping relation being a mapping relation from the traffic scene information of the impediment vehicle to the running track of the impediment vehicle;

segmenting the running track of the impediment vehicle along a time axis to obtain track segments of the impediment vehicle;

filtering out an effective track segment of the impediment vehicle from the track segments of the impediment vehicle;

learning a third mapping relation using an unsupervised learning apparatus, the third mapping relation being a mapping relation from the effective track segment of the impediment vehicle to the driving behavior information of the impediment vehicle; and

determining the first mapping relation using a supervised learning apparatus based on the second mapping relation and the third mapping relation.