Systems and methods for autonomous driving

1. A system for autonomous driving, comprising:

at least one storage medium including a set of instructions; and

at least one processor in communication with the at least one storage medium, wherein when executing the set of instructions, the at least one processor is directed to cause the system to:

obtain driving information associated with a vehicle;

determine a state of the vehicle;

determine one or more candidate control signals and one or more evaluation values corresponding to the one or more candidate control signals based on the driving information and the state of the vehicle by using a trained control model;

select a target control signal from the one or more candidate control signals based on the one or more evaluation values according to a predetermined rule; and

transmit the target control signal to a control component of the vehicle.

2. The system of claim 1 , wherein the driving information associated with the vehicle includes perception information within a first predetermined range of the vehicle and map information within the first predetermined range of the vehicle.

3. The system of claim 2 , wherein the perception information includes road condition information and obstacle information.

4. The system of claim 1 , wherein the state of the vehicle includes a velocity of the vehicle, an acceleration of the vehicle, and a steering wheel angle of the vehicle.

5. The system of claim 1 , wherein the trained control model is determined with a training process, the training process comprising:

obtaining a preliminary control model;

obtaining preliminary sample driving information, the preliminary sample driving information including preliminary sample perception information within a second predetermined range of a sample vehicle and preliminary sample map information within the second predetermined range of the sample vehicle;

obtaining a preliminary sample state of the sample vehicle;

determining a preliminary sample control signal based on the preliminary sample driving information, the preliminary sample state of the sample vehicle, and the preliminary control model;

determining a preliminary sample evaluation value associated with the preliminary control model based on a preliminary reward value corresponding to the preliminary sample control signal, wherein the preliminary reward value is associated with at least one of a collision condition associated with the preliminary sample control signal, a lane condition associated with the preliminary sample control signal, or a stop condition associated with the preliminary sample control signal;

determining whether the preliminary sample evaluation value satisfies a preset condition; and

designating the preliminary control model as the trained control model in response to the determination that the preliminary sample evaluation value satisfies the preset condition.

6. The system of claim 5 , wherein the training process further includes:

in response to the determination that the preliminary sample evaluation value does not satisfy the preset condition, performing one or more iterations for updating the preliminary control model until an updated sample evaluation value satisfies the preset condition, wherein each of the one or more iterations includes:

determining an updated control model based on a reward value in a previous iteration;

obtaining updated sample driving information;

estimating an updated sample state of the sample vehicle based on a sample control signal in the previous iteration and a sample state of the sample vehicle in the previous iteration;

determining an updated sample control signal based on the updated sample driving information, the updated sample state of the sample vehicle, and the updated control model;

determining an updated sample evaluation value associated with the updated control model based on an updated reward value corresponding to the updated sample control signal and a sample evaluation value in the previous iteration; and

determining whether the updated sample evaluation value satisfies the preset condition.

7. The system of claim 1 , wherein the trained control model includes a Deep Deterministic Policy Gradient (DDPG) network model.

8. A method implemented on a computing device having at least one processor, at least one storage medium, and a communication platform connected to a network, the method comprising:

obtaining driving information associated with a vehicle;

determining a state of the vehicle;

determining one or more candidate control signals and one or more evaluation values corresponding to the one or more candidate control signals based on the driving information and the state of the vehicle by using a trained control model;

selecting a target control signal from the one or more candidate control signals based on the one or more evaluation values according to a predetermined rule; and

transmitting the target control signal to a control component of the vehicle.

9. The method of claim 8 , wherein the driving information associated with the vehicle includes perception information within a first predetermined range of the vehicle and map information within the first predetermined range of the vehicle.

10. The method of claim 9 , wherein the perception information includes road condition information and obstacle information.

11. The method of claim 8 , wherein the state of the vehicle includes a velocity of the vehicle, an acceleration of the vehicle, and a steering wheel angle of the vehicle.

12. The method of claim 8 , wherein the trained control model is determined with a training process, the training process comprising:

obtaining a preliminary control model;

obtaining preliminary sample driving information, the preliminary sample driving information including preliminary sample perception information within a second predetermined range of a sample vehicle and preliminary sample map information within the second predetermined range of the sample vehicle;

obtaining a preliminary sample state of the sample vehicle;

determining a preliminary sample control signal based on the preliminary sample driving information, the preliminary sample state of the sample vehicle, and the preliminary control model;

determining a preliminary sample evaluation value associated with the preliminary control model based on a preliminary reward value corresponding to the preliminary sample control signal, wherein the preliminary reward value is associated with at least one of a collision condition associated with the preliminary sample control signal, a lane condition associated with the preliminary sample control signal, or a stop condition associated with the preliminary sample control signal;

determining whether the preliminary sample evaluation value satisfies a preset condition; and

designating the preliminary control model as the trained control model in response to the determination that the preliminary sample evaluation value satisfies the preset condition.

13. The method of claim 12 , wherein the training process further includes:

in response to the determination that the preliminary sample evaluation value does not satisfy the preset condition, performing one or more iterations for updating the preliminary control model until an updated sample evaluation value satisfies the preset condition, wherein each of the one or more iterations includes:

determining an updated control model based on a reward value in a previous iteration;

obtaining updated sample driving information;

estimating an updated sample state of the sample vehicle based on a sample control signal in the previous iteration and a sample state of the sample vehicle in the previous iteration;

determining an updated sample control signal based on the updated sample driving information, the updated sample state of the sample vehicle, and the updated control model;

determining an updated sample evaluation value associated with the updated control model based on an updated reward value corresponding to the updated sample control signal and a sample evaluation value in the previous iteration; and

determining whether the updated sample evaluation value satisfies the preset condition.

14. The method of claim 8 , wherein the trained control model includes a Deep Deterministic Policy Gradient (DDPG) network model.

15. A vehicle configured for autonomous driving, comprising:

a detecting component, a planning component, and a control component, wherein the planning component is configured to:

obtain driving information associated with the vehicle;

determine a state of the vehicle;

determine one or more candidate control signals and one or more evaluation values corresponding to the one or more candidate control signals based on the driving information and the state of the vehicle by using a trained control model;

select a target control signal from the one or more candidate control signals based on the one or more evaluation values according to a predetermined rule; and

transmit the target control signal to the control component of the vehicle.

16. The vehicle of claim 15 , wherein the driving information associated with the vehicle includes perception information within a first predetermined range of the vehicle and map information within the first predetermined range of the vehicle.

17. The vehicle of claim 15 , wherein the trained control model is determined with a training process, the training process comprising:

obtaining a preliminary control model;

obtaining preliminary sample driving information, the preliminary sample driving information including preliminary sample perception information within a second predetermined range of a sample vehicle and preliminary sample map information within the second predetermined range of the sample vehicle;

obtaining a preliminary sample state of the sample vehicle;

determining a preliminary sample control signal based on the preliminary sample driving information, the preliminary sample state of the sample vehicle, and the preliminary control model;

determining a preliminary sample evaluation value associated with the preliminary control model based on a preliminary reward value corresponding to the preliminary sample control signal, wherein the preliminary reward value is associated with at least one of a collision condition associated with the preliminary sample control signal, a lane condition associated with the preliminary sample control signal, or a stop condition associated with the preliminary sample control signal;

determining whether the preliminary sample evaluation value satisfies a preset condition; and

designating the preliminary control model as the trained control model in response to the determination that the preliminary sample evaluation value satisfies the preset condition.

18. The vehicle of claim 17 , wherein the training process further includes:

in response to the determination that the preliminary sample evaluation value does not satisfy the preset condition, performing one or more iterations for updating the preliminary control model until an updated sample evaluation value satisfies the preset condition, wherein each of the one or more iterations includes:

determining an updated control model based on a reward value in a previous iteration;

obtaining updated sample driving information;

estimating an updated sample state of the sample vehicle based on a sample control signal in the previous iteration and a sample state of the sample vehicle in the previous iteration;

determining an updated sample control signal based on the updated sample driving information, the updated sample state of the sample vehicle, and the updated control model;

determining an updated sample evaluation value associated with the updated control model based on an updated reward value corresponding to the updated sample control signal and a sample evaluation value in the previous iteration; and

determining whether the updated sample evaluation value satisfies the preset condition.

19. The vehicle of claim 16 , wherein the perception information includes road condition information and obstacle information.

20. The vehicle of claim 15 , wherein the state of the vehicle includes a velocity of the vehicle, an acceleration of the vehicle, and a steering wheel angle of the vehicle.