System and method for using human driving patterns to manage speed control for autonomous vehicles
A system and method for using human driving patterns to manage speed control for autonomous vehicles are disclosed. A particular embodiment includes: generating data corresponding to desired human driving behaviors; training a human driving model module using a reinforcement learning process and the desired human driving behaviors; receiving a proposed vehicle speed control command; determining if the proposed vehicle speed control command conforms to the desired human driving behaviors by use of the human driving model module; and validating or modifying the proposed vehicle speed control command based on the determination.
1. A system comprising:
a data processor; and
a vehicle speed control module, executable by the data processor, configured to:
receive a proposed vehicle speed control command prior to commanding a vehicle control subsystem of an autonomous vehicle to perform a maneuver corresponding to the proposed vehicle speed control command;
determine if the proposed vehicle speed control command conforms to standards of human driving behaviors by use of a human driving model module, executable by the data processor;
validate or modify the proposed vehicle speed control command based on the determination; and
cause the autonomous vehicle to perform a maneuver corresponding to the validated or modified vehicle speed control command.
2. The system of claim 1 wherein the vehicle speed control module is further configured to train the human driving model module using a reinforcement learning process comprising a simulation training, wherein data generated by the human driving model module is run against data corresponding to the human driving behaviors during the simulation training phase.
3. The system of claim 1 wherein the vehicle speed control module is further configured to train the human driving model module using a reinforcement learning process comprising an actual on-the-road training phase, wherein data generated by the human driving model module is run against data corresponding to the human driving behaviors captured by sensors of the autonomous vehicle.
4. The system of claim 1 wherein the vehicle speed control module is further configured to train the human driving model module by modifying parameters in the human driving model module based on a reinforcement learning process, wherein the parameters are used for validating or modifying the proposed vehicle speed control command.
5. The system of claim 4 wherein the parameters are corresponding to at least one of a speed, a braking, and a heading of the autonomous vehicle.
6. The system of claim 1 wherein the human driving model module is trained with data corresponding to the human driving behaviors.
7. The system of claim 1 wherein the vehicle speed control module is further configured to train the human driving model module by determining a current state of the autonomous vehicle and determining a deviation between the current state of the autonomous vehicle and a state corresponding to the human driving behaviors, wherein parameters in the human driving model module are modified based on the deviation between the current state of the autonomous vehicle and the state corresponding to the human driving behaviors.
8. The system of claim 7 wherein a trained human driving model module is having modified parameters, wherein the deviation determined by the human driving model module is larger than the deviation determined by the trained human driving model module.
9. A method comprising:
receiving a proposed vehicle speed control command prior to commanding a vehicle control subsystem of an autonomous vehicle to perform a maneuver corresponding to the proposed vehicle speed control command;
determining if the proposed vehicle speed control command conforms to standards of human driving behaviors by use of a human driving model module, executable by a data processor;
validating or modifying the proposed vehicle speed control command based on the determination; and
cause the autonomous vehicle to perform a maneuver corresponding to the validated or modified vehicle speed control command.
10. The method of claim 9 further comprising training the human driving model module with data corresponding to the human driving behaviors, wherein the data corresponding to the human driving behaviors is encoded as a neural network or a rules set.
11. The method of claim 9 further comprising outputting the validated or modified vehicle speed control command to the vehicle control subsystem, causing the autonomous vehicle to perform the maneuver, wherein the maneuver comprises following a trajectory.
12. The method of claim 9 further comprising capturing data through vehicle sensor subsystems and driving simulation data to model the human driving behaviors.
13. The method of claim 9 further comprising training the human driving model module with data corresponding to the human driving behaviors, wherein the data corresponding to the human driving behaviors is captures by sensors of the autonomous vehicle, wherein the data corresponding to the human driving behaviors comprises at least a position, an acceleration, an incremental speed, and a target speed of the autonomous vehicle.
14. The method of claim 9 comprising training the human driving model module based on a reinforcement learning process comprising an actual on-the-road training phase, with data corresponding to the human driving behaviors, wherein the data corresponding to the human driving behaviors comprises incremental speeds of the autonomous vehicle at time steps.
15. The method of claim 14 wherein data of the incremental speeds of the autonomous vehicle at the time steps, generated by the human driving model module, is a first data, wherein data of the incremental speeds of the autonomous vehicle at the time steps, captured by sensors of the autonomous vehicle, is a second data, wherein the first data is compared with the second data during the actual on-the-road training phase.
16. The method of claim 15 wherein a difference between the first data and the second data is calculated and compared to a threshold.
17. The method of claim 16 wherein parameters of the human driving model module are used for validating or modifying the proposed vehicle speed control command, wherein if the difference between the first data and the second data is greater than the threshold, the corresponding parameters are modified.
18. A non-transitory machine-useable storage medium embodying instructions which, when executed by a machine, cause the machine to:
receive a proposed vehicle speed control command prior to commanding a vehicle control subsystem of an autonomous vehicle to perform a maneuver corresponding to the proposed vehicle speed control command;
determine if the proposed vehicle speed control command conforms to standards of human driving behaviors by use of a human driving model module, executable by the machine;
validate or modify the proposed vehicle speed control command based on the determination; and
cause the autonomous vehicle to perform a maneuver corresponding to the validated or modified vehicle speed control command.
19. The non-transitory machine-useable storage medium of claim 18 being further configured to train the human driving model module by modifying parameters in the human driving model module based on a reinforcement learning process comprising an actual on-the-road training phase, wherein data generated by the human driving model module is a first data, wherein data captured by sensors of the autonomous vehicle is a second data, wherein a difference between the first data and the second data is calculated and compared to a threshold.
20. The non-transitory machine-useable storage medium of claim 19 wherein the parameters are used for validating or modifying the proposed vehicle speed control command, wherein if the difference between the first data and the second data is greater than the threshold, the corresponding parameters are modified.