Systems and methods for dynamic predictive control of autonomous vehicles

1. An in-vehicle control system for a semi-truck, comprising:

a processor; and

a computer-readable memory in communication with the processor and having stored thereon computer-executable instructions to cause the processor to at least:

receive a desired trajectory for autonomously driving the semi-truck and a vehicle status of the semi-truck,

determine a lateral dynamic model and a longitudinal kinematic model of the semi-truck based on the desired trajectory and the vehicle status,

model movement of the semi-truck over a prediction time horizon using the lateral dynamic model and the longitudinal kinematic model,

generate at least one control command for controlling a movement of the semi-truck based on the modelled movement, and

output the at least one control command for the autonomous driving of the semi-truck along the desired trajectory.

2. The system of claim 1 , wherein the memory further has stored thereon computer-executable instructions to cause the processor to:

update a dynamic non-linear model of the semi-truck based on the desired trajectory and the vehicle status,

wherein determining the lateral dynamic model and the longitudinal kinematic model is further based on an approximation of the dynamic non-linear model.

3. The system of claim 2 , wherein updating the dynamic non-linear model is based on small angle error assumptions.

4. The system of claim 2 , wherein the dynamic non-linear model is configured to predict the movement of the semi-truck in response to a potential control command during autonomous driving of the semi-truck.

5. The system of claim 2 , wherein determining the lateral dynamic model and the longitudinal kinematic model includes transforming a reference frame for the dynamic non-linear model from a global inertial coordinate frame to a reference frame coordinate centered on a current position of the semi-truck on a road.

6. The system of claim 1 , wherein modelling movement of the semi-truck comprises:

determining a first quadratic program problem based on the lateral dynamic model, and

determining a second quadratic program problem based on the longitudinal kinematic model.

7. The system of claim 6 , wherein the memory further has stored thereon computer-executable instructions to cause the processor to:

solving the first quadratic program problem to generate a first solution, and

solving the second quadratic program problem to generate a second solution, wherein generating at least one control command is further based on the first solution and the second solution.

8. The system of claim 5 , wherein the memory further has stored thereon computer-executable instructions to cause the processor to:

determine a road referenced error model based at least in part on a comparison between the transformed dynamic non-linear model and the desired trajectory.

9. The system of claim 8 , wherein the road referenced error model comprises a heading error and a relative lateral position error.

10. The system of claim 1 , wherein the modelling of the movement of the semi-truck comprises:

predicting the movement of the semi-truck using the lateral dynamic model and the longitudinal kinematic model, an average reference trajectory velocity of the semi-truck, and a curvature at a waypoint of the desired trajectory.

11. The system of claim 1 , wherein:

determining the lateral dynamics model is further based at least in part on one or more internal dynamic parameters of the semi-truck.

12. The system of claim 11 , wherein the one or more internal dynamic parameters include one or more of the following: a mass of the semi-truck, a moment of inertia of the semi-truck, a trailer angle, a cornering stiffness, and an articulation between a tractor and a trailer of the semi-truck.

13. A non-transitory computer readable storage medium having stored thereon instructions that, when executed, cause at least one computing device to at least:

receive a desired trajectory for autonomously driving a semi-truck and a vehicle status of the semi-truck;

determine a lateral dynamic model and a longitudinal kinematic model of the semi-truck based on the desired trajectory and the vehicle status;

model movement of the semi-truck over a prediction time horizon using the lateral dynamic model and the longitudinal kinematic model;

generate at least one control command for controlling a movement of the semi-truck based on the modelled movement; and

output the at least one control command for the autonomous driving of the semi-truck along the desired trajectory.

14. The non-transitory computer readable storage medium of claim 13 , wherein:

the modelling of the movement of the semi-truck comprises determining a first quadratic program problem based on the lateral dynamic model and a second quadratic program problem based on the longitudinal kinematic model, and

the generating of the at least one control command comprises solving the first and second quadratic program problems.

15. The non-transitory computer readable storage medium of claim 14 , wherein the first and second quadratic program problems decouple lateral and longitudinal control objectives of a dynamic linear time-invariant model.

16. The non-transitory computer readable storage medium of claim 14 , wherein determining the first quadratic program problem and the second quadratic program problem is based on minimizing an error state at every reference point.

17. The non-transitory computer readable storage medium of claim 14 , wherein determining the first quadratic program problem and the second quadratic program problem is based on penalizing using command inputs and change in the command inputs over time.

18. A method for controlling movement of a semi-truck, comprising:

receiving a desired trajectory for autonomously driving a semi-truck and a vehicle status of the semi-truck;

determining a lateral dynamic model and a longitudinal kinematic model of the semi-truck based on the desired trajectory and the vehicle status;

modelling movement of the semi-truck over a prediction time horizon using the lateral dynamic model and the longitudinal kinematic model;

generating at least one control command for controlling a movement of the semi-truck based on the modelled movement; and

outputting the at least one control command for the autonomous driving of the semi-truck along the desired trajectory.

19. The method of claim 18 , wherein the desired trajectory defines at least one of a position, heading, speed, and time in a local coordinate frame of the semi-truck.

20. The method of claim 18 , wherein the desired trajectory is discretized at a time interval to define a plurality of waypoints and a speed profile for the semi-truck.