Systems and methods for navigating a vehicle
A navigational system for a host vehicle may comprise at least one processor. The processor may be programmed to receive an image representative of an environment of the host vehicle; analyze the image to identify a navigational state associated with the host vehicle; and determine, based on the navigational state, a navigational action for the host vehicle based on a policy that maps possible navigational actions to sensed states. The navigational action may be based on a safety constraint applicable to the navigational state, the safety constraint including a safety distance constraint associated with the host vehicle, wherein the safety distance constraint is based on a determined speed of the host vehicle and a determined speed of a detected target object. The processor may cause an adjustment of a navigational actuator of the host vehicle to implement the determined navigational action.
1. A navigation system for a host vehicle, the system comprising:
at least one processor comprising circuitry and a memory, wherein the memory incudes instructions that when executed by the circuitry cause the at least one processor to:
receive at least one image acquired by an image capture device, the at least one image being representative of an environment of the host vehicle;
analyze the at least one image to identify a navigational state associated with the host vehicle;
determine, based on the navigational state, a navigational action for the host vehicle based on a policy that maps possible navigational actions to sensed states, wherein the navigational action is based on at least one safety constraint applicable to the navigational state, wherein the at least one safety constraint includes a safety distance constraint associated with the host vehicle, wherein the safety distance constraint is based on a determined speed of the host vehicle and a determined speed of a detected target object, and wherein determining the navigational action includes identifying a plurality of future navigational states of the host vehicle and analyzing each of the plurality of future navigational states to determine a respective cost of each of the plurality of future navigational states, the navigational action being determined based on a comparison of the respective costs, wherein analyzing each of the plurality of future navigational states to determine the respective cost of each of the plurality of future navigational states includes applying a trained machine learning model to the plurality of future navigational states; and
cause at least one adjustment of a navigational actuator of the host vehicle to implement the determined navigational action.
2. The system of claim 1 , wherein the determined navigational action is further based on a projected trajectory of the host vehicle and a projected trajectory of the target object.
3. The system of claim 1 , wherein the safety distance constraint comprises at least one of a safe longitudinal distance to the target vehicle or a safe lateral distance to the target vehicle.
4. The system of claim 2 , wherein the predicted trajectory of the target object is based on at least one of the determined speed of the target object, a determined heading of the target object, or a heading of the host vehicle.
5. The system of claim 1 , wherein the at least one safety constraint includes a road curvature constraint.
6. The system of claim 1 , wherein the target object comprises a detected pedestrian associated with the navigational state, and wherein the safety distance constraint prevents the host vehicle from entering a buffer zone around the pedestrian.
7. The system of claim 6 , wherein the buffer zone is based on at least one of a projected trajectory of the pedestrian or a relative motion between the detected pedestrian and the host vehicle.
8. The system of claim 7 , wherein the projected trajectory of the pedestrian is determined from a speed or a heading of the detected pedestrian.
9. The system of claim 7 , wherein the relative motion between the detected pedestrian and the host vehicle is based on a projected trajectory of the host vehicle.
10. The system of claim 1 , wherein the navigational action is further based on at least one possible driving trajectory for the host vehicle, and wherein each of the at least one possible driving trajectory is associated with a respective future navigational state.
11. The system of claim 10 , wherein the navigational action is further based on a cost associated with a trajectory of the target object.
12. The system of claim 1 , wherein a first future navigational state of the plurality of future navigational states is associated with a first cost having a first value and a second future navigational state of the plurality of future navigational states is associated with a second cost having a second value, and wherein the first value and the second value are calculated according to a cost function.
13. The system of claim 12 , wherein navigational action is determined based on a comparison of the first value to the second value.
14. The system of claim 1 , wherein the navigational action is based on a change in the respective costs over a time-dependent quantity.
15. The system of claim 14 , wherein the time-dependent quantity is based on a future state of the host vehicle.
16. The system of claim 1 , wherein the at least one safety constraint further comprises at least one of a buffer time increasing an arrival time of the host vehicle relative to a leaving time of the target object at an intersection point or a give way designation assigned to the target object, the give way designation requiring the host vehicle to give way to the target object.
17. The system of claim 1 , wherein the navigational action comprises a lane change.
18. The system of claim 1 , wherein the target object comprises a detected stationary object, and wherein the safety distance constraint prevents the host vehicle from colliding with the stationary object.
19. The system of claim 1 , wherein the at least one safety constraint further includes a take way designation assigned to the target object.
20. The system of claim 19 , wherein the take way designation is based on at least one of a detected traffic sign or traffic light.
21. The system of claim 1 , wherein the determined navigational action gives way to the target object based on an observed rule break by the target object.
22. The system of claim 1 , wherein the at least one safety constraint includes at least one of a plurality of safety constraints that are assigned a different priority based on the navigational state or a second safety distance constraint associated with a second target object.
23. The system of claim 1 , wherein the navigational action is based on a determination that the target object is changing lanes.
24. The system of claim 23 , wherein the determination that the target vehicle is changing lanes is based on the speed and a determined heading of the target object.
25. The system of claim 23 , wherein the at least one processing device is further programmed to determine that the target object will likely cut into a lane of the host vehicle, and wherein the navigational action comprises a reduction in the speed of the host vehicle.
26. The system of claim 1 , wherein the navigational actuator includes at least one of a steering mechanism, an accelerator, or a brake.
27. A method for navigating a host vehicle, the method comprising:
receiving at least one image acquired by an image capture device, the at least one image being representative of an environment of the host vehicle;
analyzing the at least one image to identify a navigational state associated with the host vehicle;
determining, based on the navigational state, a navigational action for the host vehicle based on a policy that maps possible navigational actions to sensed states, wherein the navigational action is based on at least one safety constraint applicable to the navigational state, wherein the at least one safety constraint includes a safety distance constraint associated with the host vehicle, wherein the safety distance constraint is based on a determined speed of the host vehicle and a determined speed of a detected target object, and wherein determining the navigational action includes identifying a plurality of future navigational states of the host vehicle and analyzing each of the plurality of future navigational states to determine a respective cost of each of the plurality of future navigational states, the navigational action being determined based on a comparison of the respective costs, wherein analyzing each of the plurality of future navigational states to determine the respective cost of each of the plurality of future navigational states includes applying a trained machine learning model to the plurality of future navigational states; and
causing at least one adjustment of a navigational actuator of the host vehicle to implement the determined navigational action.
28. A non-transitory computer readable medium containing instructions that when executed by at least one processor, cause the at least one processor to perform operations comprising:
receiving at least one image acquired by an image capture device, the at least one image being representative of an environment of the host vehicle;
analyzing the at least one image to identify a navigational state associated with the host vehicle;
determining, based on the navigational state, a navigational action for the host vehicle based on a policy that maps possible navigational actions to sensed states, wherein the navigational action is based on at least one safety constraint applicable to the navigational state, wherein the at least one safety constraint includes a safety distance constraint associated with the host vehicle, wherein the safety distance constraint is based on a determined speed of the host vehicle and a determined speed of a detected target object, and wherein determining the navigational action includes identifying a plurality of future navigational states of the host vehicle and analyzing each of the plurality of future navigational states to determine a respective cost of each of the plurality of future navigational states, the navigational action being determined based on a comparison of the respective costs, wherein analyzing each of the plurality of future navigational states to determine the respective cost of each of the plurality of future navigational states includes applying a trained machine learning model to the plurality of future navigational states; and
causing at least one adjustment of a navigational actuator of the host vehicle to implement the determined navigational action.
29. The non-transitory computer readable medium of claim 28 , wherein the determined navigational action is further based on a projected trajectory of the host vehicle and a projected trajectory of the target object.
30. The non-transitory computer readable medium of claim 28 , wherein the target object comprises a detected pedestrian associated with the navigational state, and wherein the safety distance constraint prevents the host vehicle from entering a buffer zone around the pedestrian.
31. The non-transitory computer readable medium of claim 30 , wherein the buffer zone is based on a projected trajectory of the pedestrian.
32. The non-transitory computer readable medium of claim 28 , wherein the at least one safety constraint further comprises at least one of: a buffer time increasing an arrival time of the host vehicle relative to a leaving time of the target object at an intersection point; or a give way designation assigned to the target object, the give way designation requiring the host vehicle to give way to the target object.
33. The method of claim 27 , wherein the determined navigational action is further based on a projected trajectory of the host vehicle and a projected trajectory of the target object.
34. The method of claim 27 , wherein the target object comprises a detected pedestrian associated with the navigational state, and wherein the safety distance constraint prevents the host vehicle from entering a buffer zone around the pedestrian.
35. The method of claim 34 , wherein the buffer zone is based on a projected trajectory of the pedestrian.
36. The method of claim 27 , wherein the at least one safety constraint further comprises at least one of: a buffer time increasing an arrival time of the host vehicle relative to a leaving time of the target object at an intersection point; or a give way designation assigned to the target object, the give way designation requiring the host vehicle to give way to the target object.