Analysis of scenarios for controlling vehicle operations
Techniques are described herein for determining one or more actions for an autonomous vehicle to perform, based on simulation of at least one possible scenario. A possible scenario may involve, for example, the autonomous vehicle interacting with an object in the environment. The possible scenario may be simulated by modifying a first internal map containing information about the autonomous vehicle and the environment. As part of the simulation, one or more parameters of the first internal map can be modified in order to, for example, determine the state of the object at a particular point in the future. Based on the modification of the one or more parameters, a second internal map representing a possible scenario is generated from the first internal map. Both the first internal map and the second internal map can be evaluated to decide which action to take.
1. A method comprising:
determining, by a controller system configured to control an autonomous operation of a vehicle, a goal to be achieved by the autonomous operation of the vehicle;
obtaining, by the controller system, a decision tree comprising nodes that represent conditions to be evaluated for making a decision on what action to take in order to achieve the goal;
generating, by the controller system, a first internal map based on sensor data from a plurality of sensors, the first internal map comprising a three-dimensional representation of an environment around the vehicle, information on a current state of the vehicle, and information about the environment;
pruning, by the controller system, the decision tree to prevent one or more of the conditions from being evaluated, wherein the decision tree is pruned based on the goal to be achieved;
modifying, by the controller system, one or more parameters of the first internal map that pertain to the conditions of the nodes that remain after pruning the decision tree;
generating, by the controller system, a second internal map from the first internal map, based on modification of the one or more parameters; and
evaluating, by the controller system, the pruned decision tree using the second internal map to make the decision on one or more actions for the vehicle to perform in order to achieve the goal.
2. The method of claim 1 , further comprising presenting, by the controller system and a user interface, the one or more actions to a user.
3. The method of claim 1 , further comprising controlling, by the controller system, the autonomous operation of the vehicle to achieve the goal using the one or more actions.
4. The method of claim 1 , wherein the modifying the one or more parameters comprises changing an attribute of at least one of a speed of an object, a direction of the object, or a distance of the object to simulate a possible scenario.
5. The method of claim 4 , wherein the object is located in the environment, wherein the one or more parameters correspond to the attribute of the object, and wherein the second internal map represents a possible state of the object.
6. The method of claim 5 , wherein multiple internal maps are generated from the first internal map, the multiple internal maps include the second internal map, each of the multiple internal maps representing a different possible state of the object, and wherein the one or more actions are determined based on evaluation of each of the multiple internal maps.
7. The method of claim 1 , further comprising selecting, by the controller system, an action from the one or more actions that maximizes a reward function, and controlling, by the controller system, the autonomous operation of the vehicle to achieve the goal using the selected action.
8. A non-transitory computer-readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform operations including:
determining a goal to be achieved by autonomous operation of a vehicle;
obtaining a decision tree comprising nodes that represent conditions to be evaluated for making a decision on what action to take in order to achieve the goal;
generating a first internal map based on sensor data from a plurality of sensors, the first internal map comprising a three-dimensional representation of an environment around the vehicle, information on a current state of the vehicle, and information about the environment;
pruning the decision tree to prevent one or more of the conditions from being evaluated, wherein the decision tree is pruned based on the goal to be achieved;
modifying one or more parameters of the first internal map that pertain to the conditions of the nodes that remain after pruning the decision tree;
generating a second internal map from the first internal map, based on modification of the one or more parameters; and
evaluating the pruned decision tree using the second internal map to make the decision on one or more actions for the vehicle to perform in order to achieve the goal.
9. The non-transitory computer-readable medium of claim 8 , wherein the operations further include presenting, by a user interface, the one or more actions to a user.
10. The non-transitory computer-readable medium of claim 8 , wherein the operations further include controlling the autonomous operation of the vehicle to achieve the goal using the one or more actions.
11. The non-transitory computer-readable medium of claim 8 , wherein the modifying the one or more parameters comprises changing an attribute of at least one of a speed of an object, a direction of the object, or a distance of the object to simulate a possible scenario.
12. The non-transitory computer-readable medium of claim 11 , wherein the object is located in the environment, wherein the one or more parameters correspond to the attribute of the object, and wherein the second internal map represents a possible state of the object.
13. The non-transitory computer-readable medium of claim 12 , wherein multiple internal maps are generated from the first internal map, the multiple internal maps include the second internal map, each of the multiple internal maps representing a different possible state of the object, and wherein the one or more actions are determined based on evaluation of each of the multiple internal maps.
14. The non-transitory computer-readable medium of claim 8 , wherein the operations further include selecting an action from the one or more actions that maximizes a reward function, and controlling, by the controller system, the autonomous operation of the vehicle to achieve the goal using the selected action.
15. A system comprising:
one or more processors;
a memory coupled to the one or more processors, the memory storing a plurality of instructions executable by the one or more processors, the plurality of instructions comprising instructions that when executed by the one or more processors cause the one or more processors to perform processing comprising:
determining a goal to be achieved by autonomous operation of a vehicle;
obtaining a decision tree comprising nodes that represent conditions to be evaluated for making a decision on what action to take in order to achieve the goal;
generating a first internal map based on sensor data from a plurality of sensors, the first internal map comprising a three-dimensional representation of an environment around the vehicle, information on a current state of the vehicle, and information about the environment;
pruning the decision tree to prevent one or more of the conditions from being evaluated, wherein the decision tree is pruned based on the goal to be achieved;
modifying one or more parameters of the first internal map that pertain to the conditions of the nodes that remain after pruning the decision tree;
generating a second internal map from the first internal map, based on modification of the one or more parameters; and
evaluating the pruned decision tree using the second internal map to make the decision on one or more actions for the vehicle to perform in order to achieve the goal.
16. The system of claim 15 , wherein the processing further comprises presenting, by a user interface, the one or more actions to a user.
17. The system of claim 15 , wherein the processing further comprises controlling the autonomous operation of the vehicle to achieve the goal using the one or more actions.
18. The system of claim 15 , wherein the modifying the one or more parameters comprises changing an attribute of at least one of a speed of an object, a direction of the object, or a distance of the object to simulate a possible scenario.
19. The system of claim 18 , wherein the object is located in the environment, wherein the one or more parameters correspond to the attribute of the object, and wherein the second internal map represents a possible state of the object.
20. The system of claim 19 , wherein multiple internal maps are generated from the first internal map, the multiple internal maps include the second internal map, each of the multiple internal maps representing a different possible state of the object, and wherein the one or more actions are determined based on evaluation of each of the multiple internal maps.