Method and system for data-driven and modular decision making and trajectory generation of an autonomous agent
A system for data-driven, modular decision making and trajectory generation includes a computing system. A method for data-driven, modular decision making and trajectory generation includes: receiving a set of inputs; selecting a learning module such as a deep decision network and/or a deep trajectory network from a set of learning modules; producing an output based on the learning module; repeating any or all of the above processes; and/or any other suitable processes. Additionally or alternatively, the method can include training any or all of the learning modules; validating one or more outputs; and/or any other suitable processes and/or combination of processes.
1. A method for operating an autonomous agent, the method comprising:
with a first learned model, constraining a set of actions available to the autonomous agent to a single action;
with a rule and based on the single action, selecting a second learned model from a set of learned models;
with the second learned model, determining a trajectory for the autonomous agent; and
operating the autonomous agent based on the trajectory.
2. The method of claim 1 , wherein the set of learned models comprises multiple learned models.
3. The method of claim 1 , wherein each of the set of learned models is trained for a particular subset of actions of the set of actions.
4. The method of claim 3 , wherein each of the particular subset of actions is one action.
5. The method of claim 1 , wherein the first learned model is part of a second set of learned models.
6. The method of claim 5 , wherein the second set of learned models comprises multiple learned models.
7. The method of claim 6 , wherein the first learned model is trained for an environmental feature associated with the autonomous agent.
8. The method of claim 7 , wherein each of the second set of learned models is trained for a particular environmental feature associated with the autonomous agent.
9. The method of claim 1 , further comprising calculating a constrained region for the autonomous agent based on the single action, wherein the trajectory is further determined based on the constrained region.
10. The method of claim 9 , wherein the constrained region comprises a set of locations which the autonomous agent can occupy based on the single action.
11. The method of claim 10 , wherein the constrained region is further defined based on a location of the autonomous agent and a map.
12. The method of claim 1 , wherein at least one of the first and second learned models is trained with reinforcement learning.
13. The method of claim 12 , wherein both of the first and second learned models is trained with reinforcement learning.
14. The method of claim 1 , wherein the rule is a predetermined rule.
15. The method of claim 14 , wherein the predetermined rule comprises a mapping.
16. A method for operating an autonomous agent with a hybrid decision making architecture, the method comprising:
with a first learned model, selecting a single action from a first set of actions;
with a predetermined mapping and based on the single action, selecting a second learned model from a set of learned models;
with the second learned model, determining a trajectory for the autonomous agent; and
operating the autonomous agent based on the trajectory.
17. The method of claim 16 , wherein each of the first and second learned models is a neural network trained with at least one of reinforcement learning and inverse reinforcement learning.
18. The method of claim 16 , further comprising calculating a constrained region for the autonomous agent based on the single action, wherein the constrained region comprises a set of locations which the autonomous agent can occupy based on the single action, and wherein the trajectory is further determined based on the constrained region.
19. The method of claim 18 , wherein the constrained region is further defined based on a location of the autonomous agent and a map.