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:
receiving a set of inputs;
selecting a first learned model from a first set of learned models based on the set of inputs;
selecting a second learned model from a second set of learned models based on an output of the first learned model;
determining a trajectory for the autonomous agent based on the second learned model; and
operating the autonomous agent based on the trajectory.
2. The method of claim 1 , further comprising validating the trajectory.
3. The method of claim 2 , wherein the trajectory is validated based on at least one of: an uncertainty associated with the trajectory and a satisfaction of a set of predetermined rules.
4. The method of claim 1 , wherein the set of inputs comprises a set of sensor inputs collected from a sensor system onboard the autonomous agent.
5. The method of claim 4 , wherein the set of inputs further comprises a context of the autonomous agent.
6. The method of claim 5 , further comprising determining the context based on a map.
7. The method of claim 6 , wherein the map prescribes a set of multiple contexts, wherein the context is selected from the set of multiple contexts.
8. The method of claim 5 , wherein the context is at least partially determined based on a fixed route associated with the autonomous agent.
9. The method of claim 1 , wherein the output comprises an action of the autonomous agent.
10. The method of claim 1 , wherein at least one of the first learned model and the second learned model is selected based on a set of mappings.
11. The method of claim 1 , wherein the first and second set of learned models comprise neural networks, wherein the neural networks are trained with at least one of: inverse reinforcement learning and reinforcement learning.
12. A system for operating an autonomous agent, the system comprising:
a sensor system;
a first set of learned models;
a second set of learned models; and
a computing system, wherein the computing system:
receives a set of inputs, wherein at least a portion of the set of inputs is received from the sensor system;
determines a first learned model based on the set of inputs and a first set of learned models;
determines a second learned model based on an output of the first learned model and a second set of learned models;
determines a trajectory for the autonomous agent based on the second learned model; and
operates the autonomous agent based on the trajectory.
13. The system of claim 12 , wherein the sensor system comprises a location sensor and an orientation sensor, and wherein the set of inputs comprises a set of poses associated with the autonomous agent.
14. The system of claim 12 , wherein the set of inputs comprises a context associated with the autonomous agent.
15. The system of claim 14 , further comprising a map, wherein the context is determined based on the map.
16. The system of claim 15 , wherein the map comprises a set of labeled contexts, wherein the context is selected from the set of labeled contexts based on a location of the autonomous agent.
17. The system of claim 14 , wherein the context is determined based on a fixed route associated with the autonomous agent.
18. The system of claim 12 , wherein the computing system further validates the trajectory based on at least one of: an uncertainty associated with the trajectory and a satisfaction of a set of predetermined rules.
19. The system of claim 12 , wherein the output comprises an action of the autonomous agent.
20. The system of claim 12 , wherein each of the first and second learned models comprises a neural network, wherein the neural network is trained with inverse reinforcement learning.