Method and system for conditional operation of an autonomous agent

1. A method for operating an autonomous vehicle, the method comprising:

selecting a first set of policies for evaluation by the autonomous vehicle, the first set of policies comprising:

a set of single action policies;

a set of multiple action policies, wherein each of the set of multiple action policies prescribes:

a set of multiple actions; and

a set of trigger conditions, wherein each of the set of trigger conditions is associated with a transition between consecutive actions of the set of multiple actions;

evaluating the first set of policies, wherein evaluating the first set of policies comprises:

for each policy of the first set of policies:

simulating, over a predetermined simulation time period in the future, a behavior of the autonomous vehicle and a behavior of each of a set of tracked agents in an environment of the autonomous vehicle in response to the autonomous vehicle executing the policy;

determining a quantitative metric for the policy based on the behavior of the autonomous vehicle and the behaviors of the set of tracked agents;

selecting a policy from the first set of policies based on the set of quantitative metrics;

operating the autonomous vehicle according to the selected policy, wherein the selected policy comprises a multiple action policy of the set of multiple action policies, comprising:

implementing a first action of the set of multiple actions of the selected multiple action policy;

while the first action is being implemented and according to a predetermined election cycle period, the predetermined election cycle period having a shorter duration than the predetermined simulation time period, selecting a second set of policies for evaluation by the autonomous vehicle;

evaluating the second set of policies and selecting a second policy based on the evaluation;

refraining from completing a remainder of the selected multiple action policy;

operating the autonomous vehicle according to the selected second policy.

2. The method of claim 1 , wherein at least a portion of the set of trigger conditions depend on a progression of the set of tracked agents.

3. The method of claim 1 , wherein at least a portion of the set of multiple action policies of the first set of policies is selected based on a location of the autonomous vehicle.

4. The method of claim 3 , wherein at least a second portion of the set of multiple action policies of the first set of policies is predetermined and selected independently of the location.

5. The method of claim 4 , wherein the second portion comprises a multiple action policy configured to maneuver around an obstacle.

6. The method of claim 3 , wherein the location is determined based on sensor data collected at a set of sensors onboard the autonomous vehicle.

7. The method of claim 6 , wherein the portion of the set of multiple action policies is further determined based on referencing a labeled map based on the location, the labeled map comprising a predetermined set of label assignments.

8. The method of claim 7 , wherein the location overlaps with a particular label assignment of the predetermined set of label assignments, the particular label assignment corresponding to a particular scenario in the environment.

9. The method of claim 8 , wherein the scenario comprises at least one of a crosswalk, intersection, or parking lot.

10. The method of claim 1 , wherein at least a portion of the set of trigger conditions are implemented in response to the set of tracked agents following a set of right-of-way driving conventions during an associated simulation performed while evaluating the first set of policies.

11. The method of claim 1 , wherein the predetermined election cycle period is less than 1/10 of the time of the predetermined simulation time period.

12. The method of claim 1 , wherein the selected second policy comprises a particular single action policy of the set of single action policies.

13. The method of claim 12 , wherein the particular single action policy comprises a second action of the selected multiple action policy.

14. A method for operating an autonomous vehicle, the method comprising:

selecting a set of policies for evaluation by the autonomous vehicle, the set of policies comprising:

a set of single action policies;

a set of multiple action policies, wherein each of the set of multiple action policies prescribes:

a set of multiple actions; and

a set of trigger conditions associated with the set of multiple actions;

evaluating the set of policies, wherein evaluating the set of policies comprises:

for each policy of the set of policies:

simulating, over a predetermined simulation time period in the future, a movement of the autonomous vehicle and a movement of each of a set of tracked agents in an environment of the autonomous vehicle;

determining a quantitative metric for the policy based on the simulation;

selecting a policy from the set of policies based on the set of quantitative metrics;

operating the autonomous vehicle according to the selected policy, wherein the selected policy comprises a multiple action policy of the set of multiple action policies, comprising:

implementing a first action of the set of multiple actions of the selected multiple action policy;

checking for satisfaction of a first trigger condition of the set of trigger conditions; and

in an event that the first trigger condition is satisfied, transitioning operation of the autonomous vehicle to a second action of the set of multiple actions.

15. The method of claim 14 , wherein at least a portion of the set of trigger conditions depend on a progression of the set of tracked agents.

16. The method of claim 14 , wherein at least a portion of the set of multiple action policies of the set of policies is selected based on a location of the autonomous vehicle.

17. The method of claim 16 , wherein at least a second portion of the set of multiple action policies of the set of policies is predetermined and selected independently of the location.

18. The method of claim 17 , wherein the second portion comprises a multiple action policy configured to maneuver around an obstacle.

19. The method of claim 17 , wherein the portion of the set of multiple action policies is further determined based on referencing a labeled map based on the location and determining a predetermined scenario label based on referencing the labeled map.

20. The method of claim 19 , wherein the predetermined scenario label comprises at least one of a crosswalk, intersection, or parking lot.