Framework for controlling devices

1. A computer-implemented method comprising:

training an anti-fish-startling model evaluation engine to output actions for accomplishing objectives while reducing a startling effect on nearby fish using training data that includes, for different, previously performed actions, (i) one or more sensed conditions associated with performing the different actions in an aquaculture environment, (ii) objectives associated with the different actions, and (iii) indications of whether nearby fish were startled when the different actions were previously performed;

determining a particular objective for a robot that is operating in the aquaculture environment;

determining one or more particular sensed conditions that are associated with the aquaculture environment;

providing the particular objective and the particular sensed conditions to the trained anti-fish-startling model evaluation engine;

receiving, from the trained anti-fish-startling model evaluation engine, one or more particular actions for accomplishing the particular objective while reducing the startling effect on nearby fish; and

transmitting the one or more particular actions to another device.

2. The method of claim 1 , wherein the particular objective is illuminating an illumination device.

3. The method of claim 1 , wherein the particular objective is moving the robot.

4. The method of claim 3 , wherein the one or more particular actions comprise illuminating an illumination device over a configured period of time.

5. The method of claim 1 , wherein the anti-fish-startling model evaluation engine evaluates an output of a machine learning model.

6. The method of claim 1 , wherein determining the one or more particular actions includes evaluating one or more rules.

7. The method of claim 1 , wherein the objective is introducing a device into the environment.

8. A non-transitory, computer-readable medium storing one or more instructions executable by a computer system to perform operations comprising:

training an anti-fish-startling model evaluation engine to output actions for accomplishing objectives while reducing a startling effect on nearby fish using training data that includes, for different, previously performed actions, (i) one or more sensed conditions associated with performing the different actions in an aquaculture environment, (ii) objectives associated with the different actions, and (iii) indications of whether nearby fish were startled when the different actions were previously performed;

determining a particular objective for a robot that is operating in the aquaculture environment;

determining one or more particular sensed conditions that are associated with the aquaculture environment;

providing the particular objective and the particular sensed conditions to the trained anti-fish-startling model evaluation engine;

receiving, from the trained anti-fish-startling model evaluation engine, one or more particular actions for accomplishing the particular objective while reducing the startling effect on nearby fish; and

transmitting the one or more particular actions to another device.

9. The non-transitory, computer-readable medium of claim 8 , wherein the particular objective is illuminating an illumination device.

10. The non-transitory, computer-readable medium of claim 8 , wherein the particular objective is moving the robot.

11. The non-transitory, computer-readable medium of claim 10 , wherein the one or more particular actions comprise the action comprises illuminating an illumination device over a configured period of time.

12. The non-transitory, computer-readable medium of claim 8 , wherein the anti-fish-startling model evaluation engine evaluates an output of a machine learning model.

13. The non-transitory, computer-readable medium of claim 8 , wherein determining the one or more particular actions includes evaluating one or more rules.

14. The non-transitory, computer-readable medium of claim 8 , wherein the objective is introducing a device into the environment.

15. A system comprising:

one or more computers; and

one or more non-transitory computer-readable media that store instructions that, when executed by the one or more computers, cause the one or more computers to perform operations comprising:

training an anti-fish-startling model evaluation engine to output actions for accomplishing objectives while reducing a startling effect on nearby fish using training data that includes, for different, previously performed actions, (i) one or more sensed conditions associated with performing the different actions in an aquaculture environment, (ii) objectives associated with the different actions, and (iii) indications of whether nearby fish were startled when the different actions were previously performed;

determining a particular objective for a robot that is operating in an in the aquaculture environment;

determining one or more particular sensed conditions that are associated with the aquaculture environment;

providing the particular objective and the particular sensed conditions to the trained anti-fish-startling model evaluation engine;

receiving, from the trained anti-fish-startling model evaluation engine, one or more particular actions for accomplishing the particular objective while reducing the startling effect on nearby fish; and

transmitting the one or more particular actions to another device.

16. The computer-implemented system of claim 15 , wherein the particular objective is illuminating an illumination device.

17. The computer-implemented system of claim 15 , wherein the particular objective is moving the robot.

18. The computer-implemented system of claim 17 , wherein the one or more particular actions comprise illuminating an illumination device over a configured period of time.

19. The computer-implemented system of claim 15 , wherein the anti-fish-startling model evaluation engine evaluates an output of a machine learning model.

20. The computer-implemented system of claim 15 , wherein determining the one or more particular actions includes evaluating one or more rules.