Multi-tiered command processing

A system comprising:

• a robot that includes:
• one or more physically moveable components,
• one or more microphones,
• a control system configured to actuate the one or more physically moveable components,
• an audio processing system that includes:
• a local recognition model configured to detect, without communication over a network and using the one or more microphones, utterance of voice commands in a set of first commands that control behavior of the robot,
• a command selection module configured to distinguish between voice commands in the set of first commands, voice commands in a set of second commands that control behavior of the robot, and voice commands in a set of application commands that include commands different from the first commands and second commands, and wherein the command selection module prioritizes detection of the first commands first, the second commands second, and the application commands third, and
• a hotword model configured to recognize utterance of a reserved word;
• and
• a remote server communicatively coupled to the robot, wherein the remote server is configured to provide a speech recognition service configured to detect voice commands in the set of second commands and voice commands in the set of application commands;
• wherein the robot is configured to receive audio data for an utterance and process the audio data for the utterance by (i) using the local recognition model evaluate the audio data for the utterance with respect to each of the set of first commands, and (ii) using the speech recognition service provided by the remote server to evaluate the audio data for the utterance with respect to the set of second commands and the set of application commands.

One or more computer-readable storage devices storing a computer program, the program comprising instructions that when executed by one or more computers cause the one or more computers to perform operations comprising:

• operating, by the robot, an audio processing system to detect utterance of voice commands in a set of first commands that control behavior of the robot and in a set of second commands that control behavior of the robot, wherein the audio processing system is configured to detect utterance of voice commands in the set of first commands using a local recognition model of the robot that does not require communication over a network, and the audio processing system is configured to detect utterance of commands in the set of second commands through communication with a server over a network;
• executing, by the robot, one or more applications that are configured to respond to voice commands in a set of application commands that include commands different from the first commands and second commands;
• and
• receiving, by the robot, audio data for an utterance;
• and
• processing, by the robot using the audio processing system, the audio data for the utterance to evaluate the utterance with respect to each of the set of first commands, the set of second commands, and the set of application commands, wherein the processing is performed according to a predetermined hierarchy that prioritizes detection of the first commands first, the second commands second, and the application commands third, wherein the audio processing system is configured to detect utterance of the first commands, the second commands, and the application commands during execution of the one or more applications by the robot.