Continual proactive learning for autonomous robot agents

A method of operating a robot agent including a movable physical component, said robot agent being a forklift, the method comprising:

• predicting, at the robot agent, at least a first action to be implemented by the robot agent in performing a task assigned to the robot agent, said predicting being based on at least based on a predefined difficulty ranking, said predefined difficulty ranking being in the form of a numeric value, corresponding to the first action to be performed;
• operating the robot agent to determine a mode of operation to be used to implement the first action based on whether the robot agent is predicted to fail or succeed in performing the first action, the robot agent determining to use an unguided mode of operation to perform the first action in response to predicting the robot agent will succeed in performing the first action and determining to use a guided mode of operation in response to predicting the robot agent will fail in performing the first action;
• and
• performing the first action by the robot agent using the determined mode of operation;
• and
• wherein the first action to be performed is one of:
• i) moving forks of the forklift in a vertical plane only while the forklift is stationary and ii) placing the forks of the forklift into a pallet located at a top shelf of a pallet rack;
• and
• wherein a first predefined difficulty ranking corresponding to the action of moving forks of the forklift in a vertical plane only while the forklift is stationary indicates a lower difficulty level than a predefined difficulty ranking corresponding to the action of placing the forks of the forklift into a pallet located at the top shelf of a pallet rack.

A robot agent comprising:

• an electro-mechanical subsystem comprising one or more electro-mechanical components;
• a sensor subsystem comprising one or more sensors;
• and
• a computer hardware subsystem to execute one or more sets of executable instructions, the one or more sets of executable instructions, when executed, manipulate the robot agent to:
• predict, based on at least a set of pre-defined actions which are actions, predicted to fail regardless of other considerations, whether the robot agent will fail or succeed in performing the first action, said predicting including predicting failure when the first action is one of the pre-defined actions, which are actions predicted to fail regardless of other considerations, said step of predicting whether the robot agent will fail or succeed in performing the first action being performed prior to the robot agent attempting to perform said first action;
• and wherein moving a pallet through a zone in which a human is expected to be present is one of the pre-defined actions in the set of pre-defined actions, which are actions predicted to fail regardless of other considerations;
• operate the robot agent to determine a mode of operation to be used to implement the first action based on whether the robot agent is predicted to fail or succeed in performing the first action, the robot agent determining to use an unguided mode of operation to perform the first action in response to predicting the robot agent will succeed in performing the first action and determining to use a guided mode of operation in response to predicting the robot agent will fail in performing the first action;
• and
• manipulate the electro-mechanical subsystem to perform the first action using the determined mode of operation.

A non-transitory computer readable medium embodying at least one set of executable instructions to manipulate at least one processor of a robot agent including a movable physical component to:

• predict at least a first action to be implemented by the robot agent in performing a task assigned to the robot agent;
• predict, at the robot agent, based on at least a set of pre-defined actions, which are actions predicted to fail regardless of other considerations, whether the robot agent will fail or succeed in performing a first action, said predicting including predicting failure when the first action is one of the pre-defined actions, which are actions predicted to fail regardless of other considerations, said step of predicting whether the robot agent will fail or succeed in performing the first action being performed prior to the robot agent attempting to perform said first action;
• and wherein moving a pallet through a zone in which a human is expected to be present is one of the pre-defined actions in the set of pre-defined actions, which are actions predicted to fail regardless of other considerations;
• control the robot agent to determine a mode of operation to be used to implement the first action based on whether the robot agent is predicted to fail or succeed in performing the first action, the robot agent determining to use an unguided mode of operation to perform the first action in response to predicting the robot agent will succeed in performing the first action and determining to use a guided mode of operation in response to predicting the robot agent will fail in performing the first action;
• and
• control the robotic agent to perform the first action using the determined mode of operation.