Hierarchical geometric plan composition (HGPC) framework for robot task planning

A method for controlling a robot in an environment comprising:

• representing, by a processor, a task as a meta-node having a meta start space representing starting configurations of the robot and the environment and a meta end space representing ending configurations of the robot and the environment after the task is completed, the meta-node being at least one of:
• a parallel meta-node including a first sub-node and a second sub-node each having a parallel start space within the meta start space and a parallel end space within the meta end space such that the meta end space is reached by successful execution of at least one of the first sub-node or the second sub-node,
• a repetition meta-node including a third sub-node having a repetition start space within the meta start space and a repetition end space outside of the meta end space such that the meta end space is reached by multiple executions of the third sub-node, or
• a sequence meta-node including a fourth sub-node having a first sequence start space in the meta start space and a first sequence stop space and a fifth sub-node having a second sequence start space in the first sequence stop space of the fourth sub-node and a second sequence stop space in the meta stop space;
• performing, by the processor, a simulation of the task by executing the parallel meta-node, the repetition meta-node or the sequence meta-node;
• and
• instructing, by the processor, the at least one actuator or motor to actuate at least a portion of the robot based on the simulation.

A method for controlling a robot in an environment comprising:

• receiving, at a processor, a definition of a meta-node representing a task and having a meta start space representing starting configurations of the robot and the environment and a meta end space representing ending configurations of the robot and the environment after the task is completed, the meta-node being at least one of:
• a parallel meta-node including a first sub-node and a second sub-node each having a parallel start space within the meta start space and a parallel end space within the meta end space such that the meta end space is reached by successful execution of at least one of the first sub-node or the second sub-node, or
• a repetition meta-node including a third sub-node having a repetition start space within the meta start space and a repetition end space outside of the meta end space such that the meta end space is reached by multiple executions of the third sub-node;
• performing, by the processor, a simulation of the task by executing the parallel meta-node or the repetition meta-node;
• and
• instructing, by the processor, the at least one actuator or motor to actuate at least a portion of the robot based on the simulation.