A NEW METHODOLOGY FOR DEALING WITH UNCERTAINTY IN ROBOTIC TASKS

摘要

A new method that evaluates the execution of robotic operations at the task level is proposed. Traditional robot controllers use variety of feedback loops at motion level to solve dynamical control and motion level planning problems asssuming that the original task plan is never modified. However, a drastical change in the operation environment of the robot might as well affect the sequence of tasks to be executed. In such a case, modifying motion plan may not be sufficient to adapt to the unexpected change in the environment. In this study, a planning environment that allows the original task plan to be modified or entirely changed according to the changing environmental conditions is proposed. A heuristic approach is taken to decide on whether the original plan should be modified or an entirely new plan should be generated. A simple case on pick-and-place sequencing of blocks-world is studied to demonstrate the idea.