A Collaborative Framework for Robotic Task Specification

Abstract

Implementation of automated robotic solutions for complex tasks currently faces a few major hurdles. For instance, lack of effective sensing and task variability – especially in high-mix, low-volume processes – creates too much uncertainty to reliably hard-code a robotic work cell. Current collaborative frameworks generally focus on integrating the sensing required for a physically collaborative implementation. While this paradigm has proven effective for mitigating uncertainty by mixing human cognitive function and fine motor skills with robotic strength and repeatability, there are many instances where physical interaction is impractical, as human reasoning and task knowledge are still needed. The proposed framework consists of key modules such as a path planner, path simulator, and result simulator. An integrated user interface facilitates the operator to interact with these modules and edit the path plan before ultimately approving the task for automatic execution by a manipulator that need not be collaborative. The software modules are integrated using the Robotic Operating System. Application of the collaborative framework is illustrated for a pressure washing task in a remanufacturing environment that requires one-off path planning for each part. The framework can also be applied to various other tasks, such as spray-painting, sandblasting, deburring, grinding, and shot peening.