Core Concepts
Balancing user control with robot autonomy is crucial in goal-oriented end-user programming tools.
Abstract
The content discusses the development of a novel end-user programming system, Polaris, that uses goal predicates as the fundamental building block of programs. Users can express high-level robot objectives or lower-level checkpoints while an off-the-shelf task planner fills in program details. The Plan Visualizer provides feedback to ensure user expectations align with robot behavior. Evaluation results show improved plan quality and user experience.
Abstract:
Goal-oriented end-user programming tools balance user control and robot autonomy.
Polaris uses goal predicates as the fundamental building block of programs.
The Plan Visualizer provides feedback on program details before runtime.
Introduction:
Demand for efficient approaches for users to specify tasks for robots is growing.
End-user programming tools enable users to create and customize robot applications.
Polaris aims to provide flexibility in program specification while leveraging robot autonomy.
Related Work:
Existing EUP tools focus on capturing user intent through various programming environments.
Polaris challenges the action-oriented paradigm by using goal predicates as the building block of programs.
System Design:
Goal automata are transition systems guiding robots in achieving goals during tasks.
Branching plans are created based on goal automata, ensuring actions match developer intent.
The Plan Visualizer allows users to view and interact with the plan in real-time.
System Evaluation:
An experiment compared participants using the Plan Visualizer versus those who did not.
Results showed improved plan quality and marginal effects on satisfaction, usefulness, and perceived competence with the Plan Visualizer.
Stats
Users can express high-level objectives or lower-level checkpoints - Users can leverage robot autonomy - Participants evaluated with 32 human participants
Quotes
"Feedback is critical for goal-oriented EUP."
"Goal-oriented programming allows for greater flexibility but may be challenging in practice."