Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Characterizing Input Methods for Human-to-robot Demonstrations

Pragathi Praveena, Guru Subramani, Bilge Mutlu, Michael Gleicher

TL;DR

The paper addresses the lack of a unifying framework for selecting input methods in human-to-robot demonstrations by introducing a design-space framework built from usage paradigms and a 12-feature set. It then operationalizes this framework through the instrumented tongs, a novel input device designed to yield high-quality, easily instrumented demonstrations for hand-scale object manipulation with two-finger grippers, and validates it with a within-subject user study against free-hand manipulation, kinesthetic guidance, and teleoperation. The study shows that instrumented tongs provide motion-quality metrics comparable to hand, improve demonstrator experience relative to teleoperation and kinesthetic methods, and offer precise measurement capabilities via integrated sensors. The work contributes a practical design-space tool for evaluating input methods, a concrete instrumented device, and empirical evidence guiding the design of demonstration interfaces with novice users. Overall, it advances how researchers choose and design input methods for robot learning and programming, with implications for improving usability, accuracy, and transfer to real-world tasks.

Abstract

Human demonstrations are important in a range of robotics applications, and are created with a variety of input methods. However, the design space for these input methods has not been extensively studied. In this paper, focusing on demonstrations of hand-scale object manipulation tasks to robot arms with two-finger grippers, we identify distinct usage paradigms in robotics that utilize human-to-robot demonstrations, extract abstract features that form a design space for input methods, and characterize existing input methods as well as a novel input method that we introduce, the instrumented tongs. We detail the design specifications for our method and present a user study that compares it against three common input methods: free-hand manipulation, kinesthetic guidance, and teleoperation. Study results show that instrumented tongs provide high quality demonstrations and a positive experience for the demonstrator while offering good correspondence to the target robot.

Characterizing Input Methods for Human-to-robot Demonstrations

TL;DR

The paper addresses the lack of a unifying framework for selecting input methods in human-to-robot demonstrations by introducing a design-space framework built from usage paradigms and a 12-feature set. It then operationalizes this framework through the instrumented tongs, a novel input device designed to yield high-quality, easily instrumented demonstrations for hand-scale object manipulation with two-finger grippers, and validates it with a within-subject user study against free-hand manipulation, kinesthetic guidance, and teleoperation. The study shows that instrumented tongs provide motion-quality metrics comparable to hand, improve demonstrator experience relative to teleoperation and kinesthetic methods, and offer precise measurement capabilities via integrated sensors. The work contributes a practical design-space tool for evaluating input methods, a concrete instrumented device, and empirical evidence guiding the design of demonstration interfaces with novice users. Overall, it advances how researchers choose and design input methods for robot learning and programming, with implications for improving usability, accuracy, and transfer to real-world tasks.

Abstract

Human demonstrations are important in a range of robotics applications, and are created with a variety of input methods. However, the design space for these input methods has not been extensively studied. In this paper, focusing on demonstrations of hand-scale object manipulation tasks to robot arms with two-finger grippers, we identify distinct usage paradigms in robotics that utilize human-to-robot demonstrations, extract abstract features that form a design space for input methods, and characterize existing input methods as well as a novel input method that we introduce, the instrumented tongs. We detail the design specifications for our method and present a user study that compares it against three common input methods: free-hand manipulation, kinesthetic guidance, and teleoperation. Study results show that instrumented tongs provide high quality demonstrations and a positive experience for the demonstrator while offering good correspondence to the target robot.

Paper Structure

This paper contains 46 sections, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. A Framework for Characterizing Input MethodsDESIGN SPACE
  4. Usage paradigmsUse Cases
  5. Direct Replay
  6. Real-time Control
  7. One-shot Learning
  8. Learning from Demonstration (LfD)
  9. Incremental Learning from Demonstration (ILfD)
  10. Apprenticeship Learning (AL)
  11. Design
  12. Features
  13. Efficient
  14. Subjective performance
  15. Desired demonstrations
  16. ...and 31 more sections

Figures (4)

  • Figure 1: Our framework and its three components: usage paradigms, design space of features, and characterization of input methods.
  • Figure 2: Design of the current prototype of instrumented tongs.
  • Figure 3: Using instrumented tongs to demonstrate Lego block stacking.
  • Figure 4: Boxplots for (A) accuracy, (B) path length, (C) jerk, (D) interaction time, and (E) subjective measures of ease of use, enjoyment and confidence for the four methods of demonstration. (F) Point plot of interaction time during the training task. Vertical bars show standard deviation. (G) Boxplots of the perceived workload on each subscale of the NASA Task Load Index.