Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Large Language Models Enable Automated Formative Feedback in Human-Robot Interaction Tasks

Emily Jensen, Sriram Sankaranarayanan, Bradley Hayes

TL;DR

The paper tackles scalable, accessible feedback generation for human-robot interaction tasks by pairing large language models with formal task specifications such as signal_temporal_logic and robustness. It argues that LLMs can translate formal assessments into formative feedback templates that are easy for non-experts to understand, enabling targeted learner improvements within existing curricula. The authors outline a pipeline, discuss potential impacts, and address challenges related to safety, knowledge representation, and integration with training systems, while calling for action and future research. The work highlights practical significance in enabling scalable, personalized HRI training and reducing instructor burden, with implications for shaping future training pipelines and learning theory in robotics.

Abstract

We claim that LLMs can be paired with formal analysis methods to provide accessible, relevant feedback for HRI tasks. While logic specifications are useful for defining and assessing a task, these representations are not easily interpreted by non-experts. Luckily, LLMs are adept at generating easy-to-understand text that explains difficult concepts. By integrating task assessment outcomes and other contextual information into an LLM prompt, we can effectively synthesize a useful set of recommendations for the learner to improve their performance.

Large Language Models Enable Automated Formative Feedback in Human-Robot Interaction Tasks

TL;DR

The paper tackles scalable, accessible feedback generation for human-robot interaction tasks by pairing large language models with formal task specifications such as signal_temporal_logic and robustness. It argues that LLMs can translate formal assessments into formative feedback templates that are easy for non-experts to understand, enabling targeted learner improvements within existing curricula. The authors outline a pipeline, discuss potential impacts, and address challenges related to safety, knowledge representation, and integration with training systems, while calling for action and future research. The work highlights practical significance in enabling scalable, personalized HRI training and reducing instructor burden, with implications for shaping future training pipelines and learning theory in robotics.

Abstract

We claim that LLMs can be paired with formal analysis methods to provide accessible, relevant feedback for HRI tasks. While logic specifications are useful for defining and assessing a task, these representations are not easily interpreted by non-experts. Luckily, LLMs are adept at generating easy-to-understand text that explains difficult concepts. By integrating task assessment outcomes and other contextual information into an LLM prompt, we can effectively synthesize a useful set of recommendations for the learner to improve their performance.
Paper Structure (6 sections, 1 figure)

This paper contains 6 sections, 1 figure.

Table of Contents

  1. Introduction
  2. Position Statement
  3. Challenges and Opportunities
  4. Potential Impact
  5. Challenges
  6. Call to Action

Figures (1)

  • Figure 1: Example of three trajectories in a 2D drone landing task. The pilot must keep the drone within the black boundary and safely reach the gray landing pad with a velocity $\leq 5~m/s$ and attitude $\leq 10^{\circ}$.