Human-mediated Large Language Models for Robotic Intervention in Children with Autism Spectrum Disorders

TL;DR

This paper enhances robotic intervention for children with ASD by integrating large language models to enable autonomous, role-based social interaction teaching, validated through expert simulations and surveys.

Contribution

It introduces a novel LLM-based pipeline for autonomous perspective-taking teaching in robots, moving beyond pre-defined scripts for ASD therapy.

Findings

GPT-2 + BART pipeline outperforms GPT-2 alone in social situation generation.

Expert simulations show the robot is perceived as safe, likable, and reliable.

No increase in perceived effort or frustration during robot-assisted sessions.

Abstract

The robotic intervention for individuals with Autism Spectrum Disorder (ASD) has generally used pre-defined scripts to deliver verbal content during one-to-one therapy sessions. This practice restricts the use of robots to limited, pre-mediated instructional curricula. In this paper, we increase robot autonomy in one such robotic intervention for children with ASD by implementing perspective-taking teaching. Our approach uses large language models (LLM) to generate verbal content as texts and then deliver it to the child via robotic speech. In the proposed pipeline, we teach perspective-taking through which our robot takes up three roles: initiator, prompter, and reinforcer. We adopted the GPT-2 + BART pipelines to generate social situations, ask questions (as initiator), and give options (as prompter) when required. The robot encourages the child by giving positive reinforcement for correct answers (as a reinforcer). In addition to our technical contribution, we conducted ten-minute sessions with domain experts simulating an actual perspective teaching session, with the researcher acting as a child participant. These sessions validated our robotic intervention pipeline through surveys, including those from NASA TLX and GodSpeed. We used BERTScore to compare our GPT-2 + BART pipeline with an all GPT-2 and found the performance of the former to be better. Based on the responses by the domain experts, the robot session demonstrated higher performance with no additional increase in mental or physical demand, temporal demand, effort, or frustration compared to a no-robot session. We also concluded that the domain experts perceived the robot as ideally safe, likable, and reliable.