Designing for Error Recovery in Human-Robot Interaction

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Wallbridge, Christopher D., Pulgarin, Erwin Jose Lopez
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866918445892763648
author Wallbridge, Christopher D.
Pulgarin, Erwin Jose Lopez
author_facet Wallbridge, Christopher D.
Pulgarin, Erwin Jose Lopez
contents This position paper looks briefly at the way we attempt to program robotic AI systems. Many AI systems are based on the idea of trying to improve the performance of one individual system to beyond so-called human baselines. However, these systems often look at one shot and one-way decisions, whereas the real world is more continuous and interactive. Humans, however, are often able to recover from and learn from errors - enabling a much higher rate of success. We look at the challenges of building a system that can detect/recover from its own errors, using the example of robotic nuclear gloveboxes as a use case to help illustrate examples. We then go on to talk about simple starting designs.
format Preprint
id arxiv_https___arxiv_org_abs_2604_12473
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Designing for Error Recovery in Human-Robot Interaction
Wallbridge, Christopher D.
Pulgarin, Erwin Jose Lopez
Robotics
Human-Computer Interaction
This position paper looks briefly at the way we attempt to program robotic AI systems. Many AI systems are based on the idea of trying to improve the performance of one individual system to beyond so-called human baselines. However, these systems often look at one shot and one-way decisions, whereas the real world is more continuous and interactive. Humans, however, are often able to recover from and learn from errors - enabling a much higher rate of success. We look at the challenges of building a system that can detect/recover from its own errors, using the example of robotic nuclear gloveboxes as a use case to help illustrate examples. We then go on to talk about simple starting designs.
title Designing for Error Recovery in Human-Robot Interaction
topic Robotics
Human-Computer Interaction
url https://arxiv.org/abs/2604.12473