[Paper Review] Improved Trust in Human-Robot Collaboration with ChatGPT

Yang Ye, Hengxu You+1 more|arXiv (Cornell University)|Apr 25, 2023

Artificial Intelligence in Healthcare and EducationMedicine3 citations

TL;DR

This paper proposes RoboGPT, a human-robot collaboration system that integrates OpenAI's ChatGPT to enable natural language interaction with a 7-DOF robotic arm, significantly improving human trust through enhanced communication. A human-subject experiment demonstrated that ChatGPT-enabled interaction increased trust by enabling more intuitive, context-aware, and responsive robot behavior during tool-fetching and placement tasks.

ABSTRACT

Human robot collaboration is becoming increasingly important as robots become more involved in various aspects of human life in the era of Artificial Intelligence. However, the issue of human operators trust in robots remains a significant concern, primarily due to the lack of adequate semantic understanding and communication between humans and robots. The emergence of Large Language Models (LLMs), such as ChatGPT, provides an opportunity to develop an interactive, communicative, and robust human-robot collaboration approach. This paper explores the impact of ChatGPT on trust in a human-robot collaboration assembly task. This study designs a robot control system called RoboGPT using ChatGPT to control a 7-degree-of-freedom robot arm to help human operators fetch, and place tools, while human operators can communicate with and control the robot arm using natural language. A human-subject experiment showed that incorporating ChatGPT in robots significantly increased trust in human-robot collaboration, which can be attributed to the robot's ability to communicate more effectively with humans. Furthermore, ChatGPT ability to understand the nuances of human language and respond appropriately helps to build a more natural and intuitive human-robot interaction. The findings of this study have significant implications for the development of human-robot collaboration systems.

Motivation & Objective

• To address the challenge of low human trust in human-robot collaboration due to poor semantic understanding and communication.
• To investigate whether large language models like ChatGPT can improve trust through more natural and context-aware human-robot interaction.
• To design and implement a robot control system, RoboGPT, that enables natural language commands for robotic manipulation tasks.
• To evaluate the impact of LLM-powered communication on user trust, perceived reliability, and task performance in a human-robot collaboration setting.
• To demonstrate the feasibility of using LLMs to bridge the communication gap between humans and robots in real-time, task-oriented environments.

Proposed method

• Developed RoboGPT, a control system that integrates OpenAI's ChatGPT as a natural language interface for a 7-degree-of-freedom robotic arm.
• Enabled human operators to issue natural language commands such as 'fetch the screwdriver' or 'place the wrench on the table' to control robot actions.
• Integrated the LLM with a robotic control pipeline that translates natural language into executable robotic motion commands.
• Designed a two-way communication loop where the robot confirms actions using natural language, improving transparency and user awareness.
• Conducted a human-subject experiment with participants interacting with the robot using natural language, measuring trust and performance.
• Used standardized trust assessment scales and task completion metrics to evaluate the impact of ChatGPT integration.

Experimental results

Research questions

• RQ1 How does integrating a large language model like ChatGPT affect human trust in human-robot collaboration?
• RQ2 To what extent does natural language interaction improve the perceived reliability and transparency of robotic systems?
• RQ3 Can LLM-powered robots understand and respond to nuanced, ambiguous, or context-dependent human instructions in real time?
• RQ4 How does the quality of communication between humans and robots influence task performance and user satisfaction?
• RQ5 What role does contextual understanding and response consistency play in building trust during human-robot interaction?

Key findings

• The integration of ChatGPT significantly increased human trust in the robot compared to non-LLM-based systems, as measured by standardized trust assessment scales.
• Participants reported higher perceived reliability and transparency when the robot used natural language to confirm actions and explain decisions.
• The robot’s ability to interpret and respond to ambiguous or context-dependent language improved user confidence and reduced frustration.
• The system demonstrated robust performance in real-time task execution, including fetching and placing tools, using only natural language input.
• Users found the interaction more intuitive and less cognitively demanding when communicating via natural language with the LLM-powered robot.
• The study confirmed that LLMs can serve as effective intermediaries to enhance semantic understanding and communication in human-robot teams.