Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Designing for Fairness in Human-Robot Interactions

Houston Claure

TL;DR

This paper tackles how autonomous robots should participate in multi-human teams with continuous resource allocation to uphold fairness and group harmony. It proposes two data-capture platforms, Co-Tetris and Multiplayer Space Invaders, to study group dynamics and fairness in HRI, and develops the strict-rate-constrained UCB algorithm to learn teammate skill while guaranteeing a minimum share for each member. Theoretical guarantees on regret accompany empirical validation with 290 participants, showing that fair resource distribution enhances user trust in robots. The work outlines future directions for fairness benchmarks, real-time adaptation, and broader theory incorporating explainability and cultural factors to improve real-world HRI outcomes.

Abstract

The foundation of successful human collaboration is deeply rooted in the principles of fairness. As robots are increasingly prevalent in various parts of society where they are working alongside groups and teams of humans, their ability to understand and act according to principles of fairness becomes crucial for their effective integration. This is especially critical when robots are part of multi-human teams, where they must make continuous decisions regarding the allocation of resources. These resources can be material, such as tools, or communicative, such as gaze direction, and must be distributed fairly among team members to ensure optimal team performance and healthy group dynamics. Therefore, our research focuses on understanding how robots can effectively participate within human groups by making fair decisions while contributing positively to group dynamics and outcomes. In this paper, I discuss advances toward ensuring that robots are capable of considering human notions of fairness in their decision-making.

Designing for Fairness in Human-Robot Interactions

TL;DR

This paper tackles how autonomous robots should participate in multi-human teams with continuous resource allocation to uphold fairness and group harmony. It proposes two data-capture platforms, Co-Tetris and Multiplayer Space Invaders, to study group dynamics and fairness in HRI, and develops the strict-rate-constrained UCB algorithm to learn teammate skill while guaranteeing a minimum share for each member. Theoretical guarantees on regret accompany empirical validation with 290 participants, showing that fair resource distribution enhances user trust in robots. The work outlines future directions for fairness benchmarks, real-time adaptation, and broader theory incorporating explainability and cultural factors to improve real-world HRI outcomes.

Abstract

The foundation of successful human collaboration is deeply rooted in the principles of fairness. As robots are increasingly prevalent in various parts of society where they are working alongside groups and teams of humans, their ability to understand and act according to principles of fairness becomes crucial for their effective integration. This is especially critical when robots are part of multi-human teams, where they must make continuous decisions regarding the allocation of resources. These resources can be material, such as tools, or communicative, such as gaze direction, and must be distributed fairly among team members to ensure optimal team performance and healthy group dynamics. Therefore, our research focuses on understanding how robots can effectively participate within human groups by making fair decisions while contributing positively to group dynamics and outcomes. In this paper, I discuss advances toward ensuring that robots are capable of considering human notions of fairness in their decision-making.
Paper Structure (8 sections, 2 figures)

This paper contains 8 sections, 2 figures.

Table of Contents

  1. Introduction & Motivation
  2. Prior Work
  3. Building platforms to capture data from multi-agent groups
  4. Co-Tetris for Online Experiments
  5. Multiplayer Space Invaders for Laboratory Experiments
  6. Building Theory on How Unfair Robot Actions Shape Human Group Dynamics
  7. Designing Algorithms that Consider Fairness
  8. Future Work

Figures (2)

  • Figure 1: Consider a scenario where a robot has to make decisions on how to distribute resources across two team members. However, one team member is more skilled than the other. How should a robot take skill level into account when dividing its resources to maximize both efficiency and also fairness? This example encapsulates the nuanced responsibility that robots have in human teams: to enhance productivity while fostering an environment of trust and equality.
  • Figure 2: Co-Tetris: a platform for observing group behavior in the face of unfair AI decisions. I ran an experiment where one player was presented as an intelligent system and disproportionately allocated falling Tetris blocks to one player (left). Multiplayer Space Invaders: a robot and a human participant competed to eliminate as many enemies on the screen as possible. A third robot disproportionately supported one player by helping them eliminate enemies for longer periods of time.