Examining Audio Communication Mechanisms for Supervising Fleets of Agricultural Robots

TL;DR

This work addresses the challenge of remotely supervising fleets of small agricultural robots under conditions of multitasking and limited technical expertise. It compares three audio-based communication modalities—earcons, single-phrase commands, and full-sentence prompts—against a visual interface using a simulated agribot fleet and a wordsearch secondary task. The study finds that single-phrase notifications are most positively perceived by users, while overall productivity gains are not statistically significant, highlighting the importance of concise, intuitive audio cues for adoption. The results inform human-robot interaction design for scalable agricultural robotics, suggesting that concise verbal prompts can improve perceived usability and reduce cognitive load in multitasking remote supervision, with potential applicability to other domains.

Abstract

Agriculture is facing a labor crisis, leading to increased interest in fleets of small, under-canopy robots (agbots) that can perform precise, targeted actions (e.g., crop scouting, weeding, fertilization), while being supervised by human operators remotely. However, farmers are not necessarily experts in robotics technology and will not adopt technologies that add to their workload or do not provide an immediate payoff. In this work, we explore methods for communication between a remote human operator and multiple agbots and examine the impact of audio communication on the operator's preferences and productivity. We develop a simulation platform where agbots are deployed across a field, randomly encounter failures, and call for help from the operator. As the agbots report errors, various audio communication mechanisms are tested to convey which robot failed and what type of failure occurs. The human is tasked with verbally diagnosing the failure while completing a secondary task. A user study was conducted to test three audio communication methods: earcons, single-phrase commands, and full sentence communication. Each participant completed a survey to determine their preferences and each method's overall effectiveness. Our results suggest that the system using single phrases is the most positively perceived by participants and may allow for the human to complete the secondary task more efficiently. The code is available at: https://github.com/akamboj2/Agbot-Sim.

Figures (3)

• Figure 1: Left: System GUI. The status bar indicates which robots have failed (stopped behind circles). Here the purple robot is behind a blue circle (Unrecoverable Failure) and the blue robot is behind a red circle (Row Collision). The red, green and yellow robots are unobstructed and continue to traverse their sections of the grid. The audio system status shows the system is waiting to hear a color indicating which robot to fix. Right: Experiment setup. The setup mimics the control center setting on an autonomous farm. The user is on an isolated desk in front of a TV screen with the GUI and is given wordsearch puzzles to work on.
• Figure 2: Mean survey responses
• Figure 3: Mean word search score and p-values of paired T-tests