Indabot: A Simulation-Based Study of Low-Cost, ArUco Marker-Based Autonomous Navigation for Indoor Plant Phenotyping

Abstract

Autonomous robotic platforms are of critical importance for high-throughput plant phenotyping in controlled agricultural environments. The vast majority of existing systems are designed for open-field deployment and rely on satellite-based navigation, which is not suitable for enclosed greenhouse settings. This study presents the design, simulation-based performance analysis of Indabot, an autonomous mobile robot developed for wheat phenotyping in indoor environments. Each wheat sample is uniquely tagged with an ArUco marker, enabling drift-free, plant-level localization without the need for any GPS infrastructure. A rotation-based random search navigation scenario was designed in the Webots R2023b environment, in which the OpenCV ArUco module was combined with discrete-time PID controllers and lidar distance feedback. The system achieved a 100% target detection rate across 13 independent simulation runs, with a mean mission completion time of 282.4 ± 6.2 s (n=13, CV=2.2%), a marker centering error of 3.06 ± 0.51 pixels, and an average approach time per target of 19.4 ± 0.4 s. The findings demonstrate that the Indabot platform, operating under ArUco-based guidance, offers a consistent, scalable, and low-cost autonomous phenotyping alternative for indoor agricultural environments. All performance metrics reported in this study were obtained exclusively within the Webots simulation environment; no physical validation of the platform has yet been conducted, and this constitutes a key limitation of the present work.