ASABE Technical Library - Abstract

Click on “Download PDF” for the PDF version or on the title for the HTML version.
If you are not an ASABE member or if your employer has not arranged for access to the full-text, view access and subscription options

GrowBot: System Design and Architecture for an AI-Optimized Controlled-Environment Plant Growth Enclosure

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org

Citation:  2026 ASABE Annual International Meeting  2600750.(doi:10.13031/aim.202600750)
Authors:   Roland Kincaid, Shekhar Suman Borah, Chandrasekar Kousik, Prabha Sundaravadivel
Keywords:   Controlled-environment agriculture (CEA), edge AI, embedded vision, Hailo-8L, Raspberry Pi 5, ESP32, YOLOv8, plant monitoring, smart agriculture, automated environmental control.

Abstract. Controlled-environment agriculture (CEA) has emerged as an important approach for improving crop productivity, resource efficiency, and year-round food production under changing environmental conditions. Recent advances in embedded artificial intelligence (AI) and edge computing enable real-time plant monitoring and autonomous environmental regulation using compact, low-power hardware platforms. This paper presents GrowBot, an AI-enabled controlled-environment plant growth enclosure integrating dual-camera machine vision, embedded edge inference, automated environmental sensing, and closed-loop actuator control. The system combines a Raspberry Pi 5 single-board computer with a Hailo-8L neural processing unit (NPU) for accelerated real-time inference and an ESP32-WROOM-32 microcontroller for environmental sensing and actuator management. A Sony IMX500 intelligent vision sensor continuously monitors the enclosure canopy and identifies regions of interest using onboard inference, while a secondary IMX219 camera mounted on a three-axis brushless gimbal performs detailed localized inspection and plant health analysis using a YOLOv8n detection framework. Environmental regulation includes temperature, humidity, carbon dioxide concentration, soil moisture, lighting intensity, airflow, and mist generation through relay-controlled and pulse-width modulation (PWM)-controlled subsystems. The enclosure is powered using a multi-rail 450 W ATX power distribution architecture supporting 12 V, 5 V, 3.3 V, and 2.5 V subsystems. Experimental implementation demonstrated stable real-time operation at approximately 28 frames s-1 with low-latency inference and coordinated environmental response. The proposed system provides a modular and scalable platform for AI-assisted plant monitoring, precision indoor agriculture, and future autonomous crop management research.

(Download PDF)    (Export to EndNotes)