[Paper Review] Robots in the Garden: Artificial Intelligence and Adaptive Landscapes

Zihao Zhang, Susan L. Epstein+4 more|arXiv (Cornell University)|May 22, 2023

Land Use and Ecosystem ServicesEnvironmental Science3 citations

TL;DR

This paper presents ELUA, an AI-integrated urban agricultural lab featuring two gantry robots that autonomously perform seeding, watering, weeding, and pruning in indoor and rooftop gardens. Equipped with sensors, computer vision, and AI-driven control systems, ELUA enables responsive, adaptive landscape design through real-time environmental feedback, demonstrating a framework for long-term, AI-augmented ecological engagement in urban settings.

ABSTRACT

This paper introduces ELUA, the Ecological Laboratory for Urban Agriculture, a collaboration among landscape architects, architects and computer scientists who specialize in artificial intelligence, robotics and computer vision. ELUA has two gantry robots, one indoors and the other outside on the rooftop of a 6-story campus building. Each robot can seed, water, weed, and prune in its garden. To support responsive landscape research, ELUA also includes sensor arrays, an AI-powered camera, and an extensive network infrastructure. This project demonstrates a way to integrate artificial intelligence into an evolving urban ecosystem, and encourages landscape architects to develop an adaptive design framework where design becomes a long-term engagement with the environment.

Motivation & Objective

• To develop a responsive, AI-augmented urban agricultural ecosystem that supports long-term ecological engagement.
• To integrate robotics and artificial intelligence into landscape architecture for adaptive, evolving design processes.
• To create a testbed for real-time environmental sensing and autonomous intervention in urban green spaces.
• To bridge disciplines across landscape architecture, computer science, and AI for sustainable urban development.

Proposed method

• Deployment of two gantry-mounted robots—one indoors and one on a rooftop—capable of seeding, watering, weeding, and pruning.
• Integration of sensor arrays to monitor microclimate, soil moisture, and plant health in real time.
• Implementation of an AI-powered camera system for continuous visual monitoring and plant state recognition.
• Use of a comprehensive network infrastructure to enable data synchronization and remote control across the robotic system.
• Application of computer vision and AI algorithms to interpret sensor and visual data for adaptive decision-making.
• Design of a responsive landscape framework where robotic actions are dynamically adjusted based on environmental feedback loops.

Experimental results

Research questions

• RQ1 How can AI and robotics be effectively integrated into urban agricultural landscapes to enable adaptive, real-time responses to environmental changes?
• RQ2 What role can robotic systems play in supporting long-term, evolving landscape design beyond static planning?
• RQ3 How can sensor networks and computer vision enable autonomous monitoring and intervention in urban green spaces?
• RQ4 What interdisciplinary frameworks are needed to support AI-driven, ecologically responsive landscape architecture?

Key findings

• The ELUA system successfully demonstrated autonomous execution of multiple gardening tasks across both indoor and outdoor environments.
• Real-time environmental data from sensors and computer vision enabled dynamic adjustments in robotic behavior.
• The integration of AI and robotics allowed for continuous monitoring and adaptive responses to plant growth and microclimate variations.
• The project established a functional model for responsive landscape architecture, where design is an ongoing, data-informed process.
• The interdisciplinary collaboration between landscape architects, roboticists, and AI researchers proved essential for developing a holistic, adaptive system.
• The system's infrastructure supports scalable deployment of similar robotic solutions in urban ecological settings.