Autonomous Robot for Small-Scale NFT Systems

Abstract

Hydroponics is currently a very successful technique for growing crops like lettuce and results in cleaner, healthier and cheaper yield. Requirement of periodic labor and a systematic arrangement of plants make it a good candidate for automation. Large-scale systems that automate the growing of lettuce already exist, but these require large capital investments and large footprints that limit the use to only large-scale growers. Our goal is to develop inexpensive robotic systems that a small-medium scale grower can afford to implement and that are compatible with the existing infrastructure. The robot is designed to pick and place plants and seedlings from one spot to other on an Nutrient Film Technique (NFT) system autonomously. We describe the design and construction of the gantry-based robot with an arm that moves on rails. By using this approach a single robot can serve a large number of plants in a greenhouse. We also describe algorithms for perception, robot path planning, and manipulation of densely planted leafy crops using a camera and Microsoft Kinect 3D imaging system. Results are demonstrated experimentally using a small AmHydro NFT system we have constructed in our lab.