A robot using a deep learning algorithm to analyse visual information has demonstrated promising results fertilising cabbage grown in strips.
The machine, which features a robotic arm fitted with cameras, low-cost sensors and spraying equipment, was able to make decisions on the amount of liquid organic fertiliser to spray based on the size of plants, when faced with rows of both white and red cabbage of different sizes.
The algorithm behind technology was ‘trained’ based on images obtained during the early growth phase of both types of cabbage, then three weeks later.
In testing using the arm attached to a platform and wheeled along planted rows, detection and characterisation of plants took place in under one second on average. Over the course of the experiments, the robot had a mean efficiency of 90.5% in detecting crops and spraying in real-time.
AI-assisted organics
The robotics experiments are part of an overall body of work called SureVEG, carried out at the Universidad Politécnica de Madrid in Spain. It aims to explore diversified strip cropping systems for intensive cultivation of organic vegetables, managed by automated machinery and incorporating waste reuse.
Priorities for further development of the technology include creating an algorithm that can distinguish between desired crops and weeds, the researchers said. This would enable the use of a valve on the robotic arm, to dispense either fertiliser or herbicide depending on what was present.
You can read the full study report in Computers and Electronics in Agriculture.
