Robotic Vertical Farming Systems

Why it Matters?

The global population is estimated to reach 10 billion by 2050 and there is a growing demand for agricultural systems that can grow more food or be more efficient to feed a larger population with the same arable land size [1]. The world’s farmland is becoming increasingly unsuitable for production, with 25% of all farmland already rated as highly degraded. Water resources are highly stressed, with more than 40% of the world’s rural population living in water-scarce areas. Australia’s horticulture industry has an opportunity to establish high-technology horticulture in or close to urban centres, close to consumers and a younger workforce.

The protected cropping industry is the fastest-growing food-producing industry in Australia. It is valued at $1.5 billion farm-gate value per annum that is equivalent to 15% of the industry [2], [3]. Development of a robotic crop management system would benefit the horticulture industry in Australia, the international horticulture industry, and its end-users or consumers. The current horticulture industry is facing challenges with the high cost of crop production requiring a large amount of labour. By developing robotic crop management systems, which can support the crop production systems in current protected cropping systems, the agricultural industry will benefit economically. Additionally, the agricultural industry could also benefit from the reduction of risks due to weather, pests and diseases or food safety from contamination.

Project overview

This project aims to develop a robotic crop management system for the growth of nutrient-dense crops within Vertical Farming Systems (VFS), which would make them more sustainable and cost-effective by increasing their productivity, reducing labour costs and minimise risks related to food security, safety and waste. The robotic system aims to complement current hydroponic growing techniques to become a complete crop management system which can perform all critical aspects of crop growth including seeding, planting, pruning, nutrient application, health monitoring, harvesting, cleaning and servicing of the growth system. The system will aim to optimise the yield and quality of the crop given environment and input resource constraints (nutrients and water).

The objective of this project is to research and develop key concepts to progress towards a complete robotic crop management system for VFS. A robotic crop management system would have the ability to:

  • actively perceive the environment and estimate the state of the crop, monitoring for risks such as pests and diseases and support a decision-making system while operating within highly occluded and unstructured natural environments;
  • an optimal sequential decision-making system which utilises all available crop and environmental information to choose the next action that optimises yield and quality of the crop while accounting for uncertainties and constraints; and
  • a crop manipulation system that can perform different crop actions such as planting, pruning, harvesting and cleaning within the growing system.

Some of the questions this project is aiming to address include:

  • Which computer vision and machine learning-based methods are more suitable for classifying crop properties such as growth stages within a complex occluded and cluttered environment posed within VFS or protected cropping systems?
  • How can we utilise crop growth models, sensor measurements and optimal control techniques to select the next action to take which maximises an objective such as crop yield, health or profit?
  • How can a robot robustly move towards and remove/harvest a crop from its growing module within a cluttered and occluded environment created by a VFS?

Real world impact

The use of robots in protected cropping system has multiple potential benefits including increasing production efficiency, reducing cost and improving labour quality and safety [4]. Automating tasks in protected cropping industry has several advantages [5]. Firstly, robots can perform repeated tasks continuously, which is mandatory to correct environmental monitoring, climate control and failure detection. Operations like precise fertilization and spraying of every single plant, precise inspection and growth evaluation of every single plant can be controlled by autonomous systems. Additionally, these systems can reduce the exposure of human operators to hazardous environments. Finally, they can also provide both high quantity and quality of information, which allows new possibilities, such as local climate control and accurate product traceability.


Other Team Members

Students:

  • Callum Veraa
  • Cameron Fox
  • Nathan Jardine
  • Joel John
  • Aaron Nebauer
  • James Snewin

Partners:

CAB FFS-CRC Greenb