A robotic solution takes hand for the banana industry

QUT’s innovative robotic arm project could revolutionise banana farming through the automation of de-handing, increasing efficiency and promising a sustainable future in agriculture innovation.

In a collaboration between Hort Innovation, Queensland University of Technology (QUT), Future Food Systems, Advanced Robotics for Manufacturing Hub (ARM), and BNL Industrial Solutions, a $2 million program is underway to transform the landscape of banana farming.

The primary focus is to prototype a robot arm that will automate a process called banana de-handing.

The scientists and researchers involved will be using technologies such as computer vison and machine learning to provide the banana industry with a solution for the repetitive and intensive banana de-handing processes.

Food and Beverage Industry News spoke with QUT lead researcher, Dr Chris Lehnert, to understand the need for this robot prototype, what is involved in the prototyping process and how a robotic solution could provide significant advancements to the industry.

Lehnert, with a background in robotics research, in recent years, has shifted his focus to vertical farming, envisioning the integration of robotics across a spectrum of harvesting processes within the food industry.

Approaching the farming challenges

At the heart of this prototype development is the Australian Banana Growers’ Council (ABGC), an organisation leading and represent banana growers, with the aim of ensuring a profitable and resilient industry future.

Across the country the group runs several grower-focused projects and advocates on behalf of banana farmers.

Recognising the need for technological innovation, the ABGC approached QUT to explore opportunities in artificial intelligence (AI) and automation.

QUT started the research process completing a feasibility study, to understand what banana farmers were struggling with, and what the best opportunities for robotic aid could be.

Through the feasibility study, researchers at QUT engaged directly with banana farmers in North Queensland.

Through these interactions, the team identified the challenging task of de-handing bananas as a key pain point within the industry.

This physically demanding and repetitive process became the focal point for a research project.

“This problem of de-handing came up, where the process is a labour-intensive task, it’s very hard to train people to do the task, and it can be quite dangerous,” said Lehnert.

Banana de-handing is the process of separating the banana fruit from the stalk, if automated, could improve processing efficiencies and provide a valuable integration point for robotics systems.

The feedback from farmers regarding their perception on utilising a robotic solution was overwhelmingly positive.

Lehnert explain from their feasibility study, that finding’s showed famers are eager to utilise technology in their sheds.

However, they are concerned about the reliability of robotics, raising questions regarding life cycle lengths of the robot, its consistency, servicing and maintenance.

“If it does the task, they will 100 per cent adopt it,” said Lehnert.

The iterative path to robotic success

The research journey at QUT is following a path of innovation through iterative design.

“The whole process that we’re going through is very iterative,” said Lehnert.

To develop an effective robotic solution, the QUT research team is focusing on two crucial systems, relating to vision and action.

Lehnert explained further, “one is going to be the vision system; this is the part that must identify where to make a cut. And the second is the part that moves the tool to cut the hand off the bunch.”

Each of the two systems will be research separately and then combined to make a successful solution for the banana industry.

Integral to their approach is the constant integration of industry feedback, involving growers, shed managers, and individuals overseeing day-to-day operations.

“We are going to get growers, shed managers and people who are running the day-to-day business in the sheds,” said Lehnert.

Previous attempts at aiding banana farmers with de-handing focused on technology through a mechanical system, this lacked adaptability to real-time situations.

“None of that’s been successful and that’s because it hasn’t been able to look at what the fruit is doing, and then adapt to what the fruit has done in real time,” said Lehnert.

Moving from prototype to commercial success

The potential success of this robotic prototype could change banana farming processes, making it more efficient, sustainable and resilient.

When the research team get to a proof of concept they will not move to the commercialising phase until the design meets farmer requirements and reliability.

Sharing his insights on the process of moving to commercialisation, Lehnert said, “We’ve already got one industry partner who’s really interested in commercialising the technology.”

The final stage would involve attracting investors who are invested in the commercial success of this transformative technology.

“There’s a lot of opportunities at the end of the project to look at getting investors, because there are investors that have a vested interest in aiding the commercial success of the technology,” he added.

Handlining the impacts

As the robotic landscape continues to evolve and intersect a range of industries, alongside the growing social pressures of creating more sustainable practices, a robotic solution for banana de-handing could be monumental.

For the banana industry, during the processing line, the point where the most waste occurs is at the de-handing system, which is the first point where the fruit is touched.

A robotic solution has the potential to significantly aid in the sustainability sector, specifically reducing product waste.

“Bananas are really sensitive to touch, and a lot of bananas get damaged at the first point of touch resulting in lot of waste,” said Lehnert.

With contingent food handling requirements, the team is designing the robot to meet and maintain industry standards.

“Anything we design and build on the tool side will be food graded, and we are also looking at a protective covering for the robot that is also food safe,” said Lehnert.

Scalability for the industry

With bananas having various species, varieties and ripeness levels, the QUT robot is focused on the de-handing process of Cavendish bananas.

This focus choice is driven by Cavendish banana prevalence as one of the most grown banana varieties.

The potential impact on efficiency could be significant, not only as an assistive device but potentially as a standalone system.

“It doesn’t have to be just replacing workers. It can assist in scaling up and meeting the processing requirements of the day,” said Lehnert.

Lehnert envisions the robotic solution addressing industry challenges of employee and skill shortages, with the possibility of this solution being a system that can run 24/7 without any assistance.

“The growers at this stage are actually struggling to get to get workers to meet the demand of how much output they need to process,” said Lehnert.

The agriculture industry is taking two approaches to robotic solutions, the first is providing robotics as a service and the second is providing robotics as a product.

The QUT prototype is expected to sit somewhere between this spectrum, given the early design stages it is not yet clear on which approach this solution will take.

“It does depend on the outcome of the prototype.” Lehnert said.

As the research lead for this prototype, Lehnert aims to address unique industry challenges, offering innovative solutions while providing the industry a robot that offers a substantial return on investment.

The collaborative efforts between Hort Innovation, QUT, and industry partners in developing a robotic solution for banana de-handing signifies a technological leap for the agricultural sector and holds the promise of reshaping the future of banana de-handing, addressing longstanding challenges within the agricultural landscape.

Send this to a friend