Condobolin Precision Ag day: Veggie robot a real RIPPA

Bots rolling around the farm 24/7 weeding, providing crop intelligence and soil sampling is closer than you think.

These autonomous robots will be compatible with other robotic systems on the farm, and will offer yield predictions to help growers drive production efficiencies.

One such system is the RIPPA (Robotic Intelligent Perception and Precision Application), a prototype robot designed specifically for the vegetable industry.

It recently made an appearance at the Central West Farming Systems Precision Agriculture seminar held at Condobolin, NSW, on June 28.

The brains behind the innovative system comprise University of Sydney's Professor of Robotics and Intelligent Systems, Salah Sukkarieh, and a team of engineers at the university's Australian Centre for Field Robotics (ACFR).

The centre is one of the largest robotic groups in the world and has focused on the autonomous cattle monitoring platform, SwagBot, horticultural robots Shrimp, Ladybird and RIPPA, and a drone for detecting invasive aquatic species.

RIPPA is an evolved and refined version of the ACFR's Ladybird unit – a solar-powered electric bot without a drivetrain that delivers configuration flexibility.

"We can open the wings for different-height crops and change the platform width for different rows, and it has a sensor pod and robotic arm underneath," Sukkarieh says.

"The robot has a seven-hour battery life – using solar panels on a clear day we get about 12 hours of endurance out of the platform.

"It's an omni-directional platform which drives up and down the headland with a GPS; we tell it the row width and then let it go. There is a large collection of hyperspectral, infra-red, thermal and vision sensors and lasers underneath the wings to collate high-resolution data from the crop.

"The robot takes in the data, classifies it and learns to detect differences between the crop and weeds. The more data it receives, the better it gets at picking up those differences.

"The vegetable industry wanted to take it from a research project to a prototype so RIPPA was born."

RIPPA is three-quarters of the weight of Ladybird and each of the four wheels has its own drive train.

The solar-powered robot has 24-hour capability depending on the weather, and travels across the paddock at 100 to 150cm per second.

Sukkarieh says the internal software is under development to allow the robot to judge its own speed to minimise energy consumption for 24/7 operation.

He adds a further five years of funding will enable the RIPPA to be advanced enough to be packaged as a product, either as intellectual property or a complete robotic system, for commercialisation.

RIPPA was used in a controlled experiment to educate sensing technologies and collect data for decision support system development and validation.

The experiment focused on the 12-week growth cycle of iceberg and cos lettuce.

RIPPA used visible-spectrum imagery, hyper and multi-spectral sensing, soil probes and 3D structure sensing to identify plant growth stages, water stress and fertiliser requirements. It then precisely sprays, mechanically removes weeds and predicts crop yield.

"We have now collected the most data in the world on that particular crop over its production cycle," Sukkarieh says.

"We have been to 10 different farms at the Lockyer Valley, Cowra, Camden, Werribee, Gippsland and northern Tasmania."

Sukkarieh says the economic benefits, including weeding, fertilizer, crop yield and variability reduction, are under analysis.

"Ultimately we have to build a system which gives a return on investment within one to two years," he explains.

"The commercialisation pathways, including the robot, algorithms or sensors, have to be worked out going forward.

"We receive grower feedback, understand what they like and don't like, where the economic returns are and develop the research program around that."

Sukkarieh adds the RIPPA had to keep abreast of trends in soil and crop sensing technology, as well as potential impacts on growers.

"We are building sub-systems to give the robot the ability to interact with the crop and soil," he says.

"We are also going through the process of automation standards to allow growers to 'plug and play' with other on-farm robotic systems, such as tractors.

"One of the biggest focuses over the next year will be to reduce crop variability by one to three per cent to surpass labour costs on the farm.

"Because the robot can detect individual plants, it can determine whether a particular plant is underperforming or not, and either add fertiliser or mechanically weed it out.

"I want to get to the point where the data is a closed loop in real time – the robot sees something and can do something about it."

In the future, multiple robots, equipped with herbicide tanks, will work across farms on rotations to avoid queuing at refill stations.

They will predict future yield plus the range of potential future yield in relation to economic risk. Combined with strategic decision making by the grower, data collected by these rolling computers on the farm can significantly improve efficiency and productivity on horticulture operations in the future.