Field robotics coming to a farm near you
The use of robots for farming is not science fiction. Sydney University Professor Salah Sukkarieh and guest speaker at the 2013 Global Access Partners Summit explains why.
Australia is an international leader in the area known as “field robotics”, which is the automation of outdoor machinery. The Australian Centre for Field Robotics at the University of Sydney has had the unique opportunity of collaborating with many industry partners in mining, stevedoring, transportation, defence and environment management where we have automated machines and placed them into successful operation. The key benefits have been improved operational and energy efficiencies, occupational health and safety (OH&S) and productivity.
Agriculture robotics has had a significant following in the United States and Europe – both nations have outlined programs that involve industry, government and university coming together to address key productivity and biosecurity issues that could be addressed by robotics and intelligent systems. Two large programs are Comprehensive Automation for Speciality Crops (CASC) and Clever Robots for Crops (CROPS).
Over the last eight years I have come into contact with many growers and industry representatives from the agriculture industry that have been interested in the use of robotics and intelligent systems for a wide range of activities including the detection of invasive species and the monitoring of crops. This includes Land and Water Australia, Meat and Livestock Australia, and the Australian Weeds Research Council for the use of robotic aircraft and intelligent software to detect aquatic and woody weeds, and collaboration with University partners and the Australian Research Council for detecting locusts. We are currently looking at using robotic aircraft to map vegetation on farms to support growers with timely information about the health of their crops.
We have been collaborating with Horticulture Australia Limited (HAL) for the last year and a half to see how autonomous systems (both machines and software) could be used in tree-crop farming operations. Our focus has been on the value-add elements to crop intelligence – that is, if we had a robot that could operate in a tree-crop farm with the right type of sensors, what can we tell about the tree and fruit that would help growers in their operation.
In this project we implement one of our electric ground vehicle robots that has been used in mining and defence applications. The robot can operate for up to four hours and has thermal, visual, and laser sensors that point towards the trees as the robot travels down the rows, as well as ground conductivity and ion sensors.
Through this system we are collecting centimetre level data from different sensors of the tree and the ground below it that provides a holistic picture of the farm. After applying various data fusion and machine learning algorithms we can do tasks such as individual fruit identification, crop yield, tree architecture modelling and identification of flowers. The field trials have occurred on apple and almond farms with an extension to tropical fruit later this year.
We were recently awarded another grant from HAL to look at developing a robot for large-scale vegetable farms. The objective is to tailor the research algorithms and software to meet the vegetable industry needs as well as build a new electric robot, with the extension to look at autonomous weeding.
The focus for now has been on crop intelligence because providing better information to the grower in a timely manner would support more informed decision-making. However the technology developed here also provides the foundation for future technologies in autonomous pruning, thinning and ultimately harvesting. It is these areas that will have the greatest transformational impact.
What has surprised me the most is the rapid appreciation and adoption of the potential for robotics to transform the agriculture industry. However the growth has been limited because of the limited funding available to take a strategic view, one in which we look beyond just the robot and start to picture autonomous operations on a grand scale. For example we can easily begin to develop multiple systems that cooperate on the farm that conduct tasks such as spraying, pruning and mapping simultaneously.
Projects that we have worked on in mining, stevedoring, defence and aerospace have demonstrated such capabilities. However any form of autonomous operation also requires a rethink about farm operations in general – the farm-of-the-future will change in structure, layout and operation as it accommodates autonomous systems. In my collaboration with other industry partners we have taken on a 5-10 year outlook so that we can look at both the robots individually as well as autonomous operations in general. It requires close collaboration between industry, government and the University and a long-term view. The technology risks are low and the impact will be transformational.

Professor Salah Sukkarieh is an international expert in the research, development, operationalisation and commercialisation of field robotic systems. He has lead a number of robotics and intelligent systems R&D projects in logistics, commercial aviation, aerospace, education, environment monitoring, agriculture and mining, and has consulted to industry including Rio Tinto, BHP, Patrick Stevedores, Qantas, BAE Systems, QLD Biosecurity, Meat and Livestock Australia, and the NSW DPI amongst others. Salah is the Professor of Robotics and Intelligent Systems at the University of Sydney, and the Director of Research and Innovation at the Australian Centre for Field Robotics. He has supervised over 10 research fellows, and graduated over 25 PhDs, 5 Masters and 60 honours students. He has received over $30m in government and industry funding, national and international. Salah is on the editorial board for the Journal of Field Robotics, Journal of Autonomous Robots, and Transactions of Aerospace Systems, and has over 300 academic and industry publications in robotics and intelligent systems.