Robotics Research

September 2017

Developing automated solutions for primary sector needs at Waikato University

The University of Waikato Agri-Engineering Group comes under the umbrella of the Agritech Group. This is an informal grouping of projects that grew out of an audit of the Engineering School. The group has 3 branches - environment, computing and agri-engineering lead by Professor Mike Duke. 

Mike’s group has a focus on technical innovation and automation for the primary industries. The group has established relationships with a number of industry players. They’re part of a consortium working with RoboticsPlus on the robotic kiwifruit harvester that is being rolled out this harvest. The consortium is also working on automated pollinating machines for apple and kiwifruit orchards. Rainer Künnemeyer is working with Plant & Food Research on a fruit sensing system (around ripeness). 

Mike says a lot of primary industry companies have processes that are ripe for automation – not merely for improved efficency, but to counteract the use of antiquated equipment, the real problem of finding seasonal labour and increasing consumer demands for better environmental practice. 

Automation is already in play in the manufacturing environments and in easily controlled indoor environments. The ‘in the field’ environment of many primary industries is a difficult area for automation, due to the many variables created by weather, terrain, crops and animals. 

There are areas where complex robotics is not the only solution. Smart automation, such as adding improvements like sensors to old technology also offers many opportunities. 

Computer engineering and ‘deep neural networks’ (DNN) are an exciting development in primary industry automation. DNN’s are essentially computer algorithms that can be trained to identify different components, for example, a pest bug. An automated weeder could have a component added that identifies this invasive pest and then sends an alert to the farmer. Further it could be set up to target that bug with a spray of insecticide. Accurate targeting of an invasive insect or weed would reduce the toxins being used in our environment. 

Asked about general perceptions that the robots are taking our jobs, Mike says many of the jobs they’re looking to automate are ones that no one wants and they’re having to import labour. Further, if New Zealand can get in early on this field automation, there is potential to grow a manufacturing base for these machines so there would be growth in manufacture, selling and marketing, and servicing work, as well as continued growth in the engineering sectors that are crying out for skilled people to design and develop these technologies. 

Through the UOW Agri-engineering group, Mike’s team is offering an “applied research pathway” – i.e they start with client need and work to deliver tangible solutions usually in the form of ‘proof of concept’ prototypes.

Mike say’s understanding the client need and problem is paramount at the start of the process as is managing expectations. “We all want a Rolls Royce for the cost of a Corolla”, he says. He acknowledges that the expense of these systems is in the initial building of prototypes.

Mike explains that each task in a particular industry has a tipping point that is vital to the work of engineers looking at automative solutions. So if a person can pick x amount of fruit in a minute, then a robotic solution must achieve this for the same cost (after the consideration of prototype development). This is where the New Zealand commitment to topping up industry research through MBIE or Callaghan Innovation is a boon. 

A forestry company first approached the university engineering department a few years back to develop a solution to an antiquated dibbler machine – a tool used for drilling holes to plant seedlings in nursery beds. 

Mike describes the machine thay had as a “medieval spoked thing that they’d tow behind a tractor”. It would rip up the ground while making the requisite holes for the seedling. 

Forestry seedings undergo stringent quality assurance in terms of straightness, diameter and root structure and getting a straight deep hole for planting helps to reduce rejects. The protoype produced by PhD student Ben McGuiness and Mike drills consistently deep, straight holes and has helped improve tree-stock productivity by about 30%. This in turn led to other developments in forestry. 

With money from Callaghan Inovation, UOW and the forestry company they developed the dibbler for, the Agri-Engineering Group is now developing an automated seed harvester. Presently the harvesting and grading of seedlings is done manually and it’s back breaking work in the muddy winter months. The key drivers for automating this work are the OSH issues. Repetitive strain, back issues and cuts from root trimming shears are a real issue. Further, it is very difficult to get staff to do this and the company is frequently paying a premium to bring in workers from off-shore. 

This machine is a complex array of components that needs to be able to lift pine seedlings, clean dirt from the roots without damaging the structure, trim the roots and then grade the seedling according to structure and stem diameter and sort into a viable/reject box. No small task and to add to it, the machine has to be able to work outside in the brutal Tokoroa winter and process 120,000 seedlings a day. 

“This is is a very difficult problem, as the grading has to be done in the fields, during the harvesting process and be able to activate a mechanism to direct ‘pass’ and ‘reject’ seedlings into the right boxes. Just to make it even more challenging, it has to process 120,000 seedlings per day,” Dr Duke says. 

Mike’s PhD student Ben McGuiness has progressed the prototype to the point where first trials will begin in a month or so. The “monster of a machine” is in essence a tough mobile factory. It is presently configured to remove the seedling and remove the dirt and then send it up a conveyer. Here it enters into a chamber for the visual viability assessment. It is photographed, with the photographs being assessed by a computer programme and from here the seedling is either sent to a reject box or a viable box.

This initial trial will be important to start the work of identifying on field issues and refining the machine. For example they think they might need to look at additional measures for the removal of dirt from the root structures – such as air blowers. 

Professor Duke is passionate about the possibilities of primary industry and smart automation. At this stage he sees the only real challenge is overcoming peoples’ resistance to change.