…can Zimbabwe benefit?
By Dr Tony Monda
THE goal of reducing reliance on manual labour, while increasing efficiency, product yield and quality remains the fundamental concept of incorporating autonomous robotics into agriculture.
With advanced machine learning, or even artificial intelligence (AI), being integrated in the future, machines could entirely replace the need for farmers to manually plough, sow, weed or monitor their crops.
Weeding and pest control are both critical aspects of plant maintenance and tasks that are perfect for autonomous robots. A few prototypes have already been developed, including Bonirob from Deepfield Robotics, and an automated cultivator – part of the UC Davis Smart Farm research initiative.
The Bonirob is about the size of a car and can navigate autonomously (freely) through a field of crops using video, LiDAR and satellite GPS.
Its developers used ‘machine learning’ to teach the Bonirob to identify weeds before removing them.
While UC Davis’ prototype cultivator is towed behind a tractor, it is equipped with imaging systems that can identify a fluorescent dye that the seeds are coated with when planted, and which transfers to the young plants as they sprout and start to grow.
The cultivator then cuts out the non-glowing weeds.
The same base machine can be equipped with sensors, cameras and sprayers to identify pests and application of insecticides.
These robots, and others like them, will not be operating in isolation on farms of the future.
They will be connected to autonomous tractors and the Internet of Things (IoT), enabling the whole operation to practically run itself.
Harvesting from field, tree and vine depends on knowing when the crops are ready; working around the weather and completing the harvest in the limited time available. Traditional combine, forage and specialty harvesters could benefit from autonomous tractor technology to crisscross the fields.
By adding more sophisticated technology with sensors and IoT-connectivity, the machines could automatically begin the harvest as soon as conditions are ideal, freeing the farmer for other tasks.
In the future, many of the wide variety of machines that are currently in use for crop harvesting would, no doubt, be suitable for automation.
To develop technology capable of delicate harvest work, engineers are working to create the right robotic components for these sophisticated tasks, such as Panasonic’s tomato-picking robot which incorporates sophisticated cameras and algorithms to identify a tomato’s colour, shape and location to determine its ripeness — then picks tomatoes by the stem to avoid bruising.
Another is the vacuum-powered apple picking robot by Abundant Robotics – a fruit picking prototype which uses computer vision to locate apples on the tree and determine if they are ready for harvesting.
Engineers are also designing robotic end effectors that will be capable of gently gripping fruit and vegetables tight enough to harvest, without causing any damage.
Camera systems are available today that span everything from standard photographic imaging, to infrared, ultraviolet and even hyper-spectral imaging.
These images enable the farmer to collect more detailed data than ever before, enhancing their abilities for monitoring crop health, assessing soil quality and planning planting locations to optimise resources and land use.
Being able to regularly perform these field surveys improves planning for seed planting patterns, irrigation and location mapping in both 2D and 3D.
With all this data, farmers can optimise every aspect of their land and crop management.
Now, there are drones designed specifically for agricultural use, such as spraying applications of liquid pesticides, herbicides or fertilisers on crops, offering the chance to automate yet another labour-intensive task.
Using a combination of GPS, laser measurement and ultrasonic positioning, crop-spraying drones can adapt to altitude and location easily, adjusting for variables such as wind speed, topography and geography.
This enables the drones to perform crop spraying tasks more efficiently, with greater accuracy and less waste.
Drones for use in seeding and planting are also being built and tested to replace the need for manual labour.
For example, several companies and researchers have worked on drones that can fire capsules containing seedpods with fertiliser and nutrients directly into the ground.
Drones with remote monitoring and analysis of fields and crops can tour the skies, getting the bird’s eyeview of plant health and soil conditions, or generating maps that will guide the robots and help the farmer plan for the farm’s next steps.
All of this will help create higher crop production and an increased availability of quality food.
The key to a truly ‘smart’ farm relies on the ability of all the machines and sensors being able to communicate with each other and with the farmer, even as they operate autonomously. The farmers of the future, unlike their forebears, whose time was mostly taken up by heavy labour, will spend their time performing tasks such as repairing machinery, debugging robot coding, analysing data and planning farm operations.
So, innovative, autonomous agricultural-robots and drones, brought all together by IoT, will make a farm a really ‘smart farm’ in the future.
Predictions were that IoT devices installed in agriculture would increase from 30 million in 2015 to 75 million by 2020.
Thus, if every farm becomes a smart farm in future, attaining the required 70 percent increase in food production projected by the UN and FAO is a certainty, even for Zimbabwe.
Dr Tony Monda is Zimbabwean socio-economic analyst and scholar. He is currently conducting veterinary epidemiology, agronomy and food security and agro-economic research in Zimbabwe and southern Africa.
For views and comments, email: tonym.MONDA@gmail.com