Small Farmers in AI & ML World Part-1
Updated: Nov 11, 2019
1) In Industrial evolution era we had the stationary robots which served as tool for improving production. 2) In India, during the 1980s majority of the farms were cultivated with pair of bulls; Today majority of the farms are cultivated by the tractors. 3) Similar changes are there if you compare the per capita phone usage and per capita consumption in 1999 & in 2019
These 3 examples speak a lot about how landscape changes with reference to time and within a one generation lifetime.
The AI revolution is a much talked word and it certainly will make paradigm shift in multiple verticals.
Let's peek into Agriculture - The primitive culture still feeding our tummies, thanks to farmers at work. But then Agriculture has multiple challenges. There are widespread instances where farmers has to take path of suicides due to economics. Also I believe everyone will agree with me on labour issues in the farming sector and that is true across the world, be it a small farm in hinterland of India or a Corporate large farm in USA or Australia. In India though there is labour availability, the labour wages in last one and half decade has disturbed the economic equation in profitability aspects. USA and Australia get their farm workers as seasonal labours and that is contributes to migration issues.
About Indian Agriculture there may be an impression of low cost labour which is true for software, manufacturing or any process or service industry but in Agriculture Labour takes the large pie of the spend when whole economics is considered. There are issues in terms of labour productivity and non-availability when its required most. This is also due to the seasonal nature of the business and almost all activities at farm have to happen in the period few days. For example, during Kharif season every one seeds farm in a 15 days’ time or when wheat is to be harvested it needs 15 days’ time. So if a machine is at your disposal to do the task it will solve your issues on low labour productivity and non-availability.
Below operations performed are during the farming season 1. Drive the machines like tractors and power tillers for land preparation 2. Remove weed that compete with the main crop for nutrition, air and sunlight. 3. Fertilizer dose infusion near root of the plant and crop protection with sprayers. 4. Harvesting flowers, grains or fruits or roots with various mechanisms, Storage and taking it to market.
Now let's take consider 1) Tractor operation and how its evolving
Today driverless cars are buzzword and being sold in Market, tomorrow on the farms we will have driverless tractors. To no surprise, self-driving tractor is a reality and here is a product is released by Mahindra Tractors. The future beacons to more sophisticated versions and it may come from start-ups in AI & ML space.
The First Driverless Tractor Technology of India by Mahindra - Video
What is the repercussion of this? In 1980s a pair of bullock was able to cultivate around 3 acres of land in a day. While today a tractor can easily cultivate 15 acres in a day. This simply means, it has put up 4 persons out of job (while population has increased). Take this further, and we have driverless tractors feefedd with GPS location and boundaries of cultivation area increase manifold. Now the tractor drivers are out of job!!
Today majority of the tractors in India are driven by farmer owners and hired drivers. There are formal informal start-ups who have aggregated drivers and tractors and they will be first movers to adopt, to reduce the cost of human labour, to improve ROI and also create USP.
2) Weed Removal: As of now this extremely human intensive activity prima facie does not look like it can be replaced by a machine. But my observations indicate that technology stack is ready to venture into this activity. The Machine Learning algorithms are capable of identifying a leaf to another type of leaf. Which means it is capable of differentiating the grass from main crop. The humanoids are not new today and infused with this machine learning models can signal actuators to remove the weed. A labourer can work only in day time while humanoid can go round the clock and if its waterproof doesn't matter if it’s raining. The fertilizer doses near to the stalk are fairly simple application that can be achieved with different actuator.
Why one should venture here. Farmers always have to be dependent on the labour and in today’s time they are taking the farmers for a ride for higher remunerations (Labourers have their own compulsions to demand for higher wages).
Harvest Automation Video
3) The crop protection segment is where we are very near in application and probably it’s the low hanging fruit.
Two incidents that I relate most will create understanding for you 1) In an unfortunate incident in Yavatmal district, 23 farmers died due to spraying of fertilizer in a day in 2017 and many from Vidarbha and Telangana were admitted to hospitals 2) Same season, farmers in Telangana procured the services from a drone sprayers and they paid 1000 per acres.
There are AI & ML based apps which are helping farmers to analyse major crops and related diseases and even recommend solutions which is a stepping stone and great example of how a machine/phone is analysing probable diseases and gives actionable intelligence. In various packets of the world there are companies and start-ups working to analyse the health of standing crop with drone imagery which gives them understanding on nutrition deficiency, with this analysis based on the damage or disease level the variable spraying of the remedial liquid can be sprayed. Drone based spraying also can save the 50 to 60% of the poisonous agrochemical that we infuse in farm ecosystem.
Due to known ill effects, awareness & incident like the one listed above, farm labourers are not taking risks of spraying and those who are taking risks are charging hefty amount from farmers. It creates opportunity for ML & AI based drones to get a quick boost in farm ecosystem.
4) Harvesting any crop is quite labour intensive and you probably have come across WhatsApp videos where there are robots which are capable of picking up the fruits.
A Bangalore based start up even moved further and they have built the robotic arm to pick the cotton, which is one of the most complex task of the lot and needs more human intelligence. It is very laborious to pick the cotton compared to harvesting other fruits. For grain based or tuber based crops there are machineries available in large farms. These are there from at least a decade or so.
With AI, ML technologies these machines will become more intelligent and effective, but somehow we have not seen replicas for small farms. The clear logic is the there is no affordability for small farmers as it high capital investment for a small farmer
Then who are the early adopters to this tech evolution. Who will afford it? It’s naturally the large farm owners and farm corporates and cooperatives. And where are they?
Pan India scenario there are only few cooperatives who are professionally managed which can shift gear to adopt. We can also add few FPOs to the party. But overall the majority of small farmers will not be able to afford this tech and will not reap its benefits. This will add to the unfavorable economics.
There is already plenty of land being cultivated at mass scale. These large farms are the natural adopters of the AI &ML technologies thanks to their ability to raise the money. This step will empower them to overcome labour issues, optimize use of inputs (fertilizers, pesticides), save water, save cost. Bottom line, scale of economies will help more ROI unlike small farmers.
Well this equation is known to the world, what's new in it.
1) A large amount of land mass is getting converted in Africa and Southern America into agrarian land. The farm sizes are in Thousands of acres 2) Farmers from Punjab are selling land in India and procuring large farms in Australia
The infographics at the top indicates how countries are procuring/ taking land on lease from geographies of Africa, Australia and southern America. That indicates large farmland owners and Corporates have read the future, and are making changes accordingly. This Scale of economies empowering large farms will just make small farmer production to be non-competitive. There are examples in India of importing pigeon pea from African countries and palm oil from East Asian countries at lower cost.
So here is warning for small farmers to read the future in groups, adopt for circumstances or perish under unfortunate bad economics of farm production.
If you try to compare this fear to the apprehensions that we talked about computers replacing clerical jobs in 90's, this is different. That time computers helped in immense job creation and evolution of an Information and Communication industry to improve the productivity while AI & ML revolution and advances in computational power will bring a different change as indicated above.
Floor is open for solutions, criticism and to see the future and write for the future and venture into future. I will write sequel to this article on how small farmers should adopt and prepare of coming change.