Do Androids dream of electric weeds

Ask anyone to describe what comes to their mind when they hear the word ‘Robot‘, and it is unlikely that agriculture would be at the top of their mind. This is ironic, given that the word itself has deep roots in agriculture, being derived from ‘Robota’, the Czech term for the forced labour that serfs owed on their Lord’s land during the heyday of feudalism. Half a millennium ago, such ‘Robots’ were used to solve the fundamental crisis of medieval agriculture – a lot of land ripe for cultivation, and not enough workers to provide the required labour. Validating the old adage that history repeats itself, agriculture today faces a similar potential crisis of labour availability that will create a requirement for a new generation of robots. However, a closer view of the landscape highlights that we are far away from a general deluge of autonomous robots taking over every farm in sight – the crisis and the opportunity are very real but so are the limits to the size of the pie. A critical understanding of the likely evolution of robot adoption across agricultural segments should help entrepreneurs and investors think about where best to channel their efforts and their investments.

Labour in Crisis

While the challenge of feeding a population of 9bn by 2050 cannot be trivialized, it is made substantially more complex by another megatrend that has perhaps not yet received sufficient attention in the public mind as well as the AgTech industry – viz. the challenge of feeding 9bn people in an environment where ever fewer people are going to be engaged in the activity of agriculture[2]. While the Agriculture industry invests billions in research on optimizing the quality and usage agricultural inputs such as seeds, chemicals and water; labour, which is an equally critical input, has not yet received the same attention – despite some alarming megatrends on the horizon. Over the last six decades fertility rates have declined dramatically all over the world[3] and simultaneously, the proportion of the labour force engaged in agriculture has dropped drastically across all economies, developed or developing. The reasons for this are not hard to identify – agricultural labour is physically demanding, seasonal and poorly remunerated. With urbanization and economic development, the attractiveness of agriculture for labour goes down – young men and women in villages across the world would much rather work in construction or drive taxis in cities than spend their youth in the field. Looking to the future, agricultural labour could suffer from a serious supply side crisis, one that could not be solved by simply raising wages or relaxing work restrictions.

The impact of this impending crisis will vary by crop and geography. In developed markets (primarily North America and Western Europe), heavy levels of mechanization have substantially lowered the labour intensity of key row crops such as Corn and Soy[5] – hence, the addressable market for Robotics solutions in these crops is likely to be very limited. The one major exception in the developed world is the cultivation of fruits, where labour requirements are substantially higher due to the lack of mechanization in harvesting operations.

Paradoxically, it is in developing countries, where farming is non-mechanized, that any radical shifts in labour away from agriculture will have a much greater impact. Almost all crops in the developing world have extremely high labour requirements across key operations like land preparation, weeding and harvesting (see Fig 5). Most of these activities are carried out with large labour crews and the increasing difficulty of sourcing such crews has made farmers in the developing world as anxious for labour substitution solutions as their counterparts in the West.

Labour substitution strategies

Shortages of labour supply have had dramatic impacts on human economic systems and have historically created conditions for technological innovation.  In the case of agriculture, the impending labour crisis can be overcome through three main technological interventions:

• Herbicides: chemicals (synthetic or otherwise) directly replace manual labour used in weeding operations, especially in developing markets
• Machinery: ranging from tractors to seed drills, these can eliminate large crews of labour at the cost of one or two human operators
• Robots: machines that can carry out complex tasks without persistent human input and can act independently over a range of external conditions could replace humans in agricultural operations that require judgment

Thus, it should suffice to say that there will be no ‘market for Robotics’ – there will be one for labour substitution, with Robotics being just one way to address labour shortage. In particular, the developing world will undergo increased mechanization and greater use of herbicides rather than shift en masse to robotics. However Robotics will be a solution for certain niche segments in Ag, the most likely being

• Fruits and vegetables in developed markets: harvesting operations in large scale plantations are likely to be more successful entry points for Robotics in Ag as compared to row crops due to the very high manpower requirements, lack of existing mechanization solutions and the magnitude of ‘value at risk’ if harvesting is not done on time
• Weed removal in developing markets: Paradoxically, the developing world could also see application of low end or frugal robots to replace manual weeding for small scale farms where the challenge of sourcing low-skilled labour for physically taxing and unremunerative work is no less acute
• Precision spraying or weeding in developed markets: adoption in this segment is driven less by labour scarcity than by the need to reducing chemical intensity – the value capture potential in this segment is accordingly limited and this niche will probably require an enabling regulatory environment to facilitate adoption

Challenge

While the need for Robotic solutions for Agriculture may be clear, it is also imperative to understand the challenges that remain to be overcome before we can see successful adoption of Robotics technology on a commercial scale in Agriculture.

• Technology: unlike industrial robots that operate under controlled environments, agricultural robots out in the field will be exposed to a greater variation of operating conditions, such as weather and terrain. In addition, to be competitive with human labour and other replacement solutions, robots will also need to demonstrate operating rates and speed of decision making on par with humans.
• Business model challenge: all through history, it has been demonstrated that technology by itself does not lead to technological revolution in the absence of a compelling economic rationale[9] . Hence, a technically sound robot by itself is unlikely to be an investible proposition – what will be necessary will be a robot that can be operated at a low enough cost over small to medium sized fields where initial adoption is likely to occur. Additionally, startups looking to address this space will need to work-out the business model viz. a service model aggregated over disaggregated demand nodes or a more traditional product sale.

It is perhaps a recognition of the enormity of some of these challenges that VC funding in the space of Robotics has been limited – Robotics based startups received $40m of funding in 2015, a pale fraction of the ~$350m invested in drones in the same year[10]. Given the impending crisis of farm labour availability, it would seem logical to expect that this situation will change in the coming few years. The scale of challenges to be overcome however, makes it likely that the successful adoption of robotics in Ag is more of a 10 year project than a 5 year one and only startups that have a clearly defined niche in which they offer a clear, ROI driven value proposition are likely to emerge as winners. It is fitting perhaps to end with a quote by Marvin Minsky that perhaps captures the spirit of the hopes from Robotics in Agriculture.

In the fifties, it was predicted that in 5 years robots would everywhere.
In the sixties, it was predicted that in 10 years robots would be everywhere. 
In the seventies, it was predicted that in 20 years robots would be everywhere.
In the eighties, it was predicted that in 40 years robots would be everywhere...

[1] Men harvesting wheat on royal demesne in England, public domain image taken from  https://en.wikipedia.org/wiki/Socage#/media/File:Reeve_and_Serfs.jpg

[2] As this article was being written, a new book titled ‘Empty Planet, The Shock of Global Population Decline’ was released, substantiating the thesis of this article with data and primary research across the globe

[3] Fertility rates in Latin America, China and India have plummeted from ~6 to 1.8-2.3 in this period, as per UN statistics

[4] ILO, Hill A. ‘Labour Supply of US Agricultural Workers’, FICCI, KPMG ‘Labour in Indian Agriculture’

[5] In the course of this article, the terms ‘developed markets’ and ‘mechanized systems of Agriculture’ will be used as proxies for each other though it should be remembered that it is possible to have non-mechanized farms in developed markets and vice-versa

[6] Rivas, Kastner and Nonhebel, ‘How much time does a farmer spend to produce my food’, most of the data for non-mechanized systems refers to India while mechanized system data pertains to the US

[7] FICCI, KPMG ‘Labour in Indian Agriculture’, van Heemst, HDJ ‘Crop Calendar, workability and labour requirements’, Jabbar MA, Faruque AKM ‘Labour requirements for major crops in Bangladesh)

[8] Pre-planting includes land preparation, N-application and weeding

[9] Roman civilization was aware of steam technology yet they did not develop a steam engine – primarily because the alternative available, slave labour, was abundant and cheap

[10] AgFunder ‘AgTech Investing Report – 2015’; no break-up of investments between drones, Robotics and Mechanization provided for subsequent years