Disrupting Food Systems: Landing Industry 4.0 Technologies

A Step Beyond Drones In Precision Agriculture.

Towards Deliberate Diffusion of Innovation

If you are a student of the history of innovations and technology adoptions, you will undoubtedly have heard versions of the theory of?Diffusion of Innovations?popularized by Everett Rogers (1962). Diffusion of Innovations theory seeks to explain how, why, and at what rate new ideas and technology spread.?

In his book?Crossing the Chasm, organizational theorist Geoffrey A. Moore advances Everett’s theory arguing?that “there is a chasm between the early adopters (the technology enthusiasts and visionaries) and the early majority (the pragmatists).”?Moore believes visionaries and pragmatists have very different expectations, and he attempts to explore those differences and suggest techniques to cross the chasm successfully.?

Several factors affect the rate of innovation diffusion, including working out the cultural dynamics, the society’s level of education, and the current levels of industrialization. The risk appetite for innovations differs across communities, affecting adoption rates. The cost, accessibility, and familiarity with technological change also gate adoption rates. On the other hand, the theory of Innovation Diffusion does not prescribe a participatory approach to adoption. However, innovation promoters may employ enthusiastic early adopters to evangelize the new technologies.

Know Thy Data!

Dimensions Of Drone Data

One of the technologies successfully landed in Malawi is drone technology. One cannot talk about precision agriculture in Malawi without straying into the specifics of drone technologies. The increased power efficiency and growing payload capacity have improved the number and types of drone sensors and diversified their fields of application. A few local start-ups invest in manufacturing drones and parts and train others.?

Using a drone or, more formally, Unmanned Aerial Vehicles (UAVs) provides innovative farming solutions applying real-time data gathering and processing to improve farm-wide decision-making and efficiency. Current use cases include surveying, mapping, crop health monitoring, and disease detection.

UAV localization typically uses a global positioning system (GPS) and multi-axis Inertial Measurement Units (IMU) for heading, leveling, and orientation measurements. They occasionally employ barometric pressure sensors to augment the GPS z-axis. The IMU extends the GPS dataset with attitude, heading, speed, yaw, pitch and roll angles, and relative position information. Typically, IMU measurements are delivered faster than GPS signal updates and may be used for heading projections when GPS data is inadequate or fails. As a general rule, the IMU provides high frequency, high accuracy data subject to drift, and the GPS unit provides low frequency, low accuracy data about absolute position. Sensor fusion allows combining sensor data from several sources to generate a more consistent, dependable, and accurate outcome.

An IMU provides position information relative to a known starting point. The IMU comprises 6 Degree of Freedom (6-DOF) from a 3-axis accelerometer (3-DOF) and 3-axis gyroscope (3-DOF). Accelerometers measure linear acceleration, and gyroscopes measure angular velocity (rotational rate). A 9-DOF IMU adds a 3-axis magnetometer for magnetic field detection. The 10th DOF usually comes from a barometric pressure sensor for altitude detection.

The IMU also allows the UAV to not rely on Ground Control Points (GCP) for Georeferencing. MEMS IMU-enabled Direct Georeferencing allows more efficient aerial photogrammetry and digital surface model creation. Direct Georeferencing can also use mapping sensors such as LiDAR systems as well as hyperspectral and multispectral imaging sensors (imaging spectroscopy).

GIS, on the other hand, links data to a map, integrating location data with all types of descriptive information, providing a foundation for mapping and analysis and helping users understand patterns, relationships, and geographic context. The fine-grained observations enabled by these highly capable drones promote accessible monitoring and mapping of yield and crop parameter data.?UAVs permit precise variable-rate prescription of agricultural resources.

How Did Drones Find Themselves At Home In Malawi?

To quote the legendary Bruce Lee,?“It’s like a finger pointing away to the moon. Do not concentrate on the finger, or you will miss all of the heavenly glory!”

Lest you believe precision agriculture in Africa is about piloting drones over fields. The primary reason for the popularity of drones is that they have become relatable; therefore, they are an easy win for many people engaging in the subject. Demystifying the technology is always phase zero of the mass adoption journey.

The drone is but a data collection and mechanical delivery tool. However, drones have been introduced with a purposeful context, thereby removing the focus on the technology itself. As a fun exercise, look at the?UAV kinematic model?and map that back to the sensing technologies previously discussed.

While there is beauty and skill in drone piloting, these are primarily pursuits in robotics, motion planning, and computer vision. On their own, they do not advance food systems. The story of the data beyond the drone-assisted collection is seldom told in full. The solutions rely entirely on human experts. The typical Malawian smallholder farmer cannot afford this.?

Naive implementations of the decision-support systems try to use the classic Expert Systems built on a series of IF-THEN rules. The issues of rapidly growing volumes, variety, and velocity of data quickly retire these solutions to oblivion. Creating a search strategy in classic Expert Systems to classify drone datasets correctly is nigh humanly impracticable. In general, Rule-Based Expert Systems cannot learn from the experience. They fail to gracefully handle noisy data or unexpected adjustments in the rules very well. In contrast, Machine Learning can robustly deal with more variability and uncertainty.

Developing robust automated decision-making requires more knowledge and resources not readily available to many. Decomposing the problem space into implementable chunks and integrating them into a coherent system requires advanced problem-solving skills. Big Data is a fundamental problem not unique to Africa.?

?At the current level of maturity, the drive now may be to create canned solutions that are easily palatable.

Accurate crop yield predictions contribute to strengthening food security strategies at all levels. Using the output of phase zero is necessary to build the subsequent phases requiring new ideas combined with locally-relevant experiences.

Precision Agriculture is an umbrella term that includes variable rate technologies (VRT) and yield prediction. VRT looks at the application of inputs based on the characteristics and the qualities of the area under consideration.

Predicting crop yields involves building AI models to make predictions based on structured and unstructured data. The first challenge is to build a quality knowledge base. Once this knowledge base is in place, we must create an inference engine that is robust in the presence of noise in the data or the environment.

These are highly technical domains.

For example, the quality of data obtained from sensors in IoT tightly couples it to measurement theory.?Differing interpretations?of measurement theory, statistical representations, probability theory, and the suitability of models all depend on the hypothesis posed. AI can break this data into umpteenth levels of gradation for found data. The human experience is required to balance the level of obscurity. The luxury of starting with a value proposition risks losing existing datasets in an environment where collecting the data is expensive. Domain knowledge is critical to determine the value and articulate the business case.

Precision is an advanced technical aspect for many people. That is where domain knowledge and science meet. The level of details, the complexity of the environment, and the nature of the questions point toward a plausible AI challenge.

The danger of focusing on singular technologies is that it distracts from the means to get the necessary value out of data.

An agriculture technology strategy needs to include the entire analytics toolchain, including the provisioning of the necessary IT infrastructures. The solutions must define clear steps from articulating the business purpose, collecting and processing the data, to extracting valuable insights.

Fitting The Silhouette Of A Regular Hero

It is tempting to romanticize the idea of a big hero swooping in and saving the day. Who doesn’t like a brilliant underdog story??

When landing technologies, it rarely works like that. We can disabuse ourselves of the idea of a regular hero parachuting into a situation, saving the day, and everyone living happily ever after. The innovation partner introducing the technology must deliberately execute an inclusive strategy that effectively grows with the communities on the ground. The protagonists typically look like a network of professionals proactively applying human-centric designs to socio-technical challenges.

In socio-technical challenges, lived experiences matter! Qualitative phenomenological research characterizes?lived experience?as a representation of personal experiences and choices and the knowledge one gains from them. It is naive and uninformed to fantasize about diving into a relatively unmapped environment and solving all problems from first principles. The innovations must incorporate the basic tenets of the culture of the target communities. Landing Industry 4.0 technologies in an African context is nuanced. Africa is that “everywhere” that is not like everywhere else.

The cautionary phrase “Nihil de Nobis, sine Nobis”?(nothing about us without us) says it all.

However we slice it, we can choose to be oblivious or pay attention to a new market connecting 1.3 billion people across 55 countries with a combined GDP valued at US$3.4 trillion. The African Continental Free Trade Agreement (AfCFTA) is shaping up to be the world’s largest free trade area with respect to the number of member countries, territories, and populations.

Understanding The Malawian Base: Literacy Challenge

Technology transfer and knowledge embedding cannot happen without the direct involvement of the target communities. The technologies in Industry 4.0 are complex. There is a noticeable gap in the expected level of entry.?UNESCO SDG target 4.6 for Malawi?(2015) estimates a combined youth literacy rate of 72.9% (Female: 73.4%, Male: 72.5%) and a combined adult literacy rate of 62.1% (Female: 55.2%, Male: 69.8%). Landing Industry 4.0 technologies must be married to improvements in?functional literacy.?

Landing Industry 4.0 technologies in Africa must resemble all the stages of building a Community of Practice to provide the necessary scaffolding, and the infrastructure must allow it.

Understanding The Malawian Base: Infrastructure Challenges

The connected demand for electricity in Malawi far exceeds the 532 MW installed generation capacity. Large-scale solar farms at Golomoti (20 MW), Nkhoma Deka (50 MW), Salima (60 MW), and the one planned for Bwengu in Rumphi are working towards alleviating the energy stress. However, even when fully delivered, these projects would only start to breach the required industrial production base capacity. Thermal energy demands exert significant pressure on forest resources, leading to forest degradation and uncontrolled deforestation, affecting hydroelectric power generation.?

The energy deficit hampers significant value-addition and data-driven strategies, which stresses agricultural capacity building. There are substantial opportunities for implementing?affordable small run-of-the-river hydropower systems?on several major rivers and outlets, which would boost the capacity with minimal environmental impact. Several renewable energy initiatives, including biogas and reclaiming energy from general household waste, also open up opportunities to ease overdependence on firewood.

Aligning With The Sub-Saharan Africa Reality

In Sub-Saharan Africa, the average farmer has yet to attain the technological sophistication demanded by high-end mechanized agricultural solutions. While most smallholder farmers own less than 2 hectares of land, we can put the large mechanized equipment as a future ambition. It is not impossible to achieve. It is merely a question of exposure, capacity building, and economic empowerment. On the other hand, the degree of penetration of education in the farming communities determines which technologies and details are relevant to inform adoption, operation, and maintenance.

Further, what skills are required to collect and process the correct datasets to add value to agricultural operations? What is the cost of data storage? What computational capacity is needed to process the data? Does it make sense to store data locally? Which edge-intelligence technologies reduce the amount of data sent back to base? Are self-serve analytics tools accessible to the farming community, or do we expect all farmers to become data scientists?

In other words, we need to size the rhetoric of precision agriculture to fit the Sub-Saharan African situation. Hoping for a different base is but that, hope. Unless we can calculate it, it is impossible to operationalize hope. Given that we know the desired goal, available infrastructure, and projected growth trajectory, perhaps it is time we reframed the question as an engineering challenge.

Let’s take a step back and look at the route to mechanization and the enablers of the science in precision agriculture.?

The World Bank defines the?International Poverty Line (IPL)?as US$1.90 per day per capita. In 2016, the World Bank estimated that in Malawi, 70.3% (roughly 12.1 million people) live under the IPL. 90% of the poor live in rural areas. According to the??CIA FACT Book (2022), 80% of the population of Malawi has a career in agriculture. Additionally, the COVID-19 pandemic and recurrent shocks such as droughts, floods, and cyclones in the recent past have significantly impacted livelihoods and infrastructure.?In this environment, technology has a human face. We must apply the tenets of the human-centric design paradigm.

Solutions that capitalize on introducing a massive shock to the status quo might find it hard to gather significant traction. Discontinuities make the users uncomfortable. On the other hand, a scaffolded approach involves a More-Knowledgeable-Other guiding the userbase across the chasm - Lev?Vygotsky’s?Zone of Proximal Development (ZPD). The Community of Practice approach offers this.

Cases In Point

The open-source communities, the spirit of collaborations in academia, and citizen science projects have shown that well-executed projects can significantly impact their ecosystems. Specifically, this author would argue the core principles at the heart of the?“Open Source Way”?provide a solid guideline.

The?Centre of Effective Altruism?writes:?“[Effective Altruism] is a research field which uses high-quality evidence and careful reasoning to work out how to help others as much as possible. It is also a community of people taking [these] answers seriously by focusing their efforts on the most promising solutions to the world’s most pressing problems.”

One of the philosophies advocated by Effective Altruism resonates very well in the current context. We need to improve the scientific establishment and add greater transparency and replication of results?to set the baseline for the aspired growth. The philosophy advocates effective collaboration plucking at the proverbial low-hanging fruits toward achieving outstanding community outcomes. These low-hanging fruits are the springboards for understanding and communicating more complex examples.

We can only talk about ephemeral datasets or data silos without a ubiquitous adoption of relatively advanced IT tools. The most relevant question remains how to bridge the chasm.

African leapfrogging narratives?in telecommunications, adoption of AI-driven healthcare and diagnostics, and FinTech have whetted the imagination of many, and that appetite is crossing over to AgriTech. The UAV technologies referenced previously were introduced in the context of health services and disaster response. Now we can talk about robotics and IoT without batting an eyelid. The adoption of AI in agriculture is enabled by robotics and IoT, serving the increased consumption and rising requirement for better yields of crops.

The?Investing in Africa Forum?is a global platform for multilateral cooperation and promoting opportunities to increase investment in Africa. The IAF’s Leapfrogging Africa’s development framework covers six investment sectors: agriculture, education, energy, finance, governance, and ICT. The education, ICT, and agricultural extension services translate the complex technologies to aid adoption. In other words, the immediate demands in the African context are self-serve tools that are easy to interpret. There is a vibrant IT community in Sub-Saharan Africa with several Centres of Excellence popping up. In the short term, this budding community is the intermediary that will continuously translate the detailed tech-speak into accessible language for Jim, Jack, Li, and Sunduzwayo.

The?Government of Malawi and UNICEF?launched Africa’s first air corridor to test potential humanitarian use cases of drones - an 80km diameter zone centered on?Kasungu Aerodrome. The drone corridor provides a controlled platform for the private sector, universities, and other partners to explore how UAVs can help deliver services that benefit communities.?The African Drone and Data Academy?in Malawi followed up to scale the local expertise. Specifically, the Academy offers Drone Basics, Logistics and Planning, and Data and GIS Analytics. These are all transferrable skills imparted in context. More importantly, they demystify the technology enough to reduce the gaps and spur innovation.

In the wake of the devastating cyclone Idai in southern Africa, UNICEF, the Malawi Red Cross, and the Department of Disaster Management Affairs (DoDMA) used UAVs to map the extent of the flooding and the damage.

How Do We Get Started?

The successful collaboration strategy between local policymakers, the private sector, and international and local NGOs for technology and scientific intervention adoption has brought a positive focus on worthwhile partnerships. The result is that despite the inadequate infrastructure and low levels of functional literacy, Malawi meets the maturity level for adopting several of the relatable Industry 4.0 technologies and spurring its fitting version of precision agriculture. These examples are not unique to Malawi.?

We must observe the correct metric to move the needle of maturity.

In the meantime, the conversation must evolve towards dealing with the hurdles of storing and processing the vast amounts of data and understanding the value of UAVs’ GIS datasets.

Conflict of Interest:?The context of the Disrupting Food System series is primarily a data-driven commentary on first-hand experiences researching, implementing large-scale socio-technical solutions to locally-relevant issues, and venturing into large-scale commercial Regenerative Precision Agriculture, ethical off-taking, and trading in Malawi, Tanzania, and South Africa.

Precision Agriculture integrates Artificial Intelligence and Internet of Things (IoT) devices to adjust for many variables affecting yield. On the other hand, regenerative agriculture is a conservation and rehabilitation approach to food and farming systems.