Wildly tenacious, methodical and thorough to a fault, Michael Mwaisakenyi’s journey to becoming a formidable innovator began with his routine visits to his grandfather’s farm. But a visit in December 2018 was unlike the others and Michael had a pressing question on his mind. The answer would change his life’s trajectory significantly. Michael’s grandfather was a seasoned farmer with an extensive farm at his disposal, but his harvests were poor. Upon realising this, Michael posed a question: why was his grandfather’s crop yield so low?
When Michael probed his grandfather about this disparity, he learned that crops would die before harvest – especially due to weeds. What was more worrying was his grandfather planting up to three seeds in one hole to try and increase the chances of at least one of the seeds surviving the scourge of weeds. With no farming experience, Michael was prompted to do some research on this. He learned that overcrowding seeds was a losing battle. These seeds ended up struggling for resources rather than benefiting from exclusive water and nutrients in their own hole.
Then a Mechatronics Engineering student at JKUAT since 2014, Michael identified a unique opportunity to merge his knowledge in technology with farming. For his final year project that began in January 2019, Michael decided to develop a tech-based solution for the struggle of small-scale farmers against weeds.
This was the start of Michael’s journey to building a weeding bot that could independently pull out the offending plants on farms. It was an ambitious project – and one that could have ended in failing the year if not executed well. It was all or nothing, but a tenacious Michael took the gamble anyway. If the project meant supporting farmers like his grandfather to have better yields, he was up for the challenge.
An Uphill Battle
From the start, Michael noticed how limited his resources would be. For starters, he had only a small budget given by his school to work with. Also, the parts he required to build the bot were not readily available in Kenya, so he enlisted the help of local artisans. They created parts, and Michael used his laptop for machine learning – but the process was painstaking.
In addition, there were no manual weeding machines available to consumers at the time. This meant that Michael was attempting to build a robot with almost no reference points. There was yet another challenge. To train a robot to recognise weeds, Michael had to delve even deeper into the research he began after visiting his grandfather’s farm to gain a foothold in agriculture. Through this continued research effort, Michael found out about the widespread use of herbicides that have major health risks and poison the very crops they are supposed to protect. This steadied the engineering student on his path to providing a non-chemical solution.
Despite the challenges, Michael was as determined as ever. Towards the end of 2019, Michael had a working proof of concept. The bot could move around fields on its own and had the tools to remove weeds.
The Path to Victory
Although Michael had a viable innovation, there were still weak areas in the technology. In November 2019, the next steps became clearer as Michael met his weed bot partner, Ken Gicira, through a project at Gearbox. With a mutual appetite for risk and zeal for tech, they pushed the weed bot to higher heights through a competition held by Microsoft called the Imagine Cup. The opportunity arose to switch over the programming of the bot to Microsoft technology – which proved to be faster.
By the end of the competition, the team members had a bot that used camera sensors to identify weeds. These cameras would then give commands to the robot’s arm to reach out and pluck the weed. Even more impressive was the bot’s 99% accuracy in identifying and plucking weeds. With this ingenious design, their top position in the EMEA region (Europe, Middle East, Africa) was a given. It was the first time that an African team had cleared the semi-finals! On top of the regional win, the teammates received a cool cash prize of 800,000 shillings each to further the innovation.
Although the bot is not yet in the hands of consumers, it represents the possibility of viable solutions for inefficient food production systems.
We salute Michael for doing his part to inspire us to believe that we can forge the solutions we need. Indeed, uplifting smallholder farmers helps build sustainable food systems and establish food security.
#KeInnovators
