East of Tokyo, the fields of Chiba Prefecture recently played host to an intriguing experiment: To determine whether robots can play a useful role on farms facing labor shortages. As Japan’s population continues to shrink, rural areas are being hit hard by an aging and shrinking farming community. To combat this trend, a team of researchers from Tokyo is developing robots that take on some jobs around the farm.
Agriculture with AI
Machine Learning Tokyo (MLT) is a volunteer-run nonprofit organization made up of people who are keen on machine learning, an increasingly popular approach to artificial intelligence (AI). Researchers from MLT went to farms affiliated with Hackerfarm, a community of technologists and farmers focused on information technology solutions, and experimented with robots that can move along a suspension system and recognize fruits and other objects. The machines can also be programmed to do photography and even use manipulators to harvest fruit. Such automated systems could work with Internet of Things (IoT) devices to continuously monitor and manage crops, lightening the workload for farmers.
“This is not only important for Japan — it’s important for developing countries because agriculture is their major source of income,” MLT founder and director Suzana Ilic told Rakuten Today on the sidelines of Rakuten Technology Conference 2019. “Having such a system could be a game-changer for developing countries.”
Ilic quotes computer scientist Andrew Ng in describing AI as “the new electricity.” MLT’s mission is to democratize this new resource, especially machine learning, with open source projects. Apart from specific projects like farm robots, it hosts open education initiatives such as bootcamps on machine learning with academic powerhouses such as the University of Tokyo, Tokyo Institute of Technology and research center RIKEN.
Founded in July 2017 with only two members, MLT differs from academic groups in Japan in a number of ways. It’s an English-speaking group and it’s dedicated to doing hands-on projects such as writing code. That approach has resonated with many people interested in AI, and membership has expanded rapidly. MLT now has nearly 4,000 members centered on a core team of 12 engineers and researchers alongside roughly 50 active contributors. In May 2019, it qualified for official nonprofit status. MLT’s efforts to open up AI were recognized again this year with the Rakuten Technology & Innovation Silver Award.
“This is an incredible opportunity for us, and I’m happy to accept this on behalf of the community,” said Ilic. “We want to encourage participation in AI for social good. This means a lot to us because this award is very well known in Japan.”
A native of Austria, Ilic did her undergraduate work in applied linguistics and became interested in the idea of using computational methods to understand language. She felt that statistical tools were not adequate for the task of processing natural language, and turned to machine learning and deep learning, which, like linguistics, are concerned with patterns. Her research has taken her from the University of Innsbruck, where she is completing a PhD, to Japan, where she has lived for over three years.
Opportunities for global talent in Japan
Japan is working to raise its profile as a global player in the field of artificial intelligence and attracting top-tier talent along the way. Ilic has done stints at the National Institute of Informatics, where she was a visiting researcher, and RIKEN, where she is currently working as a visiting scientist. She has also worked on chatbot systems at Google Japan.
“Japan offers a lot of opportunities, and I thought that there would be no better place for me to be than Japan,” she says. “I was interested in machine learning and started to experiment with new things. I learned that it’s very important to plan well ahead to be aware of certain risks, the potential, as well as the limitations and capabilities of deep learning.”
Aside from her academic work, Ilic is focused on continuing to attract talented members to MLT, which will soon present the results of its farm robot research at a Vancouver conference on machine learning for the developing world. She’s also committed to supporting an inclusive debate on the potential of AI.
“AI is changing everything and it’s heavily dependent on data. Data is human-made — there’s a lot of bias and ethical issues that we need to consider, which is the best reason for diversity in AI,” she says. “We need all kinds of people, and women are really underrepresented in AI and in technology in general. That’s a problem — we need very diverse teams.”
For more about Rakuten Technology Conference, visit here.