This maker spotlight was brought to us through Maker Faire Rome. You’ll be able to find them and many more creative and exciting makers at Maker Faire Rome: The European Edition on October 18-20. Get your tickets now!
Who are you?
Hiroshi Momose, team member of team: Laboratory for Integrated NanoSystems (LINAS).
Where are you located?
Sapporo, Hokkaido, Japan.
What is your day job?
I am a research fellow at Hokkaido University. My research interests comprise Edge AI systems and promoting Edge AI’s open-innovation platform to target a wide range of users from school kids to makers who have programming skills but limited technical knowledge of AI.
What makerspace/hackerspace/fablab do you attend, if any?
I haven’t attended the spaces/lab. My work involves the following: In 2018, I and the other members of team: LINAS, set up a community platform for our Edge AI system on our website. We currently open the website only in Japan and want to increase the outreach of our community to the global audience, starting from the autumn of 2019. In case of any questions or requests, you can feel free to drop us an email at: fpga2i-support@ist.hokudai.ac.jp. Our activities to promote the community include the following:
- Participating in the exhibitions of the Maker Fairs
- Delivering lectures at high schools, junior colleges of technology, universities, and laboratories
- Conducting workshops for public beta testing
- Promotions in technical magazines and at academic conferences.
What kinds of stuff do you make?
The Arduino would have been the best choice; however, it cannot handle the AI smoothly. Therefore, we designed and developed the FPGA2I shield considering the AI calculation (especially, for the learning process). We set up and manage the community support forum in Japan. The users can download the AI libraries and application examples from the website. The details of our platform are as follows:
- Hardware: FPGA AI (FPAG2I) shield for Arduino as an AI accelerator (training/inference)
- Software: AI libraries of Arduino sketch and FPGA configuration data
- Community support forum: Web server.
The various types of sample applications provided by us are:
- Basic samples: Logic, auto volume control, distance control, etc.
- Application samples: Line predictor, image recognition, and auto driving
Daily activities and basic samples:
Auto driving (reinforcement learning)
Line predictor (recurrent neural network)
Rock-paper-scissors (image recognition)
Which links would you like to include?
The details of our “FPGA AI (FPGA2I) Shield for Arduino rev.2 (Maker Faire Rome 2019)” platform and several application samples can be found at the links below.
1. FPGA AI (FPGA2I) Shield for Arduino rev.2 (Maker Faire Rome 2019):
2. Reconfigurable AI shield for embedded microcontrollers (Maker Faire Rome 2019)
How did you get started making stuff?
Except for the big and expensive machines, there are very few devices on the edge that you can touch, try, learn, and enjoy AI with. Therefore, we undertook the FPGA2I shield development under the open innovation platform for the Edge AI. This work is based on the results obtained from a project commissioned by the New Energy and Industrial Technology Development Organization (NEDO).
What is something you’ve made that really stands out, that you’re proud of?
- We have built a unique device that provides the capabilities for both learning (on-line learning) and inference with a single board. Therefore, you can use the FPGA2I on-site with only Arduino and the I/O (sensors, actuators, and robots).
- You can try, learn, and enjoy AI. You can also build their own AI.
- School kids who have the knowledge of Arduino sketch programming can use AI by modifying our samples in the first stage.
- Makers with the knowledge of re-configuration of the FPGA circuits can create new AI on the FPGA2I with their original neural network, thereby realizing their novel ideas.
What do you have on your horizon?
We visualize the Edge AI world together with you and your creative and novel AI ideas.
What is something you’d like to work with but haven’t yet?
We want to increase the outreach of our community to the global audience, starting from the autumn of 2019. Additionally, we hope to sell the FPGA2I shield by the end of 2020.
Any advice for people reading this?
We invite you to visit our booth in Rome or our online community. We are sure that through your experience, your creativity would be stimulated to pioneer a novel AI idea. In case of any questions or requests, you can feel free to drop us an email at: fpga2i-support@ist.hokudai.ac.jp.
ADVERTISEMENT
