IPAB Seminar - 01/08/2024

Mako Miyatake 

Title: Cooking robotics through the lens of chefs

Abstract: Daily cooking involves many complex tasks that require using the five senses and delicately controlling one's hands. In this talk, as a researcher in digital food fabrication and as a chef with experience in a culinary school and restaurants, I will organize the tasks performed in real-world kitchens and share insights into the possibilities of cooking robotics. By including discussions with chefs and food tech companies, I will highlight the challenges faced in various kitchen settings, including food factories, chain restaurants, high-end restaurants, and homes. I hope this presentation will help robotics researchers in identifying new research topics.

Short bio: Mako Miyatake is a freelance chef and researcher in the interdisciplinary field of food and engineering. She has been working on digital food processing systems that enable chefs to extend their creativity without technical expertise. She will be starting her PhD in Mechanical Engineering at UC Berkeley. She received a Bachelor's and Master's degree in engineering from the University of Tokyo. She also learned professional cooking skills at Tsuji Culinary Institute and obtained a chef's license in Japan.

Tatsuya Matsushima 

Title: Towards Scalable Robot Learning Systems

Abstract: Research of generalizable robot systems that are robust to changes in tasks and environments in the real world has been active in recent years, utilizing large-scale language models (LLMs) and vision language models (VLMs) such as GPT and CLIP as recognition and planning modules. On the other hand, research is underway to develop ''robotic foundation models'' more closely related to robot control by utilizing data accumulated through the robot's actions in the real world. In this presentation, we describe the line of research and our efforts to scale models and datasets for constructing such robotic foundation models.

Short bio: Tatsuya Matsushima is currently a postdoc researcher at the University of Tokyo, focusing on robot learning and its application to service robotics, leading robot learning research Unit in Matsuo-Iwasawa Lab. He participated in several competition on household service robotics, World Robot Summit (in 2021) and RoboCup@Home (in 2020, 2022, 2023 and 2024), and won prizes. Website of robotics research unit (TRAIL): https://trail.t.u-tokyo.ac.jp/