ACTIVITY NOTE
Participated in IROS2024 @Abu Dhabi
Oct 14, 2024
I participated in IROS (IEEE International Conference on Intelligent Robots and Systems) in Abu Dhabi with a Late Breaking Poster (LBP) presentation. As an extension of the Predictive-Coding inspired Variational RNN (PV-RNN), I analyzed how the internal state representations change when the number of learned motions is gradually increased through incremental learning, using principal component analysis (PCA). Since these are preliminary results, I presented them as an LBP, but I plan to further develop this work in more interesting directions in the future.
I attended the IEEE International Conference on Intelligent Robots and Systems (IROS) 2024, held in Abu Dhabi from October 14 to 18, 2024.
This was my first time attending an international conference in person since the COVID-19 pandemic, and I was genuinely excited even before the trip.
In my presentation, I extended previous research on the Predictive-Coding inspired Variational RNN (PV-RNN). Specifically, I analyzed how the internal state representations change when a robot incrementally learns new motions, using principal component analysis (PCA) to capture these changes.
Since PV-RNN learns each motion in a nonlinear manner, I hypothesized that incremental learning might lead to emergent novel behaviors. To test this, I incrementally increased the number of learned motions from four to nine by guiding a humanoid robot’s two arms. The results showed that the internal representations of different motions gradually began to blend together, as indicated by the expanding distributions observed in PCA space.
Unfortunately, even with nine motions, no creative or entirely novel behaviors emerged. I am currently considering how to further develop this line of research, but for now, I will put this study on hold as I am heading to France for an internship. Below are the poster presented at IROS and my participation certificate.
発表ポスターはこちら
参加証はこちら
