no code implementations • 1 Feb 2024 • Koki Inami, Koki Yamane, Sho Sakaino
The proposed method enables to increase in the amount of motion data available for training and the accuracy of the inverse dynamics computation.
no code implementations • 18 Jan 2024 • Koki Yamane, Sho Sakaino, Toshiaki Tsuji
However, these approaches face a critical challenge when processing data from multiple modalities, inadvertently ignoring data with a lower correlation to the desired output, especially when using short sampling periods.
no code implementations • 7 Dec 2021 • Yuya Nogi, Sho Sakaino, Toshiaki Tsuji
In this paper, we propose an external force estimation method based on the Mel spectrogram of the force obtained from a force sensor.
no code implementations • 22 Nov 2021 • Masahiro Aita, Keito Sugawara, Sho Sakaino, Toshiaki Tsuji
By combining two separately trained VAE models in a hierarchical structure, it is possible to generate trajectories with high reproducibility for both local and global features.
no code implementations • 11 Mar 2021 • Yuki Saigusa, Ayumu Sasagawa, Sho Sakaino, Toshiaki Tsuji
In this paper, we propose a variable speed motion generation method for multiple motions.
Imitation Learning Robotics
no code implementations • 12 Nov 2020 • Ayumu Sasagawa, Sho Sakaino, Toshiaki Tsuji
Owing to the structure and autoregressive learning of the proposed model, the proposed method can generate the desirable motion for successful tasks and have a high generalization ability for environmental changes.
no code implementations • 27 Feb 2020 • Masahide Oikawa, Kyo Kutsuzawa, Sho Sakaino, Toshiaki Tsuji
In this study, we propose a methodology that uses reinforcement learning (RL) to achieve high performance in robots for the execution of assembly tasks that require precise contact with objects without causing damage.