Multisensory Learning Framework for Robot Drumming
This work addresses the challenge of robot manipulation tasks, specifically drumming, by enabling multisensory learning, though it appears incremental as it builds on existing sensorimotor learning advancements.
The paper tackles the problem of learning non-linear sensorimotor mappings for a humanoid drumming robot by developing an open-source framework for collecting large-scale, time-synchronized synthetic data from disparate sensory modalities like audio, video, and proprioception, resulting in the generation of novel motion sequences from desired audio data using cross-modal correspondences.
The hype about sensorimotor learning is currently reaching high fever, thanks to the latest advancement in deep learning. In this paper, we present an open-source framework for collecting large-scale, time-synchronised synthetic data from highly disparate sensory modalities, such as audio, video, and proprioception, for learning robot manipulation tasks. We demonstrate the learning of non-linear sensorimotor mappings for a humanoid drumming robot that generates novel motion sequences from desired audio data using cross-modal correspondences. We evaluate our system through the quality of its cross-modal retrieval, for generating suitable motion sequences to match desired unseen audio or video sequences.