Papers
32
Total Citations
452
H-Index
12
About
Hiroki Mori is a robotics and machine learning researcher whose work sits at the intersection of deep learning, embodied cognition, and robot manipulation. His research focuses on enabling robots to perform complex, real-world tasks through predictive learning, multisensory integration, and reinforcement learning — areas where he has made substantial and varied contributions. Mori's most impactful work (81 citations) demonstrates how embodied predictive models can support multitask learning for whole-body robot control, such as door opening and entry — tasks central to deploying robots in everyday human environments. He has tackled notoriously difficult manipulation challenges, including in-air rope knotting with dual-arm systems, flexible object manipulation using vision and tactility, and peg-in-hole tasks under variable real-world conditions. His earlier work on generating composite tasks from discrete learned behaviors using deep neural networks laid important groundwork for generalizable robot learning. Beyond manipulation, Mori has explored the cognitive foundations of robotics, investigating agency perception in spiking neural networks and grounding linguistic representations in robot actions through embodied word embeddings. His work on chaotic itinerancy further bridges computational neuroscience and adaptive robot behavior. With over 300 cumulative citations across a diverse portfolio, Mori stands as a compelling figure advancing robots that sense, adapt, and reason across complex, contact-rich environments.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5
- 6In-air Knotting of Rope using Dual-Arm Robot based on Deep Learning26 citations · 2021
- 7
- 8Embodying Pre-Trained Word Embeddings Through Robot Actions19 citations · 2021
- 9
- 10