Nodoka Shibasaki

About Me

I am a fourth-year Electrical and Computer Engineering student at Northeastern University. Under the supervision of Professor Taskin Padir at RIVeR Lab, I research robot manipulation.

I am fortunate to be researching warehouse manipulation at the Innovation Lab at Amazon Robotics and building new features for SONiC as a Software Engineering Intern.

In the past, I gained valuable industry research experience as a Research Intern at OMRON SINIC X in Tokyo where I researched contact-rich manipulation for soft robotics. During my undergraduate studies, I also had the opportunity to develop autonomous multi-modal locomotion robots under Professor Alireza Ramezani, and uncertainty-aware exoskeletons under Professor Max Shepherd.

Selected Projects

* Highlighted are ongoing projects.

Cool Project

May 2026 - Present | Amazon Robotics

Researching robot manipulation for warehouse applications at the Innovation Lab. Details are confidential for now, so stay tuned!

Image from A. Allison et al.

HASHI

Jan 2026 - Present | RIVeR Lab

Investigates imitation learning for contact-rich tasks using an xArm robot with a custom chopsticks gripper. The project explores mapping human teleoperation data to dexterous manipulation policies.

Tactile Memory with Soft Robots

Jul 2025 – Dec 2025 | OMRON SINIC X

TaMeSo-bot uses transformer-based learning to process tactile, force-torque, and proprioceptive signals for retrieval-based manipulation. Achieved 88.5% and 62% success rates under seen and unseen peg-in-hole conditions (varying shapes, sizes, textures, orientations).

Published in IEEE RA-L 2026 · arXiv

Active vs. Passive Control with FT feedback

Jul 2025 – Dec 2025 | OMRON SINIC X

Comparative study of passive compliance (soft wrist, position control) versus active compliance (Forward Dynamics Compliance Controller, rigid wrist) on a UR5e using SAC reinforcement learning with force-torque feedback only.

Multi-Modal Locomotion Robot

Jan 2025 – July 2025 | Silicon Synapses Lab

M4 platform uses a reproducible sim-to-real pipeline built with ROS 2, ArduPilot, and Docker for multi-modal locomotion. Reduced deployment time by 40% and improved energy efficiency by 18% across aerial and ground modes.

Ankle Exoskeleton

Sep 2024 – Jun 2025 | Shepherd Lab

Modular ankle exoskeleton uses flexible CAD system and integrated sensor pipelines for rapid prototyping and biomechanical testing. Accelerated hardware-control iteration cycles, enabling foundation for ML-driven gait assistance.

Hip Exoskeleton

Jun 2024 – Feb 2025 | Shepherd Lab

Hip exoskeleton uses optimized power system with RL-based control policy for adaptive gait assistance under variable human movement. Reduced metabolic cost by 38.5% and improved stability, enabling extended human-in-the-loop trials.

Blog

Powered by Jekyll and Minimal Light theme.