Legged locomotion
Closed-loop gait coordination, phase regulation, touchdown/liftoff events, and robust locomotion under disturbances.
BIO-INSPIRED ROBOTICS
Seokhyun Kim
Robotics researcher focused on legged locomotion, current-based proprioception, decentralized control, and bio-inspired robots that can move robustly in constrained environments.
4×
Independent leg actuation
33 gsmall quadruped
>90%stance/swing classification
About
I build robots that use simple hardware and intelligent feedback to move effectively.
I am a graduate researcher in Mechanical Engineering at Carnegie Mellon University. My work sits at the intersection of locomotion, sensing, control, and robot design, with a particular interest in biologically inspired coordination strategies for legged and articulated robots.
I am especially interested in systems where sensing and actuation are tightly constrained. Recent work explores using motor current as a source of proprioceptive information for centimeter-scale quadrupeds, reducing dependence on joint encoders and dedicated contact sensors.
Research directions
What I work on
Bio-inspired control
Decentralized coordination, oscillatory control, body–limb interaction, and animal-inspired strategies for whole-body motion.
Small-scale robotics
Designing sensing, actuation, electronics, and control around strict mass, volume, power, and computational constraints.
Selected projects
Robots & research
Replace the gray project panels with your own photos, CAD renders, videos, or paper figures.
Little Quadruped Robot
Current-based proprioception · Small-scale locomotion
Little Quadruped Robot: Encoder-free leg-state estimation for a 33 g quadruped
A fully untethered centimeter-scale quadruped with four independently actuated rotary legs, onboard computation, IMU sensing, and per-motor current measurement. A neural-network classifier estimates stance and swing from current signals, providing contact events for closed-loop gait coordination.
33 gmass
4independent legs
>90%state-classification accuracy
Cat-inspired Robot
↻
Aerial self-righting · Bio-inspired robotics
Cat-Inspired Robot
A cat-inspired robot combining an actuated 2-DoF spine and actuated tail for closed-loop aerial self-righting.
HUMRS
∿∿∿
Underwater robotics · State estimation
HUMRS underwater snake robot
Research on an articulated underwater robot using ROS and vision-based state estimation, with mechanical design considerations for waterproof operation.
QUADRUPED LOCOMOTION
Decentralized coordination · Animal-inspired control
Whole-body quadruped coordination
Research on emergent inter-limb coordination and trunk–limb interactions without prescribing complete foot trajectories.
Skills
Technical toolkit
Robotics & Control
Legged locomotion, feedback control, decentralized control, state estimation, system modeling
Software
Python, MATLAB, C/C++, ROS, MuJoCo, machine-learning training and deployment
Perception
VINS-Fusion, IMU-based sensing, current-based proprioception, vision-based state estimation
Hardware
CAD, 3D printing, PCB integration, embedded systems, motor drivers, sensor integration
Education
Academic background
2025–present
Carnegie Mellon University
M.S. student, Mechanical Engineering
2021–2025
Tohoku University
B.Eng., Electrical, Information and Physics Engineering
Contact
Interested in robotics, locomotion, or research collaboration?
Add your email, GitHub, Google Scholar, and LinkedIn links below.