CV PDF

SIMEON NEDELCHEV

simkaned@gmail.com · s.nedelchev@innopolis.university
simeon-ned.github.io · Presentation · github.com/simeon-ned · Google Scholar

Professional Experience

Researcher & Senior RL Engineer Feb 2026 – Present
Institute of Artificial Intelligence

Developing learning-based whole-body control, data-driven motion generation, and estimation methods for humanoid and articulated robotic systems. Building simulation pipelines, motion libraries, and deployable control stacks in collaboration with academic and industry partners.

Senior Optimal Control & RL Engineer Aug 2023 – Dec 2025
Sber Robotics Center, Sber

Developer of the RL whole-body control (WBC) and locomotion policies demonstrated on the Green humanoid at AIJ Conference 2025. Work centered on control, with modeling and MuJoCo-based simulation and training/evaluation pipelines; partial system identification (mainly motor and actuator parameters) (video).

Senior Lecturer Apr 2022 – Present
Institute for Robotics and Computer Vision, Innopolis University

Advanced courses in robotics, mathematical modeling, motion planning, applied and nonlinear control. Curriculum development and student mentoring in dynamics, simulation, and embodied AI.

Junior Researcher Sep 2019 – Sep 2023
Center for Technologies in Robotics and Mechatronics Components, Innopolis University

Mathematical modeling, identification, and control of twisted-string actuators; optimal nonlinear control and trajectory optimization; mechanical and electrical prototyping of novel robotic systems.

Research Assistant Mar 2017 – Jun 2019
BioRobotics Laboratory, Korea University of Technology and Education

Analysis, prototyping, and nonlinear control of twisted-string actuators (TSA); hardware implementation of advanced control algorithms.

Education

Ph.D. studies in Computer Science (Robotics track), completed Jul 2023
Innopolis University · GPA 4.87/5 (97/100). Institute of Robotics and Computer Vision.

Thesis: Dynamic parameter estimation and adaptive control over robotic systems: generalized momentum approach.

M.Sc. in Mechanical Engineering Feb 2019
Korea University of Technology and Education · GPA 4.45/4.5 (98/100)

Thesis: Design of Robotic Gripper with Constant Transmission Ratio Based on Twisted String Actuator.

M.Sc. in Robotics & Mechatronics Jun 2018
Moscow State Technological University STANKIN · GPA 4.9/5 (98/100)

Thesis: The Dynamics of Controlled Motion of Industrial Robots.

B.Sc. in Robotics Jun 2016
Moscow State Technological University STANKIN · GPA 4.35/5 (87/100)

Thesis: Development and research of dynamic model of robot manipulator.

Selected Research & Software Projects

WBC-Mjlab: Whole Body Control in MuJoCo Lab (web demo) — shared MDP for universal whole-body motion tracking in mjlab: multi-clip training, paper-specific task presets (ZEST, BeyondMimic-style RSI, etc.), motion-data pipeline, and sim-to-real export for humanoids.

PSM: Predictive Style Matching for Natural Locomotion (project page) — offline predictor maps lower-body state and velocity commands to upper-body and gait targets during RL training; deployed on Unitree G1 with task-only inference at run time. arXiv:2606.07083 (2026).

LocoGen: Command-Conditioned Locomotion Generation — combines learned motion matching and flow matching for online full-body locomotion on legged robots. G1-first but robot-agnostic; MuJoCo-grounded state, velocity commands (extensible to waypoints and trajectories), and interactive Viser control with seamless mode transitions. Inference at ~800 FPS (LMM) and ~150 FPS (flow); PoC training from a handful of LAFAN clips. Walking and tracking validated with BeyondMimic / ZEST in simulation and on Unitree G1 hardware. Code forthcoming.

Locomotion and Whole Body Control for Sber Green (AIJ 2025 presentation) — developer of the RL locomotion and whole-body control policies demonstrated on the Green humanoid at AIJ 2025: improved naturalness and human-likeness of locomotion; RL-based whole-body control trained on synthetically generated reference trajectories (without acquiring mocap datasets or retargeting) and unified tracking for teleoperation, dance, and zero-shot skills.

Open Source Contributions

Pinocchio — contributor to the rigid-body dynamics library (inertia parametrizations and physical consistency).

Pink — contributor to differential inverse kinematics for Pinocchio (tasks, constraints, and examples).

Mink — contributed EqualityConstraintTask for closed-chain mechanisms; humanoid and manipulation examples in the MuJoCo port of Pink.

GMR: General Motion Retargeting — contributor (SOMA BVH format and multi-robot configs).

mujoco-sysid — contributor; system identification in MuJoCo and MJX (parameter estimation, dynamics regressors).

MJINX: Differentiable GPU-accelerated Numerical IK — co-author (JAX and MuJoCo MJX).

Publications

Preprints
Journals
Conference papers

Teaching

Innopolis University — Senior Lecturer (2022–present), Teaching Assistant (2019–2022)

FORC: Fundamentals of Robot Control — introductory course on state-space modeling, stability, linear and nonlinear control, and feedback linearization for robotic systems (lectures and Colab/Python labs).

MCP: Modern Control Paradigms — course materials on modern control via numerical methods and convex optimization, with interactive notebooks.

Technical Skills

Programming: Python, C/C++, Bash
Robotics & ML: MuJoCo, mjlab, Isaac Sim, Isaac Lab, RL/WBC training pipelines, system identification
Embedded: STM32, ARM, ESP32, RP2040 (MicroPython, HAL, FreeRTOS, mbed)
Tools: Linux, LaTeX, Docker, uv, pixi

Honors

Research Interests

Nonlinear and learning-based control; whole-body control and humanoid locomotion; system identification and estimation; differentiable simulation; data-driven motion; novel actuators; convex optimization for control.

Download full CV (PDF)