SHEET 01 / 07 · ROBOTICS / MECHATRONICS · ISO 2768-m

Mechanisms, from concept to tolerance.

Mechanical engineering student working across RL-based control, machining dynamics & tool-condition monitoring, and validated simulation studies.

20 MSS VERIFIED MANUSCRIPTS · 2026

01 → ASSEMBLIES 03 → PAPERS

FIG. 1 — GRASHOF CRANK–ROCKER, COUPLER CURVE SHOWN DASHED · DIMS IN mm

Assemblies

RLC-001 RLCONTROLSROBOTICS

Cart-Pole Swing-Up & Balance, Sim-to-Real Ready

MATL CART-POLE · TOOLS Python Â· Gymnasium Â· SB3 PPO

PPO balances 100% over 20 episodes, matching LQR; energy-pump swing-up catches the pole at 6.4 s. Trained in 1.5M steps, ~6 min CPU.

RLC-002 RLCONTROLSROBOTICS

Furuta Pendulum, End-to-End PPO Swing-Up

MATL ROTARY PENDULUM · TOOLS Python Â· Gymnasium Â· SB3 PPO

End-to-end PPO swings up 20/20 and reaches upright in 0.71 s, 3x faster than a hand-tuned energy controller that managed 7/10 after five tuning rounds.

RLC-003 RLCONTROLSROBOTICS

Reaction-Wheel Pendulum on a Momentum Budget

MATL REACTION-WHEEL PENDULUM · TOOLS Python Â· Gymnasium Â· SB3 PPO

Energy swing-up 10/10; against a +0.8 rad/s body poke, PPO recovers 12/20 where the pure LQR loses all 20. Both balance 20/20 near upright.

RLC-004 RLCONTROLSROBOTICS

Two-Wheel Balancing Robot, Velocity Tracking

MATL TWO-WHEEL ROBOT · TOOLS Python Â· Gymnasium Â· SB3 PPO

PPO tracks velocity to 0.123 m/s mean error vs the LQR 0.163 m/s, ~25% tighter; the LQR is marginally more robust (20/20 vs 19/20).

RLC-005 RLCONTROLSROBOTICS

Planar Ballbot, Balance & Station Keeping

MATL BALLBOT (PLANAR) · TOOLS Python Â· Gymnasium Â· SB3 PPO

Both PPO and LQR survive 20/20 and recover every shove; LQR holds station tighter (0.235 m vs 0.338 m), PPO settles faster in the demo (2.3 s vs 3.0 s).

RLC-006 RLCONTROLSHW

Magnetic Levitation, LQR-Primed RL vs Magnet Drift

MATL MAGLEV 1-DOF · TOOLS Python Â· Gymnasium Â· SB3 PPO Â· BC

BC-from-LQR warm start plus PPO holds 10/10 across Â±30% magnet-constant drift where the LQR drops to 5/10; PPO-from-scratch failure documented.

HW-001 HWCONTROLS

Arduino PID Motor Rig: Sim-First Design

MATL DC MOTOR + ENCODER · TOOLS Python · Arduino · L298N

Discrete PID tuned in Python (Kp12/Ki22/Kd1): 13% overshoot, 1.45 s settle, 0 steady-state error — firmware mirrors the sim 1:1.

HW-002 HWSIM

Folding Parallelogram Staircase

MATL STEEL 4-BAR LINKAGE · TOOLS FreeCAD · Python macro

Reverse-engineered retractable stair: parallel rails keep 6 treads level through a 900 mm rise; parametric FreeCAD macro poses the full fold live.

Experience

2026 — PRESENT RESEARCH

Independent Research Portfolio

Self-directed (Daejeon target)

20 verified manuscripts across machining dynamics, tool-condition monitoring, and validated-analysis audits; all numbers reproduced from code.
2026 RESEARCH

RL Control Series

Self-directed

Seven PPO-vs-LQR control studies (cart-pole through maglev and chatter suppression), each with a custom Gymnasium env and live demo.
2025 — 2026 WORK

Manufacturing Intern (CNC)

⚙ Factory, Urgench

Hands-on CNC machining, CAM, drilling and metal cutting on the shop floor; sourced the cutting-parameter study access.
2023 — 2027 EDU

B.S. Mechanical Engineering

New Uzbekistan University

Focus on robotics, controls and manufacturing; graduating May 2027.

3D Viewports

Orbit with mouse / touch. ⚙ Replace public/models/placeholder.glb with your own SolidWorks / Fusion glTF exports — see README.

BRACKET, BOSS-MOUNT — placeholder.glb

FOLDING STAIR, DEPLOYED — fold-stair.glb

GENEVA DRIVE, 4-SLOT — DESIGN DEMO

Mechanism Reference

Fundamental mechanisms, drawn true to their kinematics. Hover, focus, or drag any figure to crank it by hand.

Four-bar linkage Grashof: s + l ≤ p + q

Slider-crank x = r cosθ + √(L² − r²sin²θ)

Geneva drive Intermittent — 4-slot 90° index

Cam + follower SHM rise: s = ½h(1 − cos φ)

Scotch yoke Pure SHM: x = r cos ωt

Peaucellier inversor Exact straight line — OP·OQ = k²−m²

Watt governor cos α = g / (Ω² L)

Reuleaux triangle Constant width — A = ½(π−√3)w²

Papers & Reports

ITEM	TITLE	TYPE	YEAR	FILE
001	A Review of Low-Cost Tool-Condition Monitoring Methods for CNC Machine Tools	Review	2026	DOCX
002	A Review of Process-Parameter Optimization in Fused Deposition Modeling: Effects on the Mechanical Properties of Polymer Parts	Review	2026	MD

Notes

Mechanical engineering student at New Uzbekistan University with real shop-floor experience (CNC, CAM, metal cutting — factory internship in Urgench) and a simulation-first research portfolio: RL control, machining dynamics, and tool-condition monitoring.
Seeking MS positions & research assistantships in Korea (robotics / smart manufacturing) [A]
Tashkent, UZ · Available from May 2027 [B]
UNLESS OTHERWISE SPECIFIED: TOLERANCES PER ISO 2768-m · SURFACES Ra 1.6

GENERAL NOTES — GOVERNING RELATIONS

01 F = m·a Newton II
02 τ = r × F Torque
03 M/I = σ/y = E/R Euler–Bernoulli beam
04 P_cr = π²EI / (K·L)² Euler buckling
05 Re = ρvL / μ Reynolds no.
06 q″ = −k·∇T Fourier conduction

Capabilities

DOMAIN	TOOL	LEVEL
CAD	SolidWorks · Fusion 360	●●●○ WORKING
MFG	CNC machining · 3D printing	●●●● PRIMARY
CODE	Python · NumPy · SciPy	●●●● PRIMARY
ML	Gymnasium · SB3 (PPO) · scikit-learn	●●●○ WORKING
CONTROL	LQR · PID · energy methods	●●●○ WORKING
SIM	FE & stability lobes (from scratch)	●●●○ WORKING
DATA	pandas · matplotlib	●●●○ WORKING
HW	Arduino · sensors	●●○○ FAMILIAR

Contact

FIG. — NEWTON'S CRADLE · Σp = const · Σ½mv² = const · CLICK TO RELEASE

REVISIONS

REV	DATE	DESCRIPTION	BY
A	2026-07-11	Initial issue	O.M.
B	2026-07-12	Real projects & document register issued	O.M.
C	2026-07-12	Live plant drawings; register trimmed to reviews	O.M.
D	2026-07-12	Assemblies pared to RL series + PID rig	O.M.

SECTION Z—Z · UNDOCUMENTED

Strandbeest

A coupler-curve walking linkage — in the tradition of Chebyshev's 1878 plantigrade machine and Jansen's Strandbeest. Four crank-rocker legs share one crankshaft at staggered phase; each foot traces a curve with a flat stance stroke, so the frame walks. You found the undocumented sheet. Well played, engineer.