Mergen Ulziibayar

NYU Tandon School of Engineering. Mechanical engineering, robotics, and validation. Research assistant at the CUSP IgNYte Lab. New York, NY.

I am a mechanical engineering student at NYU Tandon working across robotics, mechanisms, and research tooling. I build the test that decides the design: simulation, CAD, software, and physical prototypes are connected through explicit acceptance criteria, then the result is used to revise the design or stop it before the expensive mistake.

My projects include vehicle modeling and a LiDAR-mount review for an F1TENTH-scale car, a simulation study of soft-actuator recalibration, and an outdoor air-quality sensor enclosure deployed with the CUSP IgNYte Lab. Every case study states its evidence level: a simulation is a prediction, a synthetic test validates tooling, and a prototype test supports only the condition that was measured.

I also fabricate with NYU Formula SAE and the RoboMaster UltraViolet team: tube fit-up fixtures, weld preparation, and launcher packaging. My CV has the one-page version.

autonomous-racing-systems: A simulation and mechanical-design study that exposed an unsupported mast-frequency requirement during drivetrain review.
autonomous-racing-systemsConverged FEA; physical test pending

news

Oct 4, 2026Multirotor recovery moved to the Crazyflie family. The sourced parameter register separates firmware defaults from physical limits; replay against public NanoBench flights comes next. The earlier V995 fixture is retired. Study roadmap.
Oct 2, 2026Soft-actuator study: v1.4 withdrawn before release. At matched recalibration cost, a cycle-count clock gives the same schedule and error as the P-V trigger on all 20 simulated actuators. The code and the audit are kept. Withdrawal note.
Sep 29, 2026uav-failsafe-composition compared its Python failsafe model against PX4's real Failsafe class on identical input sequences. The first run disagreed by exactly one update period, which exposed an ordering bug in the model; after the fix, all 11 sequences agree update for update. Repository.
Sep 29, 2026The LiDAR-mast assembly was built unattended on the SOLIDWORKS host and accepted against an independent CadQuery oracle: placement, per-part volumes, a known move-and-restore test, and a STEP round trip all match. Case study.
Sep 27, 2026The micro-UAV program fixed its physical platform to a stock Veeniix V995, with a separate whole-aircraft fixture register and a parametric load-cell cradle; the gates derived for the earlier designed vehicle were retired rather than carried over. Case study.
Sep 26, 2026The engineering-audit CAD loop solved its first part: a 6061-T6 mounting plate in tension, meshed at four refinements to 73,935 elements, whose peak stress at the hole sits within 1.2% of the Heywood and Howland closed-form value. Case study.
Sep 26, 2026Baseline sensor-enclosure geometry became parametric CAD, accepted against an independent closed-form oracle on volume, bounding box, solid count and STEP re-import, with a parameter re-drive recovering the predicted volume change exactly. Case study.
Sep 25, 2026A three-case modal study on the LiDAR mast showed most of the 13.5% hand-versus-FEA gap came from attaching the tip mass to a single wall node; a coupled, centred mass sits 2.8% from the hand model. Case study.
Sep 22, 2026Evidence base published for the senior design project on a compliant assistive upper body: 205 annotated sources, a requirements table where every number is cited or marked inferred, and three findings that changed the project's shape. Case study.
Sep 19, 2026Survivable-set study added to the micro-UAV program: it ranks whole design packages by the release conditions they recover from, with exact bounds and a registered drop-test prediction. Preregistration.
Sep 18, 2026A CAD build-and-validate loop joined the engineering-audit verifier: parametric geometry is built on a SOLIDWORKS host, checked against a closed-form volume oracle, and exported to STEP for FEA. Loop contract.
Sep 16, 2026Observability program for the soft-actuator study reported three limits on the health indicator: dispersion breaks its invariance in value, post-onset life estimates are not precise enough to report, and a transferred estimator misses its bar on short-lived units. Case study.
Sep 9, 2026Adaptive-Gripper repository created: a manual two-finger gripper with interchangeable rigid and compliant printed fingers, and a preregistered 60-trial pickup comparison. Repository.
Sep 9, 2026Parametric CAD generators with geometry checks in CI added for the LiDAR mast and the micro-UAV motor envelope. Mast generator.
Sep 6, 2026Soft-actuator study: a methods correction to the archived v1.3 manuscript was filed and publication placed on hold pending author review. Correction notice.
Sep 3, 2026Two research-design repositories created: segmented-shroud-aeromechanics (seam and clearance effects on rotor-shroud performance) and uav-failsafe-composition (native UAV recovery after command loss). Both are measurement-first designs with no results yet. Shroud repository.
Aug 31, 2026Soft-actuator recalibration study: preprint v1.3 published as a frozen GitHub release. Release.
Aug 10, 2026informal-road-mapping repository created with the Sentinel-2 corridor extractor and a known-truth synthetic demonstration (5/5 scenes). Case study.
Aug 6, 2026autonomous-racing-systems published on GitHub: bicycle-model identification checked on held-out telemetry from the same simulator model, pure pursuit / LQR / MPC / EKF comparison, a ROS 2 telemetry bridge, and a LiDAR-mast analysis with a 285.5 Hz FEA prediction and a static-compliance protocol. A later drivetrain review reopened the mounting requirement. Case study.
Aug 5, 2026engineering-audit benchmark results documented: 17 of 17 committed cases matched their annotated failure modes. Benchmark report.
Jul 17, 2026Micro-UAV: measured-authority gate frozen. Static thrust on the propulsion bench, not simulation, decides whether fabrication continues. Gate contract.
Jul 8, 2026Soft-actuator study: preprint v1.3 tagged with the P-V loop-area health indicator and the recalibration-policy comparison. Case study.
Jul 2, 2026engineering-audit v1.0.0: batch evaluation CI gate, stated limitations, and claim discipline for the LLM calculation verifier. Repository.
Jun 25, 2026Micro-UAV: closed-loop 6-DoF release-to-stabilization simulation and recovery-envelope figure committed. Enclosure: first-order solar self-heating model added the same day. Case study.
Jun 24, 2026Soft-actuator study: P-V loop-area indicator and triggered-recalibration study completed on a 2,000-trace actuator-split dataset; preprint draft v1 written. Repository.
Jun 20, 2026Micro-UAV paper design: 135.7 g estimated mass and 3.59 calculated thrust-to-weight after a 36.8 A current screen rejected the 450 mAh pack. This aircraft was never built. Repository.
Jun 18, 2026soft-actuator-recalibration repository created: pneumatic, fatigue, and sensing simulators for pressure-only soft-gripper proprioception. Repository.
Jun 12, 2026sensor-enclosure-thermal-design repository created with a 26-source literature matrix, PI deliverables, and manuscript v1 for the outdoor air-quality enclosure. Repository.
Jun 1, 2026engineering-audit v0.1.0: minimal synthetic benchmark verifier that recomputes mechanics independently of the model's answer. Repository.
May 27, 2026engineering-audit repository created: a CLI and benchmark for checking LLM-generated mechanical-engineering calculations. Case study.
May 12, 2026Nomad repository created: a Next.js app that maps an engineering query to ranked papers, researchers, and project directions from OpenAlex. Case study.
May 6, 2026multirotor-recovery-dynamics repository created for the simulated micro-quad release-recovery study. Repository.
Apr 20, 2026Air-quality enclosure placed outdoors for a provisional 22-day unattended run (95.7% upload success, no recorded brownouts in the window; dates inferred from the log's temperature signature). Case study.
Feb 2026Joined NYU Formula SAE (suspension and steering manufacturing) and the NYU RoboMaster UltraViolet gimbal sub-team. Launcher case study.
Dec 2025Started as an undergraduate research assistant at the NYU CUSP IgNYte Lab, owning the outdoor air-quality sensor enclosure. Case study.
Apr 2025V-Aid Climbing tactile pin array showcased as a functional prototype after four printed generations; supported by a $2,000 NYU Prototyping Fund award. Case study.
Jan 2025V-Aid Climbing started: a 12 x 12 tactile pin array that converts a climbing-route image into a raised pattern for blind climbers. Case study.

selected projects

soft-actuator-recalibration

A 2,000-trace, actuator-split simulation study of P-V loop area as a recalibration trigger, and the audit that stopped its paper: at matched cost, a cycle-count clock ties the trigger exactly.

2026 · Actuator-split simulation

autonomous-racing-systems

A simulation and mechanical-design study that exposed an unsupported mast-frequency requirement during drivetrain review.

2026 · Converged FEA; physical test pending

V-Aid Climbing Tactile Pin Array

A 12 x 12 tactile pin array that converts a climbing-route image into a raised pattern a blind climber can read by hand.

2025 · Physical functional testing

All case studies, grouped by evidence type →