Birmingham, UK · MSc Robotics 2026

Piero Flores

Mechatronics & robotics engineer. I build machines that measure to 25 µm, arms that place to 0.0002 mm, and robots that navigate hills on their own.

View CVGet in touch

01Featured work

Laser Triangulation Metrology Machine (OD/ID)

2026

MSc Thesis · Individual project · University of Birmingham

Non-contact optical inspection machine measuring OD, ID, roundness and wall thickness of machined steel cylinders (bores from 127 mm, OD to 203 mm, length 610 mm) with automated pass/fail QC reporting.

  • GUM/NIST uncertainty budget targeting expanded uncertainty U(k=2) ≤ 25 µm against a 254 µm tolerance (U/T ≤ 10%), validated with Gage R&R
  • 160-item engineering BOM and 43-part CAD assembly with collision resolution; direct vendor engagement for laser profilers, encoders and linear drives
  • Architecture aligned to a ~US$25,000 BOM for affordable deployment in machine shops

PythonSolidWorksLaser profilersGUM/NISTGage R&RFull design withheld pending IP review — public foundations & uncertainty demos in the repo.

3D plot of the QArm end-effector trajectory during a sorting cycle, showing pick-and-place paths to three color-coded fruit baskets.

Autonomous Fruit Sorting — Quanser QArm + RealSense D415

2025–2026

Applied Robotics · Team of 4 · University of Birmingham — my role: camera-arm calibration, fruit-detector tuning, voice control, GUI concurrency (70/185 commits)

End-to-end pick-and-place on a 4-DOF QArm with an RGB-D camera, sorting 14 fruits into 3 baskets in autonomous and teleoperated modes.

  • Analytical inverse kinematics with Newton-Raphson refinement: 0.0002 mm max round-trip error across four solution branches
  • 13-state finite-state-machine controller with cubic rest-to-rest spline trajectories and Z-height collision safeguards
  • Hand-eye calibration via closed-form Umeyama SVD; HSV + circularity classifier with pixel-to-world back-projection

Python 3.13NumPyOpenCVQuanser SDK (HIL)MATLAB/Simulink

System architecture map of the Moose navigation stack: four color-coded workstreams (robot setup, EKF fusion, 3D mapping, slope-aware navigation) with labeled data handoffs between subtasks.

Autonomous Navigation for an All-Terrain Robot

2025–2026

Team of 4 · University of Birmingham — my role: 3D mapping & EKF sensor fusion (13/18 commits)

Mapping and perception for a Clearpath Moose 8-wheel grass-cutting robot on hilly terrain with slope hazard avoidance (>20°).

  • Wrote the Extended Kalman Filter — 6-state [x, y, z, roll, pitch, yaw] fusing GPS, IMU and compass — for drift-free localization on slopes (RMSE 4 mm)
  • OctoMap probabilistic occupancy (depth 12) with dynamic recentring, plus a 200×200 elevation grid (86.6% coverage) with slope/roughness traversability costs
  • LiDAR pipeline: range filtering (0.3–50 m), voxel downsampling (0.1 m), ICP scan matching over 4,735+ frames (800K+ points); integrated with the team's A* planner

C (2,600 lines)WebotsEKFOctoMapICP

FrED Factory Station 1: a white UFACTORY xArm 6 cobot with a Datalogic vision gripper over a conveyor, HMI panel and PLC-driven fixtures on a lab bench.

Automated Manufacturing Station — xArm 6 (with MIT)

2023

Team of 4 · Tecnológico de Monterrey in partnership with MIT

Semi-autonomous assembly station for MIT’s FrED device: cobot pick-and-place, PLC coordination, machine vision QC and a digital twin.

  • Siemens PLC (7 Ladder Logic networks): conveyor control, photoelectric sensors, robot–PLC bidirectional signaling, HMI status
  • Datalogic machine vision over a TCP/IP vision link for real-time coordinate transfer and camera-to-gripper offset compensation
  • 125.9 s measured phase cycle on real hardware (target ≤2 min)
  • Digital twin in Siemens Plant Simulation: 245 units, 41 TPH, 81.45% value-added; meets the ≤2 min/station target

PythonxArm SDKSiemens TIA PortalDatalogic visionPlant Simulation

All projects →