Built real-time 4D reconstruction for continuous in-process monitoring on robotic additive manufacturing—geometry that updates as the part is built.
Open to research & engineering roles · Melbourne
Subash Gautam
Research Engineer · Computer Vision, AI & Robotics
I ship real-time vision, geometric digital twins, and AI pipelines on robotic cells—from additive manufacturing QA to soft robotics and architectural floor-plan intelligence.
- 5+ yrs research
- 4 RCIM papers
- 1 patent
- PhD RMIT × CSIRO
What I bring to your team
Production-minded research engineer — in-process QA, not post-mortem inspection.
- Real-time 4D reconstruction & geometric digital twins on industrial robotic cells
- Soft-robotics digital twins for simulation, sensing, and control-oriented modelling
- AI vision pipelines that turn architectural floor plans into structured, queryable models
- Patent co-inventor · peer-reviewed RCIM author · ROS2/C++ lab lead at RMIT
Professional Summary
PhD Research Engineer (Mechatronics) with 5+ years building vision, AI, and digital-twin systems for high-rate robotic manufacturing and applied robotics research. At CSIRO I delivered in-process 4D reconstruction, geometric digital twins (gDT-AM), ML defect monitoring, and multi-sensor calibration methods now published in Robotics and Computer-Integrated Manufacturing and protected by provisional patent 2025904901. At RMIT I extend that work into soft-robotics digital twins and AI floor-plan vision pipelines, while leading robotics labs in C++, Python, ROS2, and computer vision. I translate research into deployable pipelines—clear documentation, measurable validation, and engineering teams can actually run.
Impact Highlights
Designed geometric digital twin (gDT-AM) and 3D-DM² deviation mapping for shape monitoring and defect detection before scrap.
Developing soft-robotics digital twins linking continuum robot geometry, sensing, and simulation for research and control workflows.
Built AI floor-plan vision pipelines—automated extraction from 2D architectural drawings into structured digital models for downstream design tools.
Co-inventor, provisional patent 2025904901 — in-process manufacturing monitoring systems.
First-author and co-author publications in Robotics and Computer-Integrated Manufacturing; closed-form multi-profiler hand–eye calibration adopted for faster cell deployment.
Selected Work
Evidence of research depth and engineering delivery — click each project for detail.
Industry & applied research
Digital Twin · 4D Reconstruction · Defect Detection
Impact: Live deposition geometry vs design on cold-spray robotic AM—catching deviation and defects during the build, not after. Includes gDT-AM shape monitoring, 3D-DM² temporal defect tracking, and ML segmentation on production-relevant data rates.
3D-DM² paper →Soft robotics · Digital twin · Simulation
Soft Robotics · Digital Twin · RMIT
Impact: Digital twin framework for soft/continuum robotic systems—linking deformable geometry, sensor feedback, and simulation-ready models for design iteration and control-oriented validation. Bridges materials-aware shape change with vision and robotics tooling used in advanced manufacturing research.
Floor plans · VLM · Document AI
Computer Vision · AI · Architectural Digitisation
Impact: End-to-end vision pipeline converting 2D architectural floor plans into structured digital models—wall geometry, rooms, and symbols extracted for downstream BIM/design workflows. Combines classical image processing with modern VLM/ML components for robust parsing of real-world drawing sets.
Closed-form · Two-stage · Multi-sensor
Robotics · Calibration · RCIM 2025
Impact: Streamlined robotic hand–eye calibration of multiple 2D laser profilers—rapid closed-form two-stage method that cuts calibration time and improves accuracy for multi-sensor vision cells.
Read paper →
RMIT · ROS2 · Teaching
Impact: Lead robotics laboratory sessions—mentoring students in C++, Python, ROS2, and computer vision through hypothesis-driven prototyping and system integration. Parallel research on vision/AI pipelines for architectural digitisation and soft robotics.
Earlier engineering foundations
Robotics · Competition · Nepal
Robot "Nepali Keto" navigated hurdles, sand, pebbles, mud, and water — from Terai to Everest — and raised the flag 3.9m high in under two minutes.
Autonomous · PID · Arduino
Five-sensor line array with obstacle detection and PID-controlled movement on Arduino Mega — foundation in autonomous navigation.
UAV · Surveillance · Logistics
Cloud-based IP camera for real-time aerial monitoring with precision payload delivery for disaster response and medical supplies.
Solar · Embedded · IEEE
Incremental conductance MPPT validated in hardware and MATLAB — 33% PV output boost in hardware. Published at IEEE ICRERA 2016.
IEEE paper →
Renewable Energy · Hydropower
Gravitational water vortex power plant pilot for rural electrification — simple, replicable design for remote villages in Nepal.
Networking · IPv6 · La Trobe
Dual-stack IPv6 deployment across campus network — awarded Outstanding Implementation at La Trobe Engineering Showcase 2019.
Experience
- Lead soft-robotics digital twin research—deformable geometry, sensing, and simulation workflows for continuum robotic systems.
- Built AI floor-plan vision pipelines that parse architectural drawings into structured digital models for design/BIM downstream use.
- Lead robotics labs mentoring students in C++, Python, ROS2, and computer vision through production-style integration projects.
- Designed real-time 4D reconstruction pipeline for robotic additive manufacturing.
- Developed in-process 3D deviation mapping and defect detection with temporal modelling.
- Created geometric digital twin framework combining physics-based and data-driven reconstruction.
- Developed closed-form two-stage hand–eye calibration for multiple 2D laser profilers.
- Co-inventor on provisional patent 2025904901; published 4 peer-reviewed papers.
- Taught C, C++, and Python — algorithms, data structures, and OOP to undergraduates.
- Delivered hands-on labs for embedded systems and automation applications.
Core Skills
Image Processing & Reconstruction
Software & AI
Robotics & Manufacturing
Research & Leadership
Publications
Geometric Digital Twin for Robotic Additive Manufacturing (gDT-AM)
In-process 3D Deviation Mapping and Defect Monitoring (3D-DM²)
arxiv.org/pdf/2511.05604Streamlined robotic hand–eye calibration of multiple 2D-profilers
doi.org/10.1016/j.rcim.2025.102984In-process 4D reconstruction in robotic additive manufacturing
doi.org/10.1016/j.rcim.2024.102784Maximum power point tracker with solar prioritizer in photovoltaic application
doi.org/10.1109/ICRERA.2016.7884494Education
PhD — Mechanical, Mechatronics & Manufacturing
RMIT University, Melbourne — Real-time 4D reconstruction, hand–eye calibration, geometric defect segmentation using ML.
Master of Information & Communication Technology
La Trobe University — 88.25% WAM. Golden Key International Honour Society (top 15%).
BEng — Electronics & Communication Engineering
Purbanchal University, Nepal — Final year project published in IEEE Xplore (ICRERA 2017).
Honours & Awards
NRNA Australia Academic Excellence Award
Outstanding performance during master's degree.
Outstanding IPv6 Campus Deployment
La Trobe University Engineering & IT Showcase.
Best Project Award — MPPT Solar System
Department of ECE, Acme Engineering College, Purbanchal University.
National Robotics Competition Winner
NCIT ROBO Drive 2013 — National College of Information Technology, Nepal.
Let's build the next vision or robotics system together.
Open to research engineer, computer vision, robotics, and digital-twin roles in Melbourne and remote-friendly teams.