Learning control Led Active now
Dyno Humanoid Robot Research and Development
Director of Robot Control and Software · Robot Mobility Expert
Projects
Funded research & engineering
Sixteen programmes, from a CubeSat in Hanoi to a humanoid that has to stay upright
Every programme below is one I was funded to work on, in the role the contract gave me — from core developer on Vietnam's first pico-satellite to technical leader of a nine-year national research centre for urban air mobility digital twins, and now director of robot control and software for a humanoid platform. Where a programme outlived my appointment, the card says so.
- Programmes
- 16
- Led
- 8
- Since
- 2009
- Papers
- 42
Dates on each card are the period I personally worked on the programme. Several grants run past that — the nine-year urban air mobility centre is funded to 2029 — and the timeline below draws the full funded span as a faint bar behind the solid one. Papers counts the 42 entries on the publications page that the cards link out to.
Programme timeline
All 16 programmes on one axis, oldest first. The solid bar is the period I worked on it; the faint bar behind it is the full funded span of the programme itself.
- F-1 pico-satellite 2009 – 2011
- Korea–Peru UAV cooperation 2011 – 2014
- Unmanned craft automatic control 2012
- Heritage 3D mapping by airship 2013
- Balloon-satellite platform 2013
- Naval ship survivability (ST-54) 2014 – 2015
- Advanced UAV reliability 2014
- GENESIS streaming platform 2015 – 2020
- Hybrid cloud technologies 2016 – 2019
- Korea–Vietnam disaster-tolerant cloud 2017 – 2018
- AI digital twin for UAM Jun 2020 – Jun 2025
- i-MeC2 framework Sep 2021 – Jun 2025
- Certified neural flight control 2022 – Jun 2025
- Future automobile R&D Mar 2022 – Jun 2025
- Helicopter avionics Jun 2025 – Mar 2026
- Dyno humanoid robot Mar 2026 – present
The programmes
Grouped into the four eras of the work, newest first. 8 led, 8 contributed to as staff or core researcher. Where the work produced papers, the identifiers link straight to the entry on the publications page.
Robotics and mobility2025 – now
Back in Vietnam and back on hardware. Humanoid robotics at VinDynamics, and the avionics function of a national rotorcraft programme before it.
Air mobility Led
Viettel Helicopter Development and Manufacture
Manager of Avionics System
The avionics function of a national rotorcraft programme: the technical vision for on-board computing, the hardware and software teams doing the early research on modern helicopter computing and communication, and the AI work on real-time flight data analytics and automatic control.
Programme runs to Dec 2027; my involvement ended with the Viettel appointment in Mar 2026.
- Develop the technical vision for the computing systems carried on a helicopter.
- Manage a team of hardware and software engineers running the initial research on computing and communication systems for a modern helicopter.
- Direct a team of AI engineers building models for real-time data analytics and for automatic control of the aircraft.
- Work directly with the board to run the programme across internal and external sources.
- Avionics architecture
- Real-time computing
- Data-link systems
- Flight data analytics
- Automatic flight control
AI digital twins for air mobility2020 – 2029
The Konkuk decade. A nine-year national research centre for UAM digital twins, a middle-career grant on resilient smart-city services, certified neural flight control, and a mobility talent programme run as applied R&D.
Air mobility Led Flagship programme
Artificial Intelligence Digital Twin Research for Smart Urban Air Mobility
Project Technical Leader · Lead Researcher
The flagship programme of the decade: an AI-driven digital twin ecosystem for urban air mobility, linking eVTOL vehicle dynamics to cloud and edge simulation, dependable platforms, blockchain and federated learning, and multi-agent AI — and the technical roadmap that held those work packages together.
Nine-year programme funded to May 2029; my involvement ended with the Konkuk appointment in Jun 2025.
- Lead the technical roadmap and research architecture for AI-driven UAM and eVTOL digital twins, connecting vehicle dynamics, cloud and edge simulation, dependable platforms, blockchain, federated learning, and multi-agent AI.
- Manage cross-functional work packages, milestones, technical risk, and research output across AI control, flight dynamics, cloud-in-the-loop simulation, and UAM service dependability.
- Guide researchers on physics-informed and data-driven dynamics, robust control, tele-operation, situation awareness, and virtual certification for safe AAM and UAM operations.
- Digital twin architecture
- Cloud-in-the-loop simulation
- Physics-informed dynamics
- Federated learning
- Blockchain
- Virtual certification
Learning control Led
Training Experts in Future Automobile R&D
Lead Researcher
A talent programme run as applied R&D: graduate projects on autonomous navigation, SLAM and deep reinforcement learning, sensor-data processing and reliable vehicle service infrastructure, aligned with what the mobility industry actually needed from software.
Programme funded to Feb 2027; my involvement ended with the Konkuk appointment in Jun 2025.
- Lead the AI and software research tasks for future-mobility talent development, aligning autonomous driving, connected vehicles, vehicle software, edge and cloud computing, and data intelligence with industry needs.
- Design and supervise graduate-level R&D projects on autonomous navigation, SLAM and deep reinforcement learning, sensor-data processing, distributed data platforms, and reliable vehicle service infrastructures.
- Coordinate student researchers, industry-style deliverables, technical training content, and the publication output that converted the education programme into applied R&D capability.
- Deep reinforcement learning
- SLAM
- Autonomous navigation
- Sensor fusion
- Edge/cloud platforms
Dependability Led
Secure and Reliable Multi-Agent Public Safety Services for Disaster Tolerance in Urban Smart Cities
Lead Researcher
The i-MeC2 framework: secure, reliable, multi-agent public-safety services for smart cities, built on bio-inspired collaboration and intelligent edge and cloud computing — and the quantitative models that tell an operator how much resilience a given architecture actually buys.
Five-year programme funded to Aug 2026; my involvement ended with the Konkuk appointment in Jun 2025.
- Lead research planning for secure, reliable, multi-agent public-safety services using AI, bio-inspired collaboration, intelligent edge and cloud computing, and resilient smart-city architectures.
- Develop dependability, security, performance, and energy models for smart surveillance, IoT disaster detection, medical and industrial IoT, VANET and UAM services, moving-target defence, and blockchain platforms.
- Translate stochastic modelling, sensitivity analysis, and fail-over results into technology-management guidance for capacity planning, risk control, and resilient service operation.
- Stochastic reward nets
- Hierarchical models
- Sensitivity analysis
- Moving-target defence
- Edge/fog computing
- Blockchain
Learning control Led
Certified and Safe Neural Control and Reinforcement Learning for eVTOL/PAV Flight
Lead Researcher · Advisor
Learned flight controllers you can argue about formally: deep neural attitude and flight control carrying exponentially stabilizing control-Lyapunov certificates, tiltrotor transition under safety constraints, and verification machinery strong enough to certify what the network learned.
Research thrust running to 2026; my involvement ended with the Konkuk appointment in Jun 2025.
- Develop deep-neural attitude and flight controllers with exponentially stabilizing control-Lyapunov functions (es-DNLC), validated in PX4 software-in-the-loop and Gazebo.
- Develop reinforcement-learning controllers for tiltrotor eVTOL transition and reconversion under safety constraints, with Lyapunov and region-of-attraction certified stability.
- Establish formal verification of learned controllers — interval analysis, branch-and-bound, counterexample-guided synthesis, and SMT solving — to certify stability and forward invariance.
- es-DNLC
- Control-Lyapunov functions
- PX4 SITL
- Gazebo
- Region of attraction
- CEGIS
- dReal SMT
Dependable cloud and network infrastructure2014 – 2020
Stochastic models pointed at real infrastructure: active-active clouds, software-defined networks, hybrid cloud economics, naval survivability, and the reliability of an early unmanned aircraft.
Dependability Led
Korea–Vietnam Cooperative Project on Disaster-Tolerant Cloud Systems
Lead Researcher
A bilateral study putting a number on business continuity: how much a geographically distributed, active-active cloud is really worth once you model the sites, the links between them, and the recovery policies that connect the two.
- Lead the Korea–Vietnam collaboration on disaster-tolerant cloud systems, coordinating technical scenarios, modelling assumptions, and joint validation with partners at IOIT/VAST.
- Design active-active GENESIS cloud reliability and availability studies quantifying the business-continuity value of distributed sites, inter-site links, and recovery policies.
- Manage cross-border execution, publication planning, and technology recommendations for resilient cloud architecture investment and service-continuity governance.
- Active-active clouds
- Stochastic reward nets
- Sensitivity analysis
- Business continuity
Dependability Contributed
High-Performance Hybrid Cloud Computing Technologies for Knowledge Information Services
Technical Staff
Model-based technology assessment for high-performance hybrid cloud services — virtualized servers, software-defined networks, data-centre networks — turned into concrete guidance on redundancy, migration, rejuvenation, capacity and the cost of an SLA.
- Develop model-based technology assessment for high-performance hybrid cloud services, focusing on virtualized servers, SDN infrastructures, cloud data-centre networks, and cost-aware availability.
- Apply stochastic reward nets, hierarchical models, and performability analysis to guide redundancy, VM migration, software rejuvenation, resource capacity, and SLA/cost trade-offs.
- Support human-resource development through lab mentoring, reproducible experiments, technical writing, and transfer of cloud reliability methods into practical systems engineering.
- Stochastic reward nets
- Hierarchical modelling
- Performability
- Software rejuvenation
- SDN
Dependability Contributed
Distributed/Parallel Platform of Streaming Big Data Processing for Logistics and Transportation
Technical Staff
The GENESIS cloud platform for distributed, parallel streaming analytics in logistics and transportation, with active-active high availability and disaster tolerance as first-class requirements rather than afterthoughts.
- Contribute to the GENESIS cloud platform for distributed and parallel streaming big-data processing in logistics and transportation, with emphasis on active-active high availability and disaster tolerance.
- Build reliability and availability models and sensitivity analyses for cloud sites, data-centre networks, virtualized servers, storage, SDN, and resource provisioning under large-scale workloads.
- Support prototype integration, technical documentation, experiment design, and the publications that informed architecture trade-offs and continuity planning.
- Streaming analytics
- Active-active HA
- Disaster tolerance
- SRN modelling
- Storage virtualization
Dependability Contributed
Analysis Technique Study (ST-54) for Improving Integrated Survivability of Naval Ships
Software Developer · Core Researcher
Defence-funded work on integrated survivability analysis for naval vessels: the software and the analysis technique behind how a ship's mission capability degrades as its compartments and systems are damaged.
- Survivability analysis
- Systems modelling
- Simulation software
Dependability Contributed
Reliability Study of Advanced UAV
Researcher
A reliability study of an advanced unmanned aerial vehicle — the first time the dependability modelling built for data centres was pointed at an airframe, a decade before the same question returned as UAM mission dependability.
- Reliability modelling
- Fault analysis
- UAV systems
Space, UAV and embedded systems2009 – 2015
The engineering apprenticeship. A CubeSat in Hanoi, then unmanned airships, balloon payloads and survey aircraft in Seoul — the systems that made redundancy and survivability feel concrete rather than theoretical.
Aerospace Contributed
UAV-Based Korea–Peru Cooperation for Resource Exploration and Environmental Monitoring
Core Software Developer · Technical Staff
A science-and-technology cooperation programme flying unmanned aircraft for resource exploration and environmental monitoring, and the embedded flight and mission software that made those surveys repeatable.
- Embedded flight software
- Mission planning
- Remote sensing payloads
Aerospace Contributed
Construction of Historical Sites 3D Spatial Information Using Unmanned Airships
Researcher
Unmanned airships used as a slow, stable camera platform to build three-dimensional spatial information for historical sites — aerial photogrammetry where a fixed-wing survey would be too fast and a crane too intrusive.
- Aerial photogrammetry
- Unmanned airships
- 3D spatial reconstruction
Related publications C04
Aerospace Contributed
Balloon-Satellite Project for Training and Education
Core Software/Hardware Systems Developer
A near-space balloon platform built as a teaching instrument: payload electronics, on-board software and telemetry cheap enough to fly with students, close enough to a real spacecraft to teach the discipline.
- Payload electronics
- Embedded software
- Telemetry and ground station
Aerospace Contributed
Unmanned Spacecraft Automatic Control System Development
Researcher
Automatic control system development for an unmanned craft — the embedded hardware and software architecture, the sensing, and the control loop, published as the unmanned airship control work that opens the aerospace strand of the publication list.
- Automatic control
- Embedded hardware design
- Real-time control software
Aerospace Led Where it started
F-1 Pico-Satellite for Training and Education in Vietnam
Core Software/Hardware Developer · Technical Subteam Leader
Where all of it starts. A 10 cm, 1 kg CubeSat built from commercial off-the-shelf parts as Vietnam's training platform and Earth-observation mission — and the first system where the answer to "will it survive?" had to be designed in, not measured after. The reliability question raised here became the M.Sc. thesis and, eventually, the whole dependability track.
- Core software and hardware developer and technical subteam leader on the F-1 pico-satellite at the FSpace Laboratory, in partnership with VAST.
- 10 cm cube, 1 kg, built from commercial off-the-shelf components; two 0.3 MP cameras, one 1.0 MP camera, and temperature, magnetic-field and current sensors.
- Power supply, on-board computer and communications subsystems made double or triple redundant for survivability against launch vibration and the on-orbit thermal and radiation environment.
- CubeSat systems
- Redundant subsystem design
- Embedded C
- Reliability analysis
Related publications C01