Skip links
Flyward
03 · Simulation & Digital Twin

Digital Twin Simulation PlatformFly every mission here first.

A commercially available, planet-scale digital twin for aircraft: 80+ aircraft models, global 3D terrain and weather, multi-modal sensors and air traffic, with 20+ runs in parallel at 50 times real time. The flight-test range, moved into software.

  • 0180+ aircraft models12 classes, from airliners and general aviation to drones and eVTOLs
  • 025+ sensor typesCamera, infrared, depth, LiDAR and mmWave radar
  • 0320+ parallel runsSimulation instances side by side, fully isolated
  • 0450× real timeAn hour of flight in just over a minute
3D visual simulation

Anywhere on Earth, flown here first.

Our own full-stack 3D flight visualisation: terrain, cities and airports that match the real world, with high-definition city scenes of Beijing, Shanghai, Shenzhen, Guangzhou, Hong Kong, Chengdu and Nanjing, plus high-precision custom scenes for specific cities, airports and campuses.

Simulation: a city, a river and distant hillsCity panorama
Simulation: an aircraft over an airport runwayAirport operations
Simulation: an aircraft over mountainsMountain terrain

We developed this software series to benchmark against top international flight simulators such as Microsoft Flight Simulator and X-Plane, offering the most realistic visuals and the most comprehensive features among domestically developed flight simulators in China.

Weather and light

One camera, any weather, any hour.

From sunrise to night, from clear skies to thunderstorms, fog, snow and sandstorms: switch the weather and time of day from the same camera at will, and test again and again in the harshest conditions.

  • Simulation: Shenzhen, sunriseSunrise06:36
  • Simulation: Shenzhen, clear skyClear sky11:14
  • Simulation: Shenzhen, cumulusCumulus16:13
  • Simulation: Shenzhen, overcastOvercast11:11
  • Simulation: Shenzhen, fogFog16:08
  • Simulation: Shenzhen, smogSmog16:05
  • Simulation: Shenzhen, sandstormSandstorm11:13
  • Simulation: Shenzhen, thunderstormThunderstorm16:06
  • Simulation: Shenzhen, nightNight21:00
Multi-camera

Every angle, rendered in sync.

Third-person, side, top-down and tail cameras render in sync with onboard forward and downward views, and you can add custom cameras wherever you need them, for perception training or flight reviews.

  • Simulation: third-person cameraThird-person10:38
  • Simulation: side cameraSide10:38
  • Simulation: top-down cameraTop-down10:38
  • Simulation: tail cameraTail10:38
  • Simulation: forward cameraForward10:38
  • Simulation: downward cameraDownward10:38
Multi-modal sensor simulation

One moment, seen by five sensors at once.

Camera, infrared, depth, LiDAR and mmWave radar are simulated together in the same scene, producing pixel-aligned sensor data with ground truth, so perception algorithms can be trained and tested in closed loop at scale.

Simulation: one frame seen through LiDAR, depth camera, RGB camera, infrared and radar Simulation · one frame, five sensors
  • LiDARMulti-beam scanning point clouds
  • Depth cameraPer-pixel ground-truth range
  • RGB cameraImages under real light, weather and time of day
  • InfraredThermal imaging of airframes, motors and the ground
  • mmWave radarRadar returns with velocity
Air traffic and multi-aircraft coordination

Run the whole sky first.

From one aircraft to large fleets: a single scenario runs 50+ aircraft, 50+ weather zones and 50+ restricted zones at once, simulating route networks, traffic flow, formation flight and conflict detection and resolution. Prove urban air mobility works here before it flies for real.

Illustrative · urban air traffic
  • Route networks and airspace layersCorridors, vertiports and altitude layers modelled together
  • Traffic flow and congestionSee flow, bottlenecks and congestion at a glance
  • Formations and cooperative tasksMany formation shapes, with tasks shared across aircraft
  • Conflict detection and resolutionSpot conflicts between aircraft early and resolve them automatically
One simulation workflow

From scenario to evidence, in one platform.

No environment to build and no code to write: pick an aircraft, set up a scenario and run it, and you get flight data you can replay, analyse and export. Use it on the desktop or in the cloud, or deploy it on your own computing cluster.

Illustration · from scenario to evidence
  1. 01Build the scenarioWeather, routes, sensors, controllers, fault injection and geofences in one place; describe a scenario in a sentence and AI builds it.
  2. 02Run it20+ runs in parallel at 50 times real time; parameter sweeps and Monte Carlo in one click, and the same input always gives the same result.
  3. 03Judge every flightEach flight is checked against criteria such as track accuracy, geofences and stall and ground-proximity warnings, so pass or fail is plain to see.
  4. 04Replay and analyse3D replay, comparison across flights and automatic event tagging, with a flight report written by AI in one click.
Simulation at scale

Ten thousand flight hours, overnight.

More than twenty simulation instances run side by side, fully isolated, each at 50 times real time: one night gives you over ten thousand hours of flight data. Sweep aircraft, weather, speed and initial state, or sample them Monte Carlo style, and run your regression tests in batches.

20 parallel runs · 50× real time 0 flights completed Illustration · Monte Carlo batch runs
  • 20+parallel runsEach instance runs on its own; one failure never touches the others.
  • 50×real timeAn hour of flight in just over a minute.
  • 1000+scenarios in one clickParameter sweeps and Monte Carlo across aircraft, weather and speed.
  • 100%reproducibleThe same input always gives the same result; save checkpoints and resume any time.
Aerodynamics and propulsion

From one-second estimates to full CFD.

Three levels of aerodynamic analysis to suit the job: instant answers for concept work, minutes for comparing configurations, CFD for sign-off, with 3D flow and pressure fields you can see. Lift, drag and lift-to-drag ratio are solved live in flight; batteries, motors, ESCs and propellers are modelled too, so charge, heat and endurance come out of the same run.

Illustration · flow around the wing
L1 · seconds

Quick estimate

  • Instant results from the geometry
  • Concept selection and sizing
  • Stability and performance included
L2 · minutes

Vortex lattice

  • Lift distribution and induced drag
  • Fast comparison across configurations
  • Stability derivatives included
L3 · CFD

Full CFD

  • Accuracy within about ±5%
  • 3D flow and surface pressure, visualised
  • Backs configuration freeze and validation
Closed data loop

Every flight, easy to find, see and compare.

Every simulation run is recorded automatically; finding the one you need among thousands takes seconds.

  1. 01

    Multi-dimensional search

    Filter by aircraft, scenario, date, speed mode and run status, and pinpoint any record among thousands in seconds.

  2. 02

    3D replay

    Trajectories with a full set of flight instruments, with altitude, speed, attitude and weather second by second; steep banks, stalls, low altitude and high load factors are flagged automatically.

  3. 03

    Flight comparison

    Overlay several flights and get the differences and errors computed for you, so you see at once what changed and whether it got better or worse.

  4. 04

    AI flight reports

    Generate a flight report in one click, with the analysis written by AI; export the data any time into your own pipeline.

Stall warning Steep bank Wind shear Altitude (m) · time (s) 0 120 240 360 480 600 Baseline Run A Run B AI flight analysis Run B hit wind shear at 300 s: angle of attack exceeded, stall warning. Run A banked around it and lost 2 m. Advice: slow down early and reroute. Run B: stall warning. Run A: steep bank. Illustration · three flights of one mission
Core capabilities

Modelled like real flight, from dynamics to weather.

Six-degree-of-freedom dynamics, weather and propulsion are solved together. On an approach into a microburst, a headwind first lifts the aircraft, then the downdraft and tailwind press it down, and every second of lift, drag and thrust comes from real physics.

Illustration · approach and go-around in a microburst

80+ aircraft models

Airliners, general aviation, drones, eVTOLs, helicopters, gliders, airships and more: twelve classes ready to fly, and custom models built for your aircraft.

High-fidelity flight dynamics

Industrial-grade six-degree-of-freedom dynamics from takeoff and landing to the ground roll; inertial, satellite-navigation, airspeed, barometric and radar-altimeter sensors with configurable noise and faults injected on demand.

Complex weather

Wind shear, microbursts, mountain waves, jet streams, icing, thunderstorms, rain, snow and fog; weather that moves, builds and decays, with extreme conditions reproduced on demand.

Propulsion and endurance

Batteries, motors, ESCs and propellers modelled as components, so discharge, heat and endurance come out of every run; swap a battery or a motor and compare.

Open integration

Connect PX4, ArduPilot, MATLAB/Simulink, C/C++ and FMI models directly for software- and hardware-in-the-loop, with a programmable interface into your own toolchain.

AI scenario generation

Describe the environment, weather and time of day in a sentence and AI builds the test scenario, so rare conditions are always at hand.

ODD coverage

See what you’ve tested, and what you haven’t.

The platform measures test coverage across 20+ operating conditions in 5+ groups of the operational design domain (ODD). Coverage is judged from measured flight telemetry, not scenario labels, building traceable evidence for airworthiness testing.

Example
Altitude
Within visual line of sight ≤120 mLow 120–500 mMedium 500–2,000 mHigh >2,000 m
Speed
HoverLow speedCruiseHigh speed
Hazardous weather
FogStrong windThunderstormIcingRainSnow
Aircraft class
MultirotorFixed-wingeVTOLHelicopterTilt-rotor / compound
Operational constraints
Geofence
Application scenarios

Whatever needs proving before first flight, prove it here.

Simulation: a new aircraft

New aircraft development and testing

Fly a new configuration in the digital twin first, map its performance and envelope, and cut high-risk flight-test items sharply.

Simulation: autonomous flight

Flight control and autonomy validation

Connect your flight controller and validate it in closed loop across lifelike terrain, cities and weather, with results you can reproduce and compare.

Simulation: urban air mobility

Urban air mobility and delivery rehearsal

Routes, vertiports and the city around them, all recreated, with whole fleets in the same sky, so operations are rehearsed end to end before launch.

Simulation: a city at night

Perception training, night and severe weather

Generate multi-modal sensor data at scale and fly again and again at night and in fog, rain, snow and strong wind, surfacing problems before they happen for real.

Work with us

Let your next aircraft fly here first.

Tell us about your aircraft and what you need to prove, and we’ll demo the platform on your own scenario.