Virtual Fly with a Brain – When AI Truly Takes Flight

Virtual Fly with a Brain – When AI Truly Takes Flight
Flying like in nature – AI simulates insects with unprecedented precision

Visualization: © Ulrich Buckenlei | Symbolic image of the digital twin of a fruit fly with lifelike motion intelligence

Digital fly, real behavior – The revolution of AI simulation

How realistically can artificial intelligence replicate the behavior of living organisms? Research teams from Janelia and DeepMind provide the answer with an impressive premiere: the first fully AI-controlled fruit fly in virtual space. This model not only moves like a real insect – it thinks, reacts, and acts with astonishing precision.

Instead of animations, a biologically informed, data-driven simulator is used. Every wingbeat, every landing, every reaction is based on real behavioral input from nature. The AI learns from these patterns and controls a complete digital skeleton in real time.

  • Trained on real behavior: The AI controls every leg, every turn, every flight angle.
  • Physics-based: The simulation considers adhesive surfaces, air displacement, and gravity.
  • Biologically inspired: Structure and control are based on anatomical models.

User interacts with volumetric display in space

Digital twin of a fruit fly with real-time response to the environment

Image: © Ulrich Buckenlei | Source: Janelia Research Campus / DeepMind

What is being created here is more than just a model – it’s a new research platform. The combination of anatomy, AI, and simulation physics takes basic research to a new level and paves the way for applications in robotics, biology, and immersive systems.

Biology you can touch – The digital fly comes to life

Recreating nature has long been a goal of research – but only now is it possible not just to visualize a fruit fly, but to bring it digitally to life. This is made possible by a precise anatomical model, coupled with artificial intelligence and a physical simulation environment.

  • Anatomically accurate: All limbs and joints correspond to the real model.
  • Behavior-based: The AI was trained on actual motion data.
  • Interactive: The fly responds dynamically to its environment.

Schematic representation of the FlexiVol technology

Digital fruit fly with realistic real-time behavior

Image: © Ulrich Buckenlei | Source: DeepMind / Janelia Research Campus

The simulation doesn’t just show movement – it demonstrates behavior. The model lifts, lowers, balances, and corrects itself like a real insect. This yields insights for robotics, neuroscience, and virtual biology.

Understanding motion – Learned by AI

The fly is controlled by an AI trained on real motion data. Every wingbeat, every step is based on biological patterns and optimized through machine learning.

  • Highly realistic: Data from high-speed video studies flows directly into the simulation.
  • Precision through training: AI controls joints, balance, and responses.
  • Real-time processing: Movements occur without delay.

Infographic on the functionality of FlexiVol

Neural AI model controls every joint of the digital fly

Image: © Ulrich Buckenlei | Source: DeepMind / Janelia Research Campus

The AI is trained through physically realistic feedback: when the fly falls or slips, it learns from it. The result is a behavior that not only functions but feels alive.

Flight behavior and adhesion simulated

Particularly impressive is the simulation of flight phases – including landing, steering, and adhesion to surfaces.

  • Adhesion physics: The fly can stick to glass or walls.
  • Reactive flight maneuvers: Threats are actively avoided.
  • Controlled landing: Simulation also succeeds in uneven environments.

Industrial use case with volumetric interaction

Reactive flight simulation including surface adhesion

Image: © Ulrich Buckenlei | Source: DeepMind / Janelia Research Campus

Especially the precise execution of leg motor control allows for realistic scenarios like turning in the smallest space or sudden takeoff.

Infographic: How the system works

The following infographic shows how anatomy, physics, and AI merge into a digital life model.

Comparison of XR interfaces with FlexiVol system

System architecture: Anatomy, behavior, real-time control

Graphic: © Ulrich Buckenlei

This modular simulation also allows for variations: other insects, robot models, and even micro-avatars could emerge using the same principles.

Biology in motion – The video

The poster image shows a snapshot – the following video clearly illustrates how fluid and complex the motion logic truly is.

Biologically inspired flight simulation in real-time – AI meets insect physics

Image: © Ulrich Buckenlei | Visual based on research data from Janelia Research Campus & DeepMind | Source: https://www.janelia.org/project-team/flyem | Video analysis: XR Stager / Visoric

New opportunities for research, exhibitions, and product design

  • Are you looking for immersive experiences for your trade show booth?
  • Would you like to make digital twins accessible as interactive simulations?
  • Or are you planning educational or research formats with virtual biology?

Then talk to the expert team at VISORIC. We develop immersive presentations, interactive research experiences, and AI-based models for industry, science, and the creative sector.

Contact Persons:
Ulrich Buckenlei (Creative Director)
Mobile: +49 152 53532871
Email: ulrich.buckenlei@visoric.com

Nataliya Daniltseva (Project Manager)
Mobile: +49 176 72805705
Email: nataliya.daniltseva@visoric.com

Address:
VISORIC GmbH
Bayerstraße 13
D-80335 Munich

The fields marked with * are required.

Arrow right icon