What I do

At AIKO I lead the Virtual Environments department, developing real-time simulation technologies for aerospace and autonomy applications. Our work supports AIKO’s autonomy efforts through training environments, synthetic data generation, validation workflows, and interactive demonstrations.

We are not the product development team; instead, we provide the simulation and experimentation layer that helps train models, validate behaviors, and showcase capabilities in realistic scenarios.

Synthetic dataset generation for AI-based navigation

We build simulators that generate photorealistic imagery and telemetry for visual navigation and pose estimation, especially in scenarios where real orbital data is scarce or impossible to collect (e.g., asteroid proximity, CubeSat rendezvous).

Reinforcement learning environments for GNC

We create closed-loop simulators where AI-based Guidance, Navigation and Control (GNC) agents are trained, tested and stressed before deployment. A concrete example is our RL-controlled agent for In-Orbit Servicing (IOS) scenarios, where the agent learns to perform orbital approach and docking between CubeSats.

ML training & validation with controllable scenarios

We design repeatable, parameterized scenarios to validate GNC algorithms, pose estimation models, and visual navigation pipelines, covering edge cases and long-tail distributions that are hard to reproduce in real missions.

Interactive VR/AR and real-time simulators

We turn complex space technologies into immersive experiences and intuitive UI tools: from VR demos for AI-based lunar landing agents to real-time simulators where operators can interact with the dynamics, debug the simulation, and challenge the RL agent against a human player.

Tools and stack for in-orbit servicing missions

Our virtual environments integrate:

• orbital and attitude dynamics simulation,
• optical payload simulation,
• hardware subsystems modeling,
• everything built on Unity3D, C#, Python, with a strong focus on real-time performance and numerical accuracy.

The stack: Unity3D, C#, Python, and a healthy dose of numerical methods.

The vision: use virtual environments to accelerate the adoption of AI in space, reduce risk, and make the development of intelligent systems more transparent and data-centric.