Doctoral research foundation | In-house calibration and R&D
Particle Dynamics Engine
Model regolith behavior from its particles.
Our Particle Dynamics Engine builds on our CEO's doctoral research results to address particle contact, friction and motion, including interaction with wheels and tools. We compare predicted sinkage, slip and traction with measurements from our own rovers, using hardware validation to calibrate the model and advance research and development. Evaluating different particle and terrain conditions helps us improve lunar-environment fidelity.
In-house development and research | Lunar optics and cameras
Rendering Engine
Recreate how a robot sees the lunar surface.
We are developing an in-house rendering engine for lunar conditions, covering incident and reflected light, shadows cast by terrain and vehicles, surface spectral characteristics and their appearance through physical cameras. Comparison with images from real cameras supports model validation and connects varied illumination and viewpoints with perception, localization and Sim2Real evaluation.
Terrain and mobility physics
Soft-terrain simulator
Model sinkage, slip and traction.
Using environments including Isaac Lab and ROS 2, we are researching how to connect our particle dynamics and rendering engines with a lunar digital twin. Our goal is to examine mobility, visual perception, task duration and power needs within the same scenarios while varying vehicle, terrain, lighting and task conditions.
Simulation–hardware comparison
Measured-data calibration
Measure the gap and update the model.
We plan to compare predicted sinkage, slip and driving behavior with ground measurements, and rendered images with real images under documented lighting conditions. Our goal is to calibrate both physical models and rendering conditions and assess errors and uncertainty for each scenario.
Integrated hardware and software expertise
Hardware and rover development
The potential to develop every system element in house.
Our team combines hardware expertise in mechanisms, thermal design, electronics, embedded systems, measurement and sensing with software expertise in simulation, perception and control. Building on nine generations of rovers developed over four years in ARES Project, we have the potential to develop every system element in house, from the chassis and electronics to control and simulation. Specialists work together on integration, validation and improvement using our own hardware.
From ground tests to space missions
A model that grows with data
Update the model with each experiment.
We plan to accumulate ground-test data and later incorporate space-mission measurements. Calibration data and ongoing model updates will underpin future digital twin and data licensing.
Reinforcement learning and physical AI
Sim2Real and autonomous control
Train thousands, validate with models, transfer to hardware.
Our Particle Dynamics Engine supports reinforcement learning (RL) with thousands of virtual rovers and model-based behavior validation. Varying terrain, vehicle and control conditions lets us compare and evaluate learned control policies within the physics model. Validation and calibration with our own hardware help narrow the simulation gap, connecting this work with rendered visual-perception evaluation through Sim2Real.
