Hanno Moes developed a multi-physics digital twin of an Epiroc Scooptram platform to support safer braking and collision-avoidance decisions underground. The work responds to a central limitation of autonomous braking: stopping distance changes with terrain, gradient, vehicle pose and the mass and position of material in the bucket.
Modelling the machine and its environment
The research combines a physics-based vehicle model with data-driven parameter estimation. The model represents multi-body dynamics, hydraulics and tyre-terrain interaction and tracks quantities such as actuator positions, hydraulic pressures, centre of gravity, mass moments of inertia and wheel loads.
An early high-fidelity implementation ran more slowly than real time, so Hanno investigated reduced-order alternatives. Polynomial centre-of-gravity estimates achieved adjusted R² values above 0.999 with root mean squared error below 0.008 in the reported synthetic-data study. The reduced model ran substantially faster than real time, making it suitable as a future predictive-control component.
From model to intervention
The digital twin is intended to estimate otherwise unmeasured quantities—particularly bucket load and centre-of-gravity position—and use them to improve predicted stopping distance and braking strategy. Hardware-in-the-loop work demonstrated the feasibility of executing the model in a real-time control setting.
The project was completed with Epiroc collaboration and connects closely to RAMMS engine, traction and machine-vision studies. It was supervised by Prof Stephan Heyns and Prof Herman Hamersma. Hanno submitted his dissertation for examination in the second quarter of 2026 and was awaiting examiner reports at the reporting date.