Position-based fluid simulation for robotic injection sealing of pavement cracks

Abstract

Automated crack sealing could significantly benefit the maintenance of road pavements, but there is difficulty in depositing the correct volume of sealant material into the hidden crack space. A simulated model of the material flow within a crack space would allow the development of a predictive control scheme, such that the repair robot can apply suitable trajectories and operational parameters to accomplish neatly sealed surfaces. For the first time, the position-based fluid (PBF) method, a computationally cheap and fast but approximate model of fluid flows, is studied for its feasibility for sealant flow simulation in the robotic injection crack sealing scenario. A Real-to-Sim experiment is performed, in which a PBF simulation of sealant in a virtual robotic crack sealing environment is mirrored from the physical lab setup. The fluid simulation is tuned to match the real-world dynamics through comparison with 132 simulation runs, varying the artificial viscosity parameters (fluid–fluid viscous interaction) and (fluid–wall viscous interaction). It was found that had a varied three-stage influence on the simulation error while 's negative influence on the simulation error only effectively applied to fluids satisfying . Through comparing the physical and virtual crack sealing results, the simulation was validated with an average fluid level error of 1.26 mm along a 3.1 mm wide, 16 mm deep and 80 mm long artificial crack, which shows the usefulness of the PBF method for robotic injection sealing. The accuracy and computational requirements of the PBF method are also discussed.