A PHYSICS-BASED SIMULATION ENVIRONMENT FOR LUNAR ROVER OPERATIONS

Abstract

This thesis extends Project Chrono, a multi-physics simulation engine, to create a robotics simulation framework for lunar applications. The work enhances Chrono’s capabilities in three key areas: lunar scene modeling, sensor simulation, and large-scale world simulation. New features were developed to accurately model lunar lighting, dust dynamics, and terrain deformation, alongside the creation of a new module, Chrono::Worlds, for managing extensive lunar environments. Motivated by the need for high-fidelity simulation tools to support lunar exploration and robotics, this project advances Project Chrono toward realistic and scalable lunar mission simulations. First, a Hapke-based Bidirectional Reflectance Distribution Function (BRDF) model was implemented to realistically render the lunar surface’s unique optical properties, including the opposition effect and high dynamic range lighting conditions. Terrain deformation visualization was enhanced using voxel-based rendering of Smoothed Particle Hydrodynamics (SPH) terrain data, enabling realistic modeling of wheel-regolith interactions. Additionally, a dust modeling and volumetric rendering pipeline was introduced to simulate and visualize dust clouds kicked up by rover mobility, accounting for their impact on sensors and visibility. Second, the Chrono::Sensor module was expanded to a novel Time-of-Flight (ToF) sensor model based on transient rendering techniques. These enhancements allow simulation of SPAD- based ToF phenomena, providing higher-fidelity virtual sensor data critical for perception algorithm development. Third, a new module, Chrono::Worlds, was developed to enable efficient management and simulation of large-scale lunar terrains, utilizing Level-of-Detail (LOD) management through

a quadtree data structure. Both NASA Lunar Reconnaissance Orbiter (LRO) datasets and procedurally generated lunar terrains were integrated to support extended rover operations over kilometer-scale domains. The developed simulation environment was validated through several applications, in- cluding recreating imagery from the NASA POLAR dataset, simulating dust-affected rover operations, and operating rovers such as VIPER and RASSOR on both real and synthetic lunar terrains. This work positions Project Chrono as a comprehensive and open-source platform for simulating complex lunar robotics scenarios with high physical and visual realism.