A Twin Robotic CT System Simulation using MuJoCo and gVXR for Medical Applications

Abstract

Twin robotic CT systems enable flexible scan trajectories by independently controlling the x-ray source and detector. This work introduces a simulation environment that integrates MuJoCo for robot dynamics and gVXR for x-ray imaging via ROS2, aiming to accelerate trajectory optimization for medical applications. A proof-of-concept scenario with a trajectory on a sphere around the target object demonstrates feasibility. The iterative SART algorithm yields high quality image reconstructions. The framework lays the foundation for exploring novel trajectory strategies, with future work focusing on modeling system uncertainties, optimizing arbitrary scan paths and task-specific image reconstruction.