Publications / 2024 Proceedings of the 41st ISARC, Lille, France
Physics-based simulations play a crucial role in the design and development of autonomous construction equipment. Currently, a key challenge in these simulations is the time-intensive task of preparing construction equipment models that accurately represent both the equipment's geometry and physics. Manual model creation for simulations becomes particularly laborious due to the integration of diverse mechanical data such as materials, joints, and drives with the geometric data. Extant methods for automatic physics-based modeling of standard modular robots are inadequate for addressing the complexities of construction equipment. Therefore, this paper investigates the feasibility of automating and streamlining the physics-based modeling process by fusing the construction equipments mechanical data into its 3D computer-aided design (CAD) model. The proposed method involves converting the construction equipment 3D CAD model into a universal scene description (USD) model for efficient data fusion. Subsequently, the method automatically configures material parameters, collision meshes, establishes component relations, and incorporates joints and drives for the USD model. To validate the efficacy of this approach, the proposed method is applied to create a physics-based model of a Caterpillar 390F LME excavator, and simulated in a scalable robotic simulator (NVIDIA Isaac Sim). The findings demonstrate that the proposed method significantly reduces the time required for physics-based modeling compared to traditional manual methods.