BUILDER: Leveraging BIM information to automate robot task planning for lunar assembly applications

In order to support permanent human presence on the Moon, a number of critical infrastructures must be built on the lunar surface. Robots are expected to play a crucial role in the construction process, performing autonomous assembly and outfitting of a variety of systems, such as tall towers for solar power generation, shelters for crew and asset protection, and lander plume containment shields for launch & landing pads. As it has been noted in state-of-the-art work, the knowledge of building data is critical for the application of robotic solutions in terrestrial construction and operation. Specifically, these works leverage the information contained in Building Information Modeling (BIM) files that describe a target infrastructure, to extract and provide the robotic agent with the geometric, semantic and scheduling data that can then be used to assist in the generation of assembly plans to be carried out by the robot. In this proposal, TRACLabs proposes BUILDER (Leveraging BIM information to automate robot task planning for lunar assembly applications), a set of software tools that facilitate the integration of BIM data - describing an assembly infrastructure - with ROS, the standard software framework to build robot applications. The BUILDER tools are expected to provide 3 types of utilities: (1) Translate geometric data (3D models and poses) from the BIM files into objects natively usable by ROS-based tools, such as motion planning libraries; (2) Translate scheduling and semantic BIM data - describing the assembly and its construction - into a format suitable to be used for generic task planning tools; (3) Build upon our existing task planning tools, we'll integrate the geometric, scheduling and semantic data now accessible to ROS-based tools, into our framework, enabling it to autonomously generate task plans that are specific to a robotic platform, with the robot plans being encoded using a hierarchical approach based on Behavior Trees.