Advancing Large Space Structure Assembly: Insights from Earth

Current space technology does not yet have suitable methods for assembling Large-Scale Space Structure (LSS), such as Space Based Solar Power stations. Current research has focused on smaller assemblies, such as large space telescopes. However, such methods will struggle to support larger assemblies due to computational scaling and the system’s ability to determine the state of the assembly (world state). This paper reviews the state-of-the-art assembly planning and modelling in space and on Earth. Literature has identified and categorised three types of assembly problems: assembly sequence planning, assembly line balancing and assembly path planning. However, the constraints and cost drivers
for LSS assembly differ significantly from those in terrestrial cases. For example, prefabricated construction requires human insight during the final assembly and research has investigated the feasibility of using robotics for this final stage. Sources indicate that currently, robotic platforms are unable to meet the complexity required for complete fabrication, as the final plug-in stage necessitates attention that robotic platforms cannot provide. The LSS case is not cleanly divided into the three assembly problems. A new paradigm needs to be constructed so that LSS assembly can occur. Newer, novel techniques, such as abstractive and surrogate modelling, are explored and shown to excel in the space environment. Finally, this paper proposes a hierarchical, abstractive, layered approach to the LSS assembly problem, featuring an accurate and computable model of the entire assembly process, while addressing potential future constraints and assumptions. This model provides a more precise description of the system’s ‘world state’, a topic often overlooked by other researchers. The first layer developed is the geometric layer, which outlines the initial plan for where modules interface to achieve the required geometry. Secondly, a layer is added that deals with the local operation of robots to accomplish a specific task, which varies depending on the application. In a space-based solar power station, this controls multiple robots that pick and place parts from a logistical depot or harbour. Lastly, a third layer is developed, in which individual control of a robot is processed, which can be achieved by integrating the robot’s kinematics or using a path planning optimisation technique. For each layer, new or existing control schemes are discussed and explored based on their ability to interface with different layers and the dynamic environment of LSS assembly.