Sequence planning and physics-informed robotic simulation for assembly of deformable beam structures

Automated construction represents a future direction for the AEC industry, with robotic assembly playing a critical role. However, assembling beams remains challenging due to mechanical constraints and structural deformations. To narrow this gap, this paper presents an integrated framework combining sequence planning with physics-informed simulation. An enhanced genetic algorithm employs heuristic initialisation and feasibility-preserving operators to generate constructible sequences, while reducing finite element analysis calls by approximately 80% compared to a baseline without caching or early termination. The co-simulation environment integrates PyBullet for robotic motion planning with OpenSeesPy for structural analysis, enabling realistic modelling of deformation behaviour and identification of failure modes. Results demonstrate that the sequencing method produces comparable or superior sequences than the iterative constraint-tightening search method, while offering improved computational efficiency, although challenges persist for high-dimensional problems. This work narrows the gap between idealised planning and physical execution, supporting advances in the automated construction of deformable structures.