This paper aims to introduce an integrated planning method for the base position (BP) selection and its optimal visiting sequence for mobile manipulators executing large-scale construction tasks. The method ensures the overall optimality of the planning results through a bilevel coupled optimization of navigation cost and manipulation cost.
A two-stage methodology is proposed. First, candidate BP clusters are generated by explicitly incorporating construction process constraints (e.g. layered assembly and staggered patterns). Second, the BP sequence planning is formally modeled as an improved Generalized Traveling Salesman Problem (IGTSP), which is solved using a bilevel cost function that integrates and minimizes both navigation cost (mobile platform path length) and manipulation cost (robotic arm end-effector path length).
Experimental evaluations confirmed that by introducing construction constraints, the method achieved significantly higher BP cluster robustness across different layers (with improvements exceeding 14%). Furthermore, compared to state-of-the-art methods, the proposed IGTSP approach successfully reduced the total comprehensive cost by more than 8%, thereby achieving a more efficient and feasible construction sequence and ensuring the smooth completion of on-site tasks.
This research introduces a process-aware, coupled optimization framework that overcomes the sub-optimality inherent in decoupled BP planning. The novelty lies in the explicit integration of construction process constraints during the BP cluster generation and the formulation of the IGTSP with a bilevel navigation-manipulation cost function, providing a significant advancement toward safe, highly efficient and automated large-scale construction robotics.
