Knowledge-Driven Workflow Optimization for Integrating Industrial Robotics into Conventional Construction Management

Abstract

Industrial robots are increasingly capable of executing construction tasks that were previously organized around human crews, yet conventional construction management systems represent work at a granularity that is too coarse for robotic execution. Work breakdown structures, critical path schedules, and Last Planner routines coordinate trades and buffers but do not encode robot reach, payload, tolerance, calibration, safety-zone, and handoff requirements. This paper develops a knowledge-driven workflow optimization framework that links construction-domain knowledge to executable robotic work packages and to robust scheduling decisions. The framework has three components. First, an ontology and knowledge graph represent tasks, work packages, robot capabilities, site states, constraints, and evidence, allowing capability matching and constraint generation to be traced to explicit knowledge rather than ad hoc rules. Second, a capability-aware eligibility model filters task-robot assignments, and a precedence model represents setup, calibration, and safety handoffs as first-class workflow dependencies. Third, a mixed-integer optimization model balances makespan, cost, risk, and schedule deviation under bounded duration uncertainty, with execution feedback updating capability estimates. The contribution is a formal integration logic for embedding industrial robotics into conventional construction management without replacing existing planning artifacts. The framework clarifies how robotic constraints alter workflow precedence, why semantic knowledge representation is necessary for reliable constraint generation, and how robust optimization can support managerial trade-offs. Its main implications concern task decomposition, handoff design, knowledge maintenance, and the boundary conditions under which robotic construction schedules remain both feasible and auditable.