Abstract

Robotic machining cells are commonly engineered independently from material-flow optimization, renewable-energy supervision, and condition-monitoring functions, which limits factory-level coordination and makes system claims difficult to verify. This case study consolidates two implemented foundations-an ABB IRB 6700 stone-machining cell and ABBP, a mixed-case palletizing decision-support prototype-and defines a safety-separated reference architecture for their future integration with photovoltaic metering and advisory artificial-intelligence-based condition monitoring. ABBP combines a height-guided anchor-point heuristic with a hybrid genetic algorithm subject to pallet boundaries, non-overlap, weight, and a 70% base-support constraint. A Python optimization service exchanges JSON records with a C++/Qt interface for real-time three-dimensional inspection and slicing. The reported benchmark used 120 boxes from 12 types, a population of 50 and 20 generations. One recorded run required 10.86 min and obtained 79.39% pallet-volume utilization, placing 58 boxes with a mass of 278.05kg, an air-exposure score of 0.84, a centre-of-gravity score of 0.78 and a composite fitness of 0.8055. These results verify the palletization prototype and supervisory visualization for the documented case; they do not validate physical robot palletizing, solar contribution, or condition-monitoring accuracy. The proposed architecture therefore keeps deterministic robot and safety control independent while using AI only for anomaly scoring, maintenance prioritization and energy-aware scheduling advice. A reproducibility protocol, evidence-status matrix, explicit results section, limitations, conclusions and staged future-validation plan are provided.

Keywords:Industrial robot; Robotic machining; Mixed-case palletizing; Genetic algorithm; 3D visualization; Photovoltaic energy; Condition monitoring; Reference architecture