Integrating QN-MHP With MTM-1: A Computational Architecture for Human Performance Prediction in Industrial Tasks

Accurate cycle-time prediction in manufacturing needs to consider both cognitive and physical complexity, but existing models address only one. The Queueing Network-Model Human Processor (QN-MHP) provides multitasking cognitive modeling but relies on simplified regression equations for motor execution. Conversely, Methods-Time Measurement (MTM-1) provides precise basic motion time standards but ignores cognitive processing. We introduce QN-MHP-MTM, a computational architecture that replaces QN-MHP’s motor execution with MTM-1 to add industrial motion accuracy while preserving its cognitive layer. An open-source prehension dataset (51 conditions; 4 object sets; reach distances 170 to 470 mm) was used to validate the architecture. Compared with pure MTM, QN-MHP-MTM decreased trial-level RMSE from 257.4 to 127.4 ms (50.5%) and condition-level RMSE from 234.6 to 73.1 ms (68.8%), and reduced MTM’s systematic underprediction (bias reduced from −226.8 to −37.6 ms). Our results show that integrating cognitive architectures with predetermined motion-time standards can provide more accurate human-task time prediction.