Measuring the Labour-Time Effects of Construction Automation: A Falsifiable Programme and an Observability Proof from Open Field Data

Construction automation decisions increasingly rely on estimates of how much labour time a technology could save, yet the key state variables required for such estimates have rarely been measured directly on real construction sites. This paper addresses that measurement gap. It asks what would be required to move from proxy-based screening to field-based measurement of the labour-time effects of construction automation. The paper makes two main contributions. First, it defines a falsifiable four-hypothesis programme for measuring the quantities that a recent Kingman-derived screening framework requires, with explicit success and kill criteria stated before any new data is collected. Second, it demonstrates, usingthe open Aalto/VTT MEP time-motion dataset, that second-level field observation of labour states on real construction sites is already feasible with openly available data. The Aalto record provides per-worker analogues of the relevant state variables at realistic field granularity, but not the station-level item-flow data needed for a full test of the analytical model. On that basis, the paper specifies a sensing architecture that combines worker-mounted inertial measurement, machine telematics, optional fixedcamera work-zone occupancy, and manual coding as ground truth. The contribution is therefore a measurement framework and an observability proof, not a re-derivation of the analytical model and not yet a field-validation study. The paper closes the gap between proxy-based decision support and future first-party field testing of labour-time effects in construction automation.