Industry
Factories still treat people as fixed capacity: Industry 5.0 survey on the missing closed loop
An Dongguan molding plant added station cameras and wristbands. The big screen can flag “fatigue high.” Dispatch still runs on the morning takt sheet: when someone is worn down or shoulder risk rises, the lead swaps people by voice while the system’s labor coefficients do not move. Sensing arrived. Scheduling did not plug in.
A University of Coimbra Industry 5.0 survey by Pereira, Ferreira, and Paulo turns that break into a checkable claim: worker-state monitoring is relatively mature, yet a full closed loop from live sensing to adaptive operations decisions to active worker feedback has not been publicly validated. This note maps plant management moves. It is not a reprint.
What you can verify in the paper
The authors run an umbrella survey of related work (roughly 2018–2026) and organize the mainline into three dimensions:
- Worker state monitoring: physical fatigue and ergonomic-risk sensing (wearables, markerless computer vision, and related ML) is relatively mature; cognitive-load methods that are industrially usable in real time remain weaker.
- Human factors in operations management: many scheduling and task-allocation models still treat people as static resources with fixed service rates (the worker-as-resource assumption). Even when fatigue or ergonomics appear, they are often analytical assumptions or population averages—not live, sensor-derived estimates on shift.
- Human-in-the-loop (HITL): trust calibration, explainability, and override rights are widely discussed, yet validated systems where workers actively feedback on operations-level adaptations and close the loop remain missing in the open literature reviewed.
Their central gap: no publicly documented and validated system connects real-time sensor-driven worker-state estimation with adaptive OM decisions (scheduling/task allocation) and worker feedback in one unified closed-loop architecture.
Source of record: arXiv:2609.26245 (PDF). Title: Human-Centricity in Industry 5.0: A Survey of Worker Sensing, Adaptive Operations, and Human-in-the-Loop Systems.
Pearl River Delta translation: find where the loop breaks
Most shops do not need another fatigue tile. They need a responsibility chain from state to dispatch to feedback. A workable minimum:
- Name the data’s job — are wearable/vision scores for EHS retrospectives, or for same-shift task changes? Write the use into the rulebook; do not stop at the dashboard.
- Audit the dispatch assumption — do labor standards, takt, and assignments still use fixed capacity coefficients? If state changes and coefficients do not, you are still in worker-as-resource mode.
- Leave workers a rewrite door — when the system suggests a swap or load cut, can the team see the reason, veto it, and leave a trace? Without explanation and veto, it is not HITL.
- Do not mistake offline redesign for a closed loop — the survey cites prior work such as Menanno et al., where offline RGB video supported ergonomic reassignment (reported ~13% ergonomic-stress reduction in that prior study). That is workstation redesign, not live sensor-driven same-shift scheduling.
Organization power here means locking sensing, dispatch, and team veto into one shop system—not buying another isolated sensor stack.
A fifteen-minute self-check
Did this week’s fatigue/ergonomic data change a dispatch decision once? Can workers see and rewrite system suggestions? Two blanks means sensing is mature and the closed loop is still open.
Public wording stays with the arXiv paper. This note only maps plant management moves.
FAQ
What is the core finding?
Worker-state monitoring (fatigue, cognitive load, ergonomic risk) is relatively mature, and OM/HITL streams have progressed on their own, but the survey finds no publicly documented validated system that connects real-time sensor-driven worker state to adaptive scheduling/task allocation and active worker feedback in one closed loop.
What three dimensions does the survey cover?
(1) worker state monitoring in industrial settings; (2) human-factor objectives inside operations-management models; (3) whether HITL architectures let worker feedback close the adaptive decision loop.
What should a Pearl River Delta plant check first?
Do not treat a wearable or vision fatigue dashboard as a finished scheduling upgrade. Check whether dispatch still uses fixed takt/capacity coefficients; whether state data ever changes same-shift tasking; and whether workers can see and override system suggestions.
Where is the source?
arXiv:2609.26245, Human-Centricity in Industry 5.0: A Survey of Worker Sensing, Adaptive Operations, and Human-in-the-Loop Systems (Pereira, Ferreira, Paulo; University of Coimbra).
