Aircraft assembly is still largely manual, and the processes under the tightest quality requirements are the ones least amenable to automation.

We built a worker fatigue model driven by worker characteristics (skill, age, motivation), task characteristics, and level of automation, then integrated it into the system model those architects already used. Running a critical process from a new aircraft programme through two scenarios, fully manual and calibrated on historical production-line data against a simulated semi-automated alternative, produced fatigue trajectories alongside error rates, time lost, cost, and overall system resilience.

The contribution is less the model than its placement: it lets a human-factors question be asked in the same language, and at the same design stage, as the engineering ones. It also exposed the limit of the approach. Without objective measurement from the floor, the model can only be as good as the literature it was built from. That limitation is what the next chapter exists to fix.

Reviewing intraoperative recordings from fifteen elective liver resections, we catalogued 154 human errors and 42 adverse events. All but one of the major events were associated with a human error, and it was failures of recognition, rather than lapses of attention, that tracked with serious harm.

Measuring that at scale needs instrumentation a surgical team will actually tolerate. Surg-STAMPS uses cap-mounted motion trackers, sensor-embedded garments, and mask-mounted microphones, compatible with a sterile field and synchronised with the operating-room black-box systems already installed. Across fifteen real procedures and thirty participants it proved acceptable in practice, and a detailed case study of an open right hepatectomy recovered stress, workload, and communication signatures that separate by surgical phase and by clinical role.

Whether such systems are deployable at all is a separate question, so we asked it directly. A nationwide survey of 831 French anaesthesia professionals, surgical professionals, and patients found broad support, with 84.8% of patients in favour, running alongside unresolved concerns about privacy and the strain of constant monitoring.

The operating room gives you realism but almost no control. This chapter trades one for the other. Puzzle-STAMPS covers 143 participants across 36 teams working a time-constrained puzzle box of thirteen physical puzzles in eight timed segments, spanning visual search, decoding, and manipulation, with a hint system that keeps teams progressing at comparable rates while leaving room for different ways of organising themselves.

Seventy-six hours of synchronised recording across nine modalities, including head and torso motion, cardiac and respiratory signals, oxygen saturation, temperature, per-person audio, and multi-angle video, sit alongside puzzle-progress logs and psychometrics covering personality, workload, cohesion, and leadership.

It is built to support work on leadership emergence, team-performance prediction, interpersonal synchrony, coordination breakdown, and robust fusion when modalities drop out mid-session. This chapter is a collaboration with Dr. Ksenia Keplinger and the Organizational Leadership and Diversity group. The first-author paper is still being written.

Trimanual control has no analogue in ordinary movement, so it was unclear how, or how quickly, people acquire it.

Working with Dr. Etienne Burdet, we trained two groups of twelve participants in virtual reality over five weeks. One group practised tasks with three independent goals, the other a single goal demanding all three limbs in concert. Both groups improved sharply, but the gains diverged: independent-goal training yielded measurably more efficient motion and lower subjective workload, while coordinated-goal training produced neither. On a separate transfer test using a gamified trimanual task, that advantage washed out and both schemes improved capability about equally.

How you structure trimanual practice, then, appears to matter less for whether people learn than for what they learn. This was also where I started caring about a question that runs through everything since: which signals from a moving human body actually tell you something about performance.