Industries / Government, education and community / Education / Trade schools and vocational training
Simulators and virtual training
Trainees practise trade tasks on simulators, virtual reality headsets and screen-based trainers, so they can repeat a task without using up materials or running a real machine.
5
Milestones achieved, from 2007 to 2021.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
2007
Painters practice without paint
By 2007 the Iowa Waste Reduction Center at the University of Northern Iowa was training spray painters on VirtualPaint, a simulator that shows the film thickness, paint used, transfer efficiency (the share of sprayed paint that lands on the part) and overspray of each pass, with no paint sprayed. The center says many painters reach a transfer efficiency of 40% or less, and that its training of more than 300 military painters raised transfer efficiency by 18.42% on average and cut paint use by 13.33%; those figures cover its whole course, not the simulator alone. Sherwin-Williams now runs VirtualPaint sessions for the painters it trains.
Credit
Iowa Waste Reduction Center, University of Northern Iowa (Joe Bolick and colleagues)
2015
Practice must match the test
In five studies with students at Purdue University and expert operators at John Deere, Chung Yin So tested how people learn excavator and wheel loader controls on simulators. Students who broke off their excavator practice to practise on a simulated loader improved less at loading a truck with the excavator than students who stayed on the excavator, and practice on tasks like the final test carried over best.
LessonOn a simulator, practice that matches the real task transfers best, and switching machines midway can slow learning.
Credit
Chung Yin So, Purdue University
2020
Practical training stops
In April and May 2020 the ILO, UNESCO and the World Bank surveyed more than 1,300 respondents in 126 countries, mostly training providers. 90% reported that training centers in their country had closed completely, 98% reported that training in workplaces was disrupted, and more than three-quarters reported postponed certification exams. About two-thirds of providers taught fully remotely, but most could not teach or assess practical skills at a distance and taught theory instead; in low-income countries fewer than one provider in five managed fully remote training.
LessonRemote teaching carried the theory, but practice and practical exams waited for the workshops to reopen.
Credit
International Labour Organization, UNESCO and the World Bank
2020
A VR crane exam tested
The National Commission for the Certification of Crane Operators (NCCCO) Foundation had 40 certification candidates, all men with an average of six years of crane experience, take a virtual reality (VR) test built to follow the structure and order of the real crane exam, and then the exam on an actual crane. Wallace Judd of Authentic Testing, who designed the study, reported that 87% got the same pass or fail on both and that 5.5% passed in VR but failed on the crane.
LessonIts author said more studies are needed before a VR test could replace the real machine for certification.
Credit
NCCCO Foundation; Wallace Judd, Authentic Testing
2021
VR results pooled
A World Bank working paper by Diego Angel-Urdinola, Catalina Castillo-Castro and Angela Hoyos pooled 92 experiments from 31 studies of virtual reality training; 78 of the experiments were randomized and 29 were in vocational or on-the-job training. On average VR training beat traditional training for technical, practical and social skills, most clearly in health and safety, engineering and technical education, and VR-trained students were up to 30% more efficient with materials and time per extra hour of training. Results in general education were mixed, and the authors did not measure whether VR costs less than real labs.
LessonVR helps most where real practice is costly or risky, and its cost against a real workshop is still unmeasured.
Credit
Angel-Urdinola, Castillo-Castro and Hoyos, World Bank
Watch for
Faster technicians, by survey
Interplay Learning, which sells simulation training for HVAC, electrical, plumbing and other trades, says in its 2025 State of Skilled Trades Training report that nearly 7 in 10 companies ramp up new technicians faster with this kind of training, saving 4.3 months on average against traditional methods. The figures are what companies told Interplay.
Achieved whenAn independent study compares technicians trained with and without the simulations and times how long each group takes to pass the same practical check.
Claimed by
Interplay Learning, December 2025
Open
01
An open paint simulator, checked
A free, open-source spray painting simulator for a standalone VR headset that predicts the paint film thickness across a panel from the path of a hand-held spray gun, and a test that compares its predictions with real panels sprayed along the same tracked paths.
The briefCredit · unclaimed02
Simulator scores against the practical
A study that sets the scores heavy equipment simulators log for each student beside the same students' results on their school's practical test on a real machine, to show whether the simulator scores predict who passes.
The briefCredit · unclaimed
Open 01 · the brief
An open paint simulator, checked
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A free, open-source spray painting simulator for a standalone VR headset that predicts the paint film thickness across a panel from the path of a hand-held spray gun, and a test that compares its predictions with real panels sprayed along the same tracked paths.
Why it is still open
Paint simulators used in training are commercial. Sherwin-Williams' page for VirtualPaint lists the numbers it reports (flow rate, transfer efficiency, film thickness) but not how they are worked out, and no published check of those numbers against measured panels was found. Simulations checked against measured paint thickness exist for robots: Chalmers' IPS Virtual Paint, sold since 2014, was compared with a car fender painted at Fraunhofer IPA, but it models the rotary bell sprayers that robots carry, not a person's gun. A GitHub search on 20 November 2026 for spray painting simulators found one unstarred hobby project.
Built with what exists
Unity or Godot with OpenXR on a Meta Quest 3; a spray deposition model of the kind used to plan robot painting paths; a headset controller clamped to a real HVLP (high-volume, low-pressure) spray gun to record each pass; water-based paint on primed panels, sprayed by an instructor in a school spray booth; a wet-film comb and a dry film thickness gauge such as the Elcometer 456.
First step
Write the deposition model for a flat panel, then record one tracked pass over a real panel and compare predicted and measured thickness at nine points.
Done looks like
On 20 panels sprayed with the tracked gun, the simulator's predicted dry film thickness is within 10 µm of the gauge reading at 9 marked points per panel for at least 80% of points, with the code, recorded paths and gauge readings published.
Open 02 · the brief
Simulator scores against the practical
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A study that sets the scores heavy equipment simulators log for each student beside the same students' results on their school's practical test on a real machine, to show whether the simulator scores predict who passes.
Why it is still open
The one comparison found is the NCCCO Foundation's crane study (see Achieved): 40 experienced candidates, a VR test built for the purpose and one kind of machine. Hoshino and colleagues, testing an excavator simulator with Hermano Igo Krebs, write that they "have not yet compared" experts' performance in the simulator with their performance in a real excavator, and no study was found that uses the scores school simulators already log. Schools hold both sets of numbers: Simlog's page describes school and prison training programs that pair its desktop simulators with NCCER's heavy equipment curriculum.
safetyandhealthmagazine.comconstruction-robots.github.iosimlog.com
Built with what exists
Score exports from the simulators a school already runs (Simlog, CM Labs Vortex or Caterpillar); the school's own records of practical test results; Python with pandas and scikit-learn; written consent and removal of names before analysis. The reader handles only records, never a machine.
First step
Ask one school that runs a simulator lab for one class's simulator exports and practical results, and match them by student.
Done looks like
For at least 150 students at three or more schools and at least two kinds of machine, the published correlation between final simulator score and practical test score, and the share of practical test failures that a simulator score cut-off would have flagged in advance, with the analysis code.