Virtual Reality (VR)
Rehearse the hard moments before they are real.
Virtual reality earns its cost when real practice is dangerous, expensive or rare, and when a space needs judging at full scale before it is built. ColdAI designs VR scenarios backwards from what a person must be able to do afterwards, sets fidelity to match that objective, and plans the comfort, accessibility and operating routine that keep a program useful long after launch.
The opportunity
Start with a decision
worth improving.
A plant needs operators to practice an emergency shutdown that cannot be rehearsed on live equipment. A VR scenario recreates the control room and records each decision for the debrief.
From possibility to a working system
Follow the flow.
Three connected decisions. One considered architecture.
Conceptual flow, not a live system or measured result.
Set the learning objective
Define the skills, decisions and errors the scenario must expose, and how competence will be assessed. VR earns its cost where real practice is dangerous, expensive or rare.
Build the environment
Model the space and equipment at the fidelity the objective requires, and no more. Design locomotion, interaction and session length to limit motion discomfort.
Debrief and operate
Record actions and timings for structured debriefs. Manage headsets centrally for content updates, hygiene, charging and device security across sites.
The work, made concrete
What we can build
with your team.
We agree scope, dependencies and acceptance criteria before delivery. Your project can start with an assessment, a pilot or an integration into an existing system.
Scenario design with assessment criteria
Clarify the system boundary, the responsible owners and the decisions the architecture needs to support.
An immersive training or review environment
Turn the agreed design into reviewable work, evaluated against representative inputs and explicit success criteria.
Debrief analytics and a fleet operations plan
Make the next step operable: document responsibilities, known limitations and the path from pilot to ongoing use.
In depth
Go deeper: evidence and comfort
Two questions decide whether a VR program lasts: can you show that it works, and can everyone who needs it use it comfortably?
- 01GuideMeasuring VR training effectiveness and ROIHow to show whether VR training works: observable objectives, Kirkpatrick levels for VR, fair telemetry, comparison groups, retention checks and a cost model.
- 02ChecklistVR comfort and accessibility design checklistA practical checklist for reducing VR motion sickness and widening access: locomotion, performance, session length, captions, seated input and alternatives.
Deciding whether a skill belongs in a headset
VR is one training medium among several, and ColdAI positions it as preparation for hands-on practice rather than a substitute for it1. These situations show where it adds something the alternatives cannot.
- If
Practicing for real would put people, equipment or the environment at risk, as with an emergency shutdown, a confined-space entry or a fire response.
ThenVR is a strong candidate for rehearsal before supervised practice.
Learners can make the mistake, see its consequence and try again without anyone being harmed.
- If
The equipment is costly to take out of service, or the situation arises so rarely that people forget what to do.
ThenUse VR for refresher practice, timed before the skill is likely to be needed.
A scenario can be repeated on demand, which the real event cannot.
- If
The skill is mostly knowledge: rules, definitions or a fixed sequence with no spatial or time pressure.
ThenUse e-learning, video or a job aid instead.
Immersion adds cost and set-up time without testing anything a short quiz cannot.
- If
Success depends on touch, force or fine hand control, such as welding or suturing.
ThenPair VR for the decisions with physical practice for the motor skill.
Controllers and hand tracking do not reproduce resistance or texture well enough to build that skill alone.
- If
People at different sites must review a design or rehearse a coordinated response together.
ThenUse multi-user sessions with voice and shared annotation.
Everyone stands in the same full-scale model without traveling, and the session can be replayed for those who missed it.
From learning objective to working scenario
A scenario is a designed assessment with an environment around it, not a 3D model with tasks added afterwards. This order keeps the objective in charge of the build.
Analyze the task with the people who do it
Walk through the real task with experienced practitioners and supervisors. Capture the cues they notice, the decisions they make, the common errors and the rare errors that are serious.
Write observable performance criteria
Turn each objective into something a scenario can detect: an action taken, an order followed, a hazard spotted within a time limit. If nobody can describe a pass, the scenario cannot assess it. The training effectiveness guide shows how these criteria feed evaluation.
Set fidelity to the objective
Decide which parts of the environment must behave realistically and which can be simplified. A valve that must be turned in the right order needs functional accuracy; the paint on the wall does not.
Script branches and consequences
Plan what follows each critical decision, including delayed consequences that surface later in the scenario. Linear scripts teach people to follow arrows; branches test judgment.
Build comfort and access in from the start
Settle locomotion, session length and seated or standing use before modeling begins, and plan an equivalent route for people who cannot use a headset. The comfort and accessibility checklist lists the decisions.
Design the debrief
Decide what the facilitator will show after each session: decisions, their order, timings and moments of hesitation. A replay that starts a discussion is worth more than a single score.
Pilot with representative learners
Test with novices and experts, people who wear glasses and people who have never used VR. When experienced staff fail a step, the fault usually lies in the scenario rather than in their skill.
Five kinds of fidelity and when each is worth paying for
| Fidelity | Worth investing when | Safe to simplify when | Effect on the build |
|---|---|---|---|
| Visual | Learners must recognize a specific gauge, label, leak or warning sign | Recognition is not part of the objective | Detailed models and textures, and a heavier performance budget on standalone headsets |
| Functional | Equipment must respond correctly to the learner's actions, in sequence | The task is about where to go or what to notice, not how to operate | Logic and state for every control, usually the largest engineering effort |
| Physical | Posture, reach or tool handling matter to the objective | Decisions matter more than movement | Props, tracked tools or a physical mock-up aligned with the virtual scene |
| Environmental | Noise, smoke, low light or time pressure change how people perform | The task is normally done in calm conditions | Audio design, effects and timers that must respect comfort and photosensitivity limits |
| Social | The skill involves other people: a team drill, a handover, a difficult conversation | The task is done alone | Multi-user networking, or scripted characters with believable responses |
Fidelity is a cost decision. Matching each dimension to the objective keeps both budgets and frame rates under control.
Full-scale design review and multi-site sessions
VR is also a review tool. Walking through a ward, a control room or a production line at full scale lets clinicians, operators and maintenance staff spot problems that drawings hide: a door that blocks a trolley route, a valve nobody can reach without a ladder, a screen placed in glare. People who will work in the space can take part even if they cannot read plans.
The engineering sits in the data path. Design models arrive from CAD or BIM tools and must be simplified for a headset without losing what reviewers need to judge, such as clearances and access zones; the spatial computing practice covers that conversion. Findings should not stay in the headset either. Comments captured in a session can be exported to the design team's issue tracker, for example in the BIM Collaboration Format (BCF), a specification maintained by buildingSMART International, so each finding points to a location in the model2.
Multi-user sessions add networking, avatars and voice, and with them governance questions: who may join, whether sessions are recorded and how long recordings are kept. Agree those rules before the first cross-site session, not after it.
How a VR program runs after launch
Device logistics such as enrollment, updates and cleaning sit with the extended reality practice. This cycle is the training side.
- Release a version
Each scenario version is tested on the target headsets before learners see it.
- Schedule sessions
Sessions are booked with a facilitator, a cleared space and charged devices.
- Run and record
Learners complete the scenario while decisions and timings are logged.
- Debrief
The facilitator replays key moments and agrees what to practice next.
- Review the evidence
Program owners look across sessions for steps most learners miss.
- Revise content
Scenario faults are fixed and procedure changes go into the next version.
Frequently asked questions
Does VR replace hands-on training?
No. VR is best used to prepare people for supervised practice and to keep rare skills fresh between real exercises. It lets learners make decisions under realistic conditions and see the consequences safely, but it cannot reproduce the full physical feel of tools and equipment. Most programs use VR before or between hands-on sessions, and keep the practical assessment that confirms someone can do the job.
Is VR training suitable for every employee?
Not for everyone, and a program should not assume it is. Some people feel motion discomfort, some find headsets hard to wear with glasses or hearing aids, and some have conditions their manufacturer's health and safety warnings advise against. A well-run program offers comfortable defaults, short sessions, seated options and an equivalent non-headset route, so that nobody is disadvantaged because the medium does not suit them.
What do we need to provide before a VR scenario can be built?
Three things matter most: a skill that is genuinely hard to practice for real, subject-matter experts who can say what good performance looks like, and the space and time for people to use headsets. Useful extras include photos, drawings or CAD models of the environment, existing procedures and incident reports, and access to the real site so designers can see the cues experienced staff rely on.
Should we build custom scenarios or buy off-the-shelf VR courses?
Off-the-shelf courses suit generic skills such as general safety awareness, where your site's layout and procedures do not change the right answer. Custom scenarios make sense when the task depends on your equipment, your procedures or your incident history, or when results must feed your own assessment records. Many programs mix the two, and either way you should check the content format, data ownership and how updates are delivered.
Can existing CAD or BIM models be used for VR design review?
Usually, with preparation. Design models carry far more detail than a standalone headset can render smoothly, so they are simplified, split into sections and given materials before review. Metadata such as room names or equipment tags should be kept, because reviewers need it to describe what they find. Large or confidential models may need a tethered high-performance set-up or controlled hosting rather than copies on every device.
Sources
- Virtual Reality (VR): scenario design, comfort and managed headset fleets — ColdAI
- BCF-XML: XML specification for BIM Collaboration Format — buildingSMART International
Connect the architecture
The next connection.
Explore the complementary capabilities that can turn an individual technology into a complete workflow.
Your next move
Bring us the
real problem.
A workflow that takes too long. A system that cannot connect. An idea that needs a technical path. Start there, and we can define what to investigate, build and measure.
- A skill that is hard to practice for real
- Subject-matter experts to define success
- Space and time for headset sessions