Ask what is being measured before asking what is possible. “Augmented reality in sports training” is not one capability — it is a stack, and the parts of that stack sit at very different levels of readiness. Marker-based drill overlays and camera-based movement capture feeding post-session review are working today. Real-time, headset-worn coaching guidance that holds outdoors, through occlusion, at low enough latency to correct a movement mid-rep, is not the same product and does not arrive on the same schedule.
That conflation is the usual reading error here. Demo footage compresses the two into one impression, and a budget request follows an undated capability claim.
What does augmented reality in sports training mean in practice?
In real use, it means one of three things, and the difference decides the hardware bill:
- Marker-based overlays — projected or tablet-rendered targets, cones, lines and timing gates. Calibration is fixed, the scene is controlled, and the failure modes are known.
- Markerless movement capture — cameras plus pose estimation, usually consumed after the session as review material. Latency is not on the critical path, so accuracy can be traded for it.
- Real-time worn AR guidance — a headset rendering corrective cues while the athlete moves. This is the part gated by hardware, calibration and outdoor lighting, and it is the part most forecasts are implicitly about.
A vendor claim that does not say which of the three it belongs to is not yet a claim you can plan against.t.
The metrics that decide whether an AR drill works
| Constraint | What to measure | Why it decides the drill |
|---|---|---|
| Responsiveness | Motion-to-photon latency, in milliseconds (a working threshold used in practice is under ~20 ms) | Above it, a corrective cue arrives after the movement it was meant to correct |
| Spatial accuracy | Joint-position error in centimetres against a reference capture system | Technique feedback is only as good as the skeleton it is drawn from |
| Robustness | Tracking uptime across a full session, under outdoor lighting and occlusion | A system that drops tracking mid-drill costs coaching time rather than saving it |
| Operational cost | Setup time per drill | Determines whether a coach uses it twice or once |
Each row is a number a supplier can be asked for. Claims that cannot be tied to one of these belong on a watch list, not in a planning document — hedging against an unmeasurable claim is a way of paying for it without deciding to.
Testing a forecast rather than reacting to it
Forecasts about sensor and headset cost curves and forecasts about real-time coaching capability behave differently: the first tracks a manufacturing trend, the second depends on calibration and computer-vision robustness that no price decline resolves. Treat them as separate claims with separate horizons.
For any AR-in-sport prediction, name three things: the unit of measurement, the time horizon, and the observation that would retire the claim. Applied here, “AR coaching overlays will replace conventional drill instruction” is retired by evidence such as sustained joint-position error too large for technique correction, or tracking hold that fails in normal outdoor conditions across a season of use. We use the same unit/horizon/falsifier discipline described in how to evaluate predictions about AI, and the underlying engineering — pose estimation, calibration, tracking under real lighting — is ordinary computer vision work rather than anything exotic. Our wider view of where these capabilities sit is on the TechnoLynx site.
We state no dated prediction of our own. What is worth carrying forward is narrower: the AR training claims that already map to a metric are procurement-relevant this year, and the rest are watch items until someone names the number that would settle them. Which of the four metrics above would your shortlisted vendor be willing to publish?
