How to Measure Trampoline Time of Flight with an IMU on a Training Mat

A trampolinist finishes a routine that felt identical to the one that scored well last month — same rhythm, same height, same rotation speed on the triples. But the video split reads two-tenths of a second short of last week's number. Nothing looked different. That gap matters more than it sounds: time of flight (ToF) is one of three judged components in competitive trampoline gymnastics, alongside difficulty and execution, scored to the hundredth of a second by timing equipment built into the competition rig. Most training gyms don't own that equipment — just a trampoline with different bed tension and a coach eyeballing a stopwatch or a slow-motion phone video.

Inertial measurement units close a meaningful part of that gap. A single IMU worn at the waist can detect the exact moment an athlete leaves the bed and lands again, because the accelerometer signal during free flight looks fundamentally different from the signal during contact — not through an impact spike, but through its absence. What follows is the physics behind that detection, a step-by-step training-mat protocol, and the calibration steps that keep numbers comparable across a different bed and across sessions where fatigue shifts the picture.

Why Time of Flight Is Scored in the First Place

Competitive routines are built from ten scored skills performed without a pause, and judging panels score three components: difficulty, execution, and time of flight. ToF earns its place because it's the one number in the sport a stopwatch measures more precisely than a human eye judges rotation or form. The International Gymnastics Federation times the routine electronically, sums the airtime across all ten skills, and converts that total directly into points — an athlete who matches difficulty and execution but stays airborne longer walks away with a higher score. Elite men's routines typically accumulate 17 to 19 seconds of total flight time; elite women's routines usually land a little lower, around 15 to 17 seconds.

The physics is simple projectile motion: once feet leave the bed, only gravity acts on the center of mass, so flight duration converts directly to peak height using h = g·t²/8, where g is 9.81 m/s². A 1.6-second flight is roughly 3.1 meters of rise; a triple flying for 1.85 seconds is roughly 4.2 meters. That's why ToF works as a scoring proxy for height and effort — you can't fake a longer flight time without jumping higher — and why it's worth measuring accurately rather than eyeballing it off video.

How an IMU Detects the Flight Phase

An accelerometer doesn't measure gravity the way intuition expects — it measures specific force, the reaction force the casing feels pushing back against gravity. Standing still, an IMU reads close to 1 g because the ground pushes back on the body. The instant an athlete leaves the bed, that reaction force disappears entirely: body and sensor are both in free fall, and the accelerometer briefly reads close to 0 g on every axis — the same signature a sensor produces in an elevator with a snapped cable. That near-zero window is the flight phase, ending abruptly the moment the feet reconnect with the bed, usually as a sharp spike well above 2 g as the fabric decelerates the body.

PoinT GO's IMU samples at 1,000 Hz and flags flight onset when acceleration magnitude drops below a 0.2 g threshold, then flags landing when it rises back above 1 g. That resolves flight phases from roughly 0.6 to 2.2 seconds — a straight bounce through a high-difficulty triple — with 1 ms resolution, regardless of the frame underneath. That's what makes it useful for reconciling a gym bed against a competition rig: the sensor reads the athlete's body, not the equipment.

Why Training-Mat Numbers Don't Automatically Match Competition Numbers

The most common mistake in field ToF testing is treating a raw flight-time reading from a gym trampoline as directly comparable to the competition rig, and it quietly derails an athlete's sense of progress. Competition trampolines use a rigid steel frame with webbing tension standardized across FIG-certified equipment; most training gyms run a softer bed, a foam-pit rig, or a double-mini with different rebound. Two athletes generating identical takeoff force can post flight times 40 to 80 milliseconds apart purely from bed stiffness — nothing to do with jumping ability.

The fix isn't a universal correction factor — bed-to-bed variation is too inconsistent for that — it's a calibration baseline specific to each trampoline a program trains on. Before comparing ToF to any competition target, run a calibration set: ten controlled straight bounces at a fixed effort (verbal cue: jump for the same height every time, no tucking), log the mean flight time and its CV, and treat that mean as the gym's own reference point. Skill and routine ToF then get tracked as a delta from that baseline — the only honest way to say flight time on skill five improved six percent, independent of which bed the athlete trained on.

Step-by-Step Measurement Protocol

The protocol below adapts cleanly to a single training trampoline or a full competitive rig, and takes about 20 minutes once a program has run it a few times.

1. Warm-up: 8–10 minutes of bed familiarization, building to normal working height, plus skill-specific mobility.
2. Sensor placement: Secure the IMU at the waist, over center of mass, with the compression belt. A waist mount tracks whole-body flight phase more reliably than a wrist or ankle mount, which can register early arm-swing or leg-tuck acceleration that isn't yet true flight.
3. Baseline calibration set: Ten straight bounces at a fixed, moderate effort, at least 2 seconds of controlled contact apart. Log mean flight time and CV as the gym baseline.
4. Single-skill testing: Perform each skill in isolation three times, with full recovery between attempts (roughly 15–20 seconds standing rest). Record flight time per attempt, not just a single best rep — a lone maximal attempt is more prone to a mistimed, slightly early landing.
5. Full-routine testing: Run the complete ten-skill routine at competition tempo, without pausing, and export the full flight-time series. This mirrors what a judging panel scores, since isolated skill testing runs 5–8% higher than the same skill mid-routine under fatigue.
6. Session logging: Record total routine ToF, per-skill ToF, and baseline CV together. A session where CV spikes above roughly 4% is a fatigue or footing flag; interpret its numbers cautiously.

Accuracy of Flight-Time Methods Compared

Flight-time measurement is well studied in jump testing, even though trampoline-specific IMU validation is still thin on the ground. The table below draws on vertical-jump flight-time literature alongside PoinT GO's internal comparison against a FIG-spec timing mat, since the underlying principle — free-fall onset to landing impact — is identical off the floor or off a trampoline bed.

The consistent finding across methods, going back to the original flight-time formula from Bosco, Luhtanen, and Komi (1983), is that this approach is accurate in principle but sensitive to landing posture: Kibele (1998) showed a change in knee-flexion angle at landing versus takeoff can bias the height estimate by several centimeters even on an identical jump. That's a real constraint on trampoline testing too — an athlete landing with more knee bend on a fatigued rep registers a longer flight time without jumping any higher, so watch landing posture, not just the ToF number, when a session's readings look inconsistent.

Interpreting Flight-Time Values by Skill and Level

These ranges come from projectile-motion norms and publicly reported FIG scoring patterns, not a single controlled study — a full-routine dataset at the individual-skill level isn't publicly available.

• Basic straight bounce (developmental): 0.7–0.9 s, roughly 0.6–1.0 m of rise.
• Single somersault, competitive youth: 0.9–1.1 s, enough airtime for one full rotation with a margin for spotting the landing.
• Double somersault, junior/senior national: 1.2–1.5 s.
• Triple somersault, elite senior: 1.6–1.9 s, the range most elite men's high-difficulty skills fall into.
• Full ten-skill routine (elite): Roughly 17–19 s for men, 15–17 s for women.

These are reference bands, not pass/fail cutoffs — an athlete's own gym-specific baseline matters more than where they sit against these ranges, since bed stiffness, body size, and training age all shift the raw number independent of ability.

Common Errors That Distort Flight-Time Readings

• Comparing raw numbers across different trampolines. Bed tension alone can shift flight time by 40–80 ms — anchor to a gym-specific calibration baseline instead.
• Reporting a single best attempt. A maximal rep more easily includes a mistimed, slightly early landing; average three to five attempts per skill.
• Mixing isolated-skill and mid-routine data. The same skill performed cold reads 5–8% longer than mid-routine under fatigue — log the two separately.
• Ignoring landing-posture drift. A softer landing with more knee flexion late in a fatigued session inflates flight time with no real height gain (Kibele, 1998). Watch video alongside the sensor trace when a number looks unusually high.
• Loose sensor mounting. A belt shifting more than a couple of millimeters mid-session adds vibration artifacts near the zero-g threshold that can register as a phantom short flight.

Building Session-to-Session Consistency

A single flight-time number, taken alone, tells a coach almost nothing about whether training is working. What matters is whether it moves in a predictable direction across weeks, tested the same way. Building that consistency means holding three things constant every session: the trampoline used, the point in the session testing happens (immediately post-warm-up is standard, since later fatigue inflates flight time via the knee-flexion mechanism above), and the calibration set anchoring the gym baseline.

A useful rule of thumb: if the ten-bounce calibration set's CV sits at or below roughly 3%, that session's skill-level ToF is trustworthy. Above roughly 5%, something in the day's conditions — fatigue, an unfamiliar surface, a loose sensor — is adding noise large enough to swamp a genuine training effect; log those numbers but flag them rather than acting on them. A program that ignores the calibration CV ends up chasing noise half the time it thinks it's chasing progress.

Key References

• Bosco, C., Luhtanen, P., & Komi, P. V. (1983). A simple method for measurement of mechanical power in jumping. European Journal of Applied Physiology, 50(2), 273–282.
• Kibele, A. (1998). Possibilities and limitations in the biomechanical analysis of countermovement jumps: A methodological study. Journal of Applied Biomechanics, 14(1), 105–117.
• Casartelli, N., Müller, R., & Maffiuletti, N. A. (2010). Validity and reliability of the Myotest accelerometric system for the assessment of vertical jump height. Journal of Strength and Conditioning Research, 24(11), 3186–3193.
• Glatthorn, J. F., et al. (2011). Validation of a simple method for measuring jump height. Journal of Strength and Conditioning Research, 25(2), 556–560.