01
Measure often
1,000 samples per second on every axis, so the fastest parts of a lift are not missed.
TECHNOLOGY · VBT SENSOR
Real velocity. On your bar.
A wireless sensor that attaches to the bar and records its movement 1,000 times per second. It is MOOVIA's first device and the data source for the platform.
VBT IN ONE MINUTE
What is velocity-based training (VBT)?
Velocity-based training (VBT) uses the speed at which you move the bar to adjust the day's load.
With the same weight, if the first rep moves slower than usual, today it is costing you more. If velocity drops a lot within a set, fatigue is building up. Measuring it turns a feeling into a number you can compare with last week's.
- Mean propulsive velocity (MPV)
- the average velocity over the part of the lift where you are pushing the bar.
- Peak velocity
- the highest velocity the bar reaches during the rep.
- Velocity loss
- how much velocity drops compared with the fastest rep of the set.
- Range of motion (ROM)
- the vertical displacement of the bar during the rep.
- Load-velocity profile
- the relationship between load and velocity for each person, which makes it possible to estimate their 1RM without a max test.
HOW IT WORKS
Three steps, no cables
01
Attach the sensor
It goes on the end of the bar. No cables.
02
Train as usual
The sensor streams the signal to your phone over Bluetooth while you lift.
03
Review the set
When you finish, the app counts the reps, calculates their metrics and keeps the full signal.
FIGURES
The prototype's figures
This is what the sensor measures today, with its source and date. When the hardware changes, this table will change.
| Item | Value |
|---|---|
| Sampling rate | 1000 Hz |
| Axes | 6: 3-axis accelerometer and 3-axis gyroscope |
| Accelerometer range | ±8 g |
| Gyroscope range | ±1000 °/s |
| Connection | Bluetooth Low Energy, 15 samples per packet |
| Capture integrity | 0 samples lost in two captures of 69.4 s and 66.8 s (22 October 2026) |
| Raw signal | Kept in full with every set and exportable as JSON |
| Battery | Rechargeable via USB-C |
Current prototype, October 2026. The next version of the sensor is being designed: components and ranges will change, and this table will be updated with its tests.
METRIC STATUS
What it measures, and what it doesn't yet
Every metric has a status. It only becomes “verified” once we have checked it and can say against what and when.
Verified
- 1000 Hz capture with no losses (22 October 2026).
- Full raw signal with every set.
In validation
- Rep counting, with its own mode for the Olympic lifts since 24 October 2026.
- Mean velocity, mean propulsive velocity (MPV) and peak velocity.
- Velocity loss within the set.
- Vertical range of motion (ROM) and eccentric and concentric tempo.
- Power and force, from the load you enter.
- Estimated 1RM from the load-velocity profile.
In development
- Automatic exercise recognition.
- Phase-by-phase metrics for the snatch and the clean and jerk.
Out of scope
- Bar path in 2D or 3D (shelved on 25 October 2026). Today we measure vertical displacement.
DATA QUALITY
What we mean by data quality
02
Never invent data
If a stretch of signal is lost, it is marked as a gap: it is not silently filled in and time is not compressed.
03
Keep the full signal
Every set keeps the raw signal. If we improve an algorithm tomorrow, your set can be recalculated.
04
Publish the status
Every metric says whether it is verified, in validation or in development, and since when. What we measure is public; how we calculate it is not.
TEST LOG
Test log
The tests behind the figures on this page, newest first.
| Date | Test | Result | Status |
|---|---|---|---|
| Date: | Test: Capture integrity | Result: 0 samples lost in two captures of 69.4 s and 66.8 s, checked against the sensor's internal timestamp | Status: Verified |
| Date: | Test: Continuous streaming on the bench, sensor at rest | Result: 0 samples lost in about 2 minutes | Status: Verified |
| Date: | Test: App-to-sensor connection | Result: The app connects, sends commands and receives the signal in real time | Status: Verified |
| Date: Next | Test: Vertical range of motion against heights measured on the rack; set-by-set comparison with a commercial VBT system | Status: Pending |
Tests carried out before 22 October 2026 will be repeated: they used a firmware version that dropped samples.
Does your lab have reference equipment? Propose a collaboration
Our dated timelineDEVICES
One device today. More to come.
The bar sensor is MOOVIA's first device. The next ones will follow the same standard: publish what they measure, with which figures and the status of every metric, before asking you to trust them.
The platform that will use this dataBar sensor
Working prototypeINSIDE
Inside the prototype
What you see is the actual prototype board, exported from the electronics design project. It is not a catalogue render.
01
Development headers
Programming and debug headers. This is how we flash the firmware and probe the sensor bus on the bench.
02
USB-C
Battery charging and power during lab tests.
03
Bluetooth Low Energy microcontroller
It reads the IMU, packs the samples and sends them to the phone.
04
6-axis IMU
The heart of the measurement: accelerometer and gyroscope at 1000 Hz. Every rep's velocity starts here.
05
Power and passives
Regulation, filtering and protection: the part nobody sees, and the one that decides whether the signal is clean or noise.
06
In-house board
Designed by us, with a continuous ground plane so the board does not pollute the sensor signal.
Development revision: the programming headers are exposed because this is the board we test the sensor with. The next version is being designed.
Frequently asked questions
What is velocity-based training (VBT)?
It is training that uses bar velocity to adjust load and manage fatigue: with the same weight, if the bar moves slower than usual, today it is costing you more.
How often does the sensor measure?
1,000 times per second on each of its 6 axes (accelerometer and gyroscope). In the tests of 22 October 2026, two captures of 69.4 s and 66.8 s, no samples were lost.
How accurate is it?
We do not publish an accuracy figure for velocity yet: it is in validation. We do publish the sampling rate, the ranges and the capture integrity, with their dates.
Does it work for Olympic weightlifting?
Rep counting has had its own mode for the Olympic lifts since 24 October 2026. Their velocities are in validation and their phase-by-phase metrics are in development.
Does it track the bar path?
No. 2D or 3D bar path has been out of scope since 25 October 2026. Today we measure vertical displacement.
Can I export my data?
Yes. The test app exports every set with the raw signal as JSON.