Six channels. A shared clock.
Three acceleration axes and three angular-velocity axes describe different components of movement. Time alignment makes their relationship usable. An IMU alone does not measure intention.
CHAPTER 05 / HUMAN SYSTEMS
Intelligence beginswith the person.
A meaningful system starts with a human question. Then comes the sensor.
Our direction is to connect external sensing, local computation and understandable feedback. Explore two design concepts, then build a first signal budget for your own experiment.
Open System Studio
DESIGN DIRECTION / OUTSIDE THE LAB
Life moves.The system needs context.
A step on a staircase and a movement of the wrist can look very different to the same sensor.
An inertial measurement unit, or IMU, combines acceleration and angular velocity. It can describe aspects of motion, but interpretation depends on placement, orientation and the task. A signal gains meaning when it is compared with a defined reference.
Repeat. Reposition. Compare.
Use a known movement on a bench rig. Repeat it with different sensor orientations, compare timestamps and record drift. Then define a voluntary, non-clinical usability study before testing a wearable concept with people.
A cue you can question.
A future interface could show a simple motion event and its uncertainty. The person should be able to correct it, pause it and understand what was actually measured.
DESIGN DIRECTION / COMPUTATION CLOSE BY
A small object.A thoughtful architecture.
Keep the first interpretation close to the signal. Decide deliberately what needs to leave the device.
Our concept combines a textile interface, a removable sensing module and local processing. The exploded view is a design exploration. The proposed architecture must still be built and evaluated for power, heat, contact quality and comfort.
Make the assumptions visible
Capture only what is needed
Start with a specific question and the smallest useful set of signals. More channels also create more data to explain and protect.
Measure the actual cost
A proposed bench test would measure acquisition, processing and transmission separately. Data volume alone cannot predict battery life or response time.
Make stopping simple
The wearer needs an understandable state, a clear pause control and a choice about data retention. Local computation is one design choice, not a complete privacy guarantee.
INTERACTIVE / SYSTEM STUDIO
Turn an ideainto a testable brief.
Choose a question, combine signals and set a recording length. See the raw data budget, the limitations and the tests your concept would need. Save the result as a brief.
CONCEPT PLANNER / LOCAL CALCULATIONSSTUDIO / 01
1 min60 min
YOUR SYSTEM BRIEFMovement
Raw signal payload1.17 KiB/s1,200 bytes/s
Recording volume703.13 KiB720,000 bytes / 10 min
Active channels616 bits per sample
Share of the raw payload100%
- IMU 1,200 bytes/s
Proposed signal path
- 01IMU
- 02Synchronize + check quality
- 03Compare with known movement
- 04Human review + correction
THE PERSON IN THE LOOP
Show a possible movement event with uncertainty. Let the person confirm, correct or pause.
What the signals cannot tell you
- IMU: placement and orientation affect the signal. No direct measurement of intention.
Proposed validation plan
- Compare with a known movement and document timestamps.
- Repeat with new sensor orientations and report drift.
- Test whether the feedback is understood and can be stopped.
Open the calculation and assumptions
(6 × 100 × 16 / 8) × 600 = 720,000 bytes
Raw payload = sum of channels × samples/second × bits/sample ÷ 8. Recording volume = raw payload × seconds. 1 KiB = 1,024 bytes; 1 MiB = 1,048,576 bytes.
These are illustrative, fixed planning assumptions, not clinical acquisition settings. The environmental example assumes one temperature channel and one ambient-light channel. Each channel stores a 16-bit value. Timestamps, packet headers, storage format, compression and other overhead are excluded.
The studio calculates a data budget. It does not estimate battery life, model accuracy, latency or clinical benefit. No sensor is connected and no personal signal is collected.
Take the thinking with you.
Save the configuration, exact calculation and proposed tests.
THE NEXT STEP / A PROPOSED DEVELOPMENT SEQUENCE
- 01
Define the question
One use case. One reference. A written measure of success.
- 02
Build a bench test
Characterize the signal and its failure cases before a richer model.
- 03
Test understanding
Explore comfort, clarity, control and the ability to stop.
- 04
Publish what was learned
Report the method, limitations and result. Let evidence guide the next step.
RESEARCH CONTEXT
Build on what can be checked.
Published work informs our questions. The concepts and calculator on this page are independent design explorations; the studies below do not describe Nano Genetics products.
Muscle signals as a digital interface
A 2025 study combined a wristband with surface electrodes and models trained across many participants to decode wrist movements, finger gestures and handwriting. New users could use the models without individual calibration in the studied tasks. This was interaction based on muscle activity, not mind reading.
Read the original studyComputation woven into a textile
Researchers integrated sensors, memory, processing and communication into an elastic fibre. In a demonstration, four fibre computers in a garment classified physical activity using local neural networks. Combining their predictions improved the result compared with individual fibres. This was an experimental demonstration, not medical validation.
Read the original study