Skip to content

About

Official repository for the User-Experienced Sound Level (UESL) protocol—a standardized framework for evaluating audible noise in wearable robotics, exoskeletons, and prostheses.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

UESL: User-Experienced Sound Level

A standard, low-cost protocol for measuring and reporting the audible noise of wearable robots (prostheses, exoskeletons, orthoses).

Users care about how loud their devices are, but there have been many different approaches to measuring sound, so published numbers can't be compared directly. UESL is the A-weighted equivalent continuous sound level (Leq,A) measured at the user's ear, during real use. You need a Class 2 sound level meter, a quiet room, and 9 trials.

📋 Read the full protocol checklist → · 🖨️ Printable PDF

Quick start

  1. Mic at the ear: Class 2 SLM (A-weighted, FAST 125 ms, ≥ 8 Hz) at the back of the head, or a binaural headset.
  2. Quiet, absorptive room: ambient < 35 dBA or ≥ 6 dB below the device; carpet or mat.
  3. Each trial: 10 s ambient (still & silent), then 20 s of steady-state use.
  4. 9 trials: 3 slow · 3 normal · 3 fast self-selected speeds.
  5. Energy-average each trial to Leq; subtract ambient if within 6 dB.
  6. Report mean ± SD over all trials & users, n trials, user mass/height, and device settings.

Benchmark data (level-ground walking)

Device Category UESL (dBA) SD (dBA) Trials
No device baseline 35.5 2.3 72
VSO – off passive 37.4 3.0 72
MBLUE ankle exoskeleton powered 37.6 2.7 72
Proteor Kinnex 2.0 quasi-passive 38.2 3.4 54
Fillauer AllPro passive 38.6 2.6 18
Ottobock Genium X4 quasi-passive 40.1 2.0 18
Nylon pants baseline 41.5 2.8 72
Ottobock Empower powered 43.6 2.3 18
VSO – on quasi-passive 43.9 1.7 72
DESR VSO passive 44.0 3.0 72
Dephy Sidekick powered 44.6 3.4 72
Open-Source Leg v2 – ankle powered 47.7 4.4 54
Össur Power Knee powered 49.5 2.6 36
Open-Source Leg v2 – knee + ankle powered 50.6 2.2 18

7 participants (4 able-bodied, 2 transtibial, 1 transfemoral). Machine-readable version: benchmarks.csv.

Preliminary design targets: in our (small, N = 7) cohort, the sound of a device began to have a moderate impact on users' willingness to wear it daily at ≈ 44.4 dBA UESL, 12.5 sone loudness (N5), and 1.95 acum sharpness (S10).

Contribute your device

Measured a device with this protocol? Please share it. Open an issue or pull request with a filled-in row of reporting_template.csv so the benchmark can grow.

Citation

Paper in preparation. Until it's published, please cite this repository (see CITATION.cff or the "Cite this repository" button on GitHub):

Z. Bons, S. Thiede, and E. J. Rouse, "Sound Matters: Standardizing Audible Noise Reporting for Wearable Robots," in preparation.

Contact

Zachary Bons · zbons@umich.edu · Neurobionics Lab, University of Michigan Supported by the NSF Graduate Research Fellowship Program (DGE-2241144).

See CHANGELOG.md for protocol version history.

About

Official repository for the User-Experienced Sound Level (UESL) protocol—a standardized framework for evaluating audible noise in wearable robotics, exoskeletons, and prostheses.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors