EDU – Course 1
Metrology and TinyML
Measuring Accurately When the World Shakes
How BioSonic GeoLabs builds instruments that refuse to be disrupted by noise.
A Whisper in a Stadium
Imagine trying to understand your friend’s whisper in a packed stadium while 50,000 people are cheering. Measuring instruments face this exact challenge every single day. They are designed to detect microscopic signals—such as a faint tremor in the ground, the resonance of an instrument, or structural stress in a material—while being constantly bombarded by external interference.
The Problem: The World Never Stands Still
In a controlled laboratory environment, instruments have it easy: stable temperature, zero vibration, no humming motors. Out in the field, everything changes. It gets hot and cold, the wind batters the housing, barometric pressure fluctuates, the ground shifts from saturated to bone-dry, or a cell tower broadcasts right across the frequency band. Even the instrument itself warms up and ages.
The consequence is Drift.
Imagine walking straight ahead with your eyes closed. You are certain you are walking in a straight line, but with every single step, you drift a microscopic fraction away from the center. After 50 meters, you find yourself completely off course, perhaps even in a ditch.
In the exact same way, measurement devices (cell phone compasses, drones, scales, weather stations) slowly accumulate errors that compound over time.
Until now, there have been two conventional workarounds:
Filtering out the noise or repeatedly recalibrating the device. Both have a major catch. If you simply put earplugs on in the stadium, you miss the whisper entirely; inside a measuring device, a heavy filter usually swallows the fine signal right along with the noise.
For instance, weather station sensors are typically recalibrated by a technician only every 12 to 18 months. Between those calibration windows, there is no way to know precisely when or by how much the device was drifting. This creates an unacceptable blind spot for fields dealing with nature, such as climate research, medicine, or robotics.
Our Approach: Errors Are Data
While others try to eliminate interference, we asked a different question:
What if we measure the interference right alongside the signal?
In addition to the primary signal, our devices continuously capture environmental vectors such as temperature, barometric pressure, humidity, and contact pressure. In total, we account for ten distinct disturbance variables, or vectors (see the Info Box below).
Our core engine, the Real Time Inversion Core (RTIC), processes all of this auxiliary data in real time, running the physics in reverse. It calculates precisely how much each interference vector distorts the reading and strips it out dynamically.
What remains is the pure, unadulterated value — what experts refer to as:
Ground Truth.
Calculating back from the effect to the root cause is known as Inversion.
You already know this principle from a kitchen scale: The empty plate weighs 200g. With food on it, the scale reads 480g, meaning the food weighs 280g. Once you place the plate on the scale and press TARE, the scale displays only the weight of the food—the exact net value you intended to measure.
The RTIC performs this exact operation at near-lightspeed, but simultaneously for ten different “plates,” analyzing their color, shape, weight, density, and moisture all at once. This defines our fundamental axiom:
What others discard as measurement error is our raw material.
System Architecture
Fine Senses: A multi-sensor array simultaneously captures mechanical vibrations, optical fields, electromagnetic interference, and chemical signatures, alongside precision weather telemetry (temperature, pressure, humidity).
Ultra-Coherent Clocking: A master timing reference with a rhythm that drifts by less than 50 femtoseconds. A femtoseconds is one-millionth of a billionth of a second—a timespan in which light, the fastest thing in existence, travels less than the thickness of a human hair.
Shielded Monolith: The most sensitive internal components are hermetically protected against thermal gradients and electromagnetic fields.
A Learning Core: A compact TinyML neural network tracks how the device hardware evolves over time, while a dedicated hardware accelerator (FPGA) executes the RTIC algorithms entirely on-board. This ensures maximum energy efficiency and total data privacy:
The intelligence lives on the device, not in some energy-hungry cloud datacenter.
Our Primary Focus: The Earth Beneath Your Feet
In a single handful of healthy soil, billions of microscopic life forms thrive: bacteria, fungi, and countless others.
Together, they form the soil microbiome.
They transform organic debris into fertile humus, aid in water retention, and some of these organisms can even break down environmental toxins. When pollutants — such as fertilizers, industrial waste, or accidental spills — enter the ground, this delicate balance collapses.
This is where our technology engages. Our sensors analyze the sub-surface without requiring excavation:
Where is water pooling, where are gas pockets forming, where are contaminants present, and how is the soil biome health trending?
The objective is to make every substance entering the earth visible down to microscopic scales as absolute Ground Truth. Only by knowing precisely what is happening in the sub-surface can we systematically assist it.
This process is called Bioremediation.
Restoring sick soil via biological agents (microbes) that decompose pollutants. Our sensors pinpoint exactly where intervention is required, dictate what needs to be done, and verify whether the remediation is succeeding.
What This Enables
The RTIC is not a single-purpose tool. Wherever environmental noise hinders accurate measurement, it can be deployed. Here are a few examples from our design pipelines:
Geophony & Environmental Drone Surveillance: Our flagship project features dual geosondes capable of detecting and evaluating subterranean anomalies. Upon triggering an alarm, autonomous drones launch and synchronize their telemetry with the probes, enabling real-time 3D plume mapping and migration tracking.
Bioreactors: An autonomous unit that converts plant waste into fertile humus while simultaneously heating residential buildings and storing surplus energy. The RTIC monitors internal biochemical states continuously.
Passive Submarine Sonar: “Passive” means emitting zero acoustic energy—listening like someone in the dark. Underwater environments are loud (waves, currents, shipping traffic). The RTIC strips out this ambient noise to isolate faint acoustic signatures.
Audio Engineering: A mastering console where thermal drift and component tolerances no longer compromise the signal: authentic analog warmth paired with absolute digital precision.
Structural Engineering & Materials Science: Bridges, buildings, tunnels, and structural components can be evaluated nondestructively and completely passively.
Robotics & Prosthetics: A robotic hand capable of discerning whether it is holding a raw egg, a solid stone, or gently gripping another human hand.
Surgical Robotics: Real-time tissue differentiation onboard a robotic surgical arm without relying on external cloud processing, enabling sub-cellular precision directly inside the human body.
Next-Generation Medicine: Non-invasive deep tissue diagnostics without ionizing radiation, contrast agents, or even physical contact.
Wherever conditions turn hostile, our system remains calm—because it does not fight its environment; it decodes it.
Where conventional engineering treats fluctuations as errors, we utilize them as foundational data to achieve truly uncompromised measurement records for the first time in modern history.
Innovation demands a fresh perspective — interdisciplinary, dynamic, and uncompromising.
Nature needs these instruments.
INFO BOX — For Engineers & Tinkering Minds
$$\text{True Value} = (\text{Raw Reading} – \text{Additive Distortions} – \text{Zero-Point Offset}) \times \text{Correction Factor}$$
Two vectors are subtracted additively (like accounting for the weight of the plate).
Eight vectors are scaled via dynamic coefficients (like compensating a device that reads 5% high in cold temperatures).
| Störeinfluss | Alltagsbild | Wird … |
| Störfelder von Elektronik, Motoren, Funk | Handy neben dem Lautsprecher: es knattert | abgezogen |
| Nullpunkt der Elektronik | Waage zeigt ohne Last schon 20 g | abgezogen |
| Anpressdruck | Stethoskop fest oder locker aufgesetzt | herausgeteilt |
| Flüssigkeitsdruck (Wasser, Blut) | nasser statt trockener Schwamm | herausgeteilt |
| Luftdruck | Druck auf den Ohren im Flugzeug | herausgeteilt |
| Temperatur | Gitarre verstimmt sich bei Kälte | herausgeteilt |
| Leitfähigkeit des Materials | Salzwasser leitet Strom besser als reines Wasser | herausgeteilt |
| Gase und Chemie | Rauchmelder schlägt beim Duschen an | herausgeteilt |
| Winkel und Abstand | Foto aus schiefem Winkel | herausgeteilt |
| Alterung | Gummiband, das mit den Jahren ausleiert | herausgeteilt |