BLE postural monitoring device
BLE sensors with movement analysis algorithms for specialist rehabilitation. Certification for medical use, interface for clinicians, secure cloud synchronization compliant with GDPR.
Intelligence goes where the data is born.
Models run on the device, not only in the cloud: less latency, less data in transit, more privacy. Nodes coordinate in a distributed network and talk to Nebula when needed.
Sensors: intelligence where the data is born.
Models run on the device, not only in the cloud: less latency, less data in transit, more privacy. Nodes coordinate in a distributed network.
The model runs on the node: the decision is born where the data is born.
Sensors wake up on the network and send only what is needed.
Nodes exchange messages and coordinate without a single center.
At the center, the platform collects, governs and returns the results to the services.
Why it matters.
Sending all data to the cloud costs bandwidth, energy and privacy. Often the useful decision can be made on the device: recognizing a pattern, filtering noise, sending only the important event.
Edge AI moves processing to where the data is born, with models small enough to run on a microcontroller. The result is less latency, less traffic and sensitive data that stays on site.
How it works, layer by layer.
From sensor to platform, in five layers.
- Sensor
- Measures physical quantities such as movement, temperature, force or position, with low power consumption.
- On-board processing
- A microcontroller filters the signal and applies a compressed, often quantized, model to recognize the event.
- Communication
- Bluetooth, low-power cellular networks or Wi-Fi send only events and summaries.
- Coordination
- Nodes exchange messages over a private network and do not depend on a single centre.
- Platform
- Nebula collects, governs and returns results to services, and distributes updates to devices.
The model on board, the platform at the center.
Each node decides on its own what it can; the rest goes up to the platform. Less data in transit, more privacy, less latency.
- Edge AI
- Models on microcontrollers and low-cost boards, such as ESP32 and Arduino.
- IoT
- Standard sensors and protocols, from BLE to low-power cellular networks.
- Wearable
- Wearable devices for biotelemetry and movement analysis.
- Distributed network
- Nodes that exchange messages without a single center, on a private network.
- Integration
- Data arrives in Nebula, which governs it and returns it to the services.
Use cases.
Anomaly detection on a machine
A node with an accelerometer recognizes an abnormal vibration by itself and reports only the event.
Autonomous environmental nodes
Battery-powered sensors measure temperature and humidity and wake up only to transmit, staying active for a long time.
Data, integrations and standards.
Firmware and models are written for microcontrollers such as the ESP32 and Arduino boards. Devices are updated remotely and communicate with open protocols; the platform tracks the version of every node.
Privacy, security and compliance.
- Data stays on site
- On-board processing means only events and aggregates are sent, not raw signals.
- Authenticated devices
- Every node has an identity and communications are encrypted: an unrecognized device does not join the network.
- Health data
- When sensors concern health, the GDPR rules for special categories of data apply and, for medical devices, the dedicated regulation.
How to start.
- 01
Analysis and scope
We define what to measure, how often, and with what constraints of energy and size.
- 02
Pilot on a real case
We build a test node and collect real data to train and verify the model.
- 03
Extension
We shrink the model and its power consumption, and prepare remote updates.
- 04
Operation and support
We prepare the platform to manage a fleet of devices, with monitoring and support.
Words worth knowing.
- Edge AI
- Artificial intelligence that runs on the device next to the data, instead of on a remote server.
- Microcontroller
- A small, low-power programmable chip that reads sensors and drives devices.
- Quantization
- A technique that reduces the precision of a model's numbers so it fits in little memory.
- Remote update
- The ability to install new firmware on devices without physically intervening.
Frequently asked questions.
What is the difference between Edge AI and cloud?
In Edge AI the model runs on the device, in the cloud on servers. On the device the response is immediate and data stays local; in the cloud larger models can be used.
How long can a battery last?
It depends on how often the device measures and transmits: with long sleep cycles, months or years; with continuous transmission, days.
Can you work on the hardware we already have?
Yes, if it has a programmable microcontroller or a data interface. Otherwise we choose standard, easy-to-source components together.
The other areas.
People
Museums, kiosks and proximity marketing: tools that speak to people in the right place.
Open the pagePlaces
Indoor and outdoor localization, augmented reality search and contextual digital signage.
Open the pageTerritory
Multi-tenant GIS for public bodies: maps, open data and aggregated flows of people.
Open the pageProcesses
Predictive maintenance and statistical process control on machines and lines.
Open the pageSoftware
Custom software and cloud platforms, one sheet at a time.
Open the pageDo you have a device you want to make smart?
Tell us what it measures and where it must decide: from sensor to model.
Let's talk