Algorized shows AFM, a foundation model for sensing people, at CIIE 2026
Algorized is presenting its Algorized Foundation Model (AFM), which gives machines an understanding of people using standard Wi-Fi, UWB or mmWave sensors, at the 9th China International Import Expo (CIIE 2026) in Shanghai from 5 to 10 November. The company says one model can serve robots, buildings and connected devices.
- Algorized presents its Algorized Foundation Model (AFM) at CIIE 2026 in Shanghai, 5 to 10 November.
- AFM uses standard Wi-Fi, UWB or mmWave sensors and runs at the edge to detect, track and sense breathing and heart rate of people.
- Algorized says it stays anonymous because it relies on wireless signals alone, not cameras.
- The company exhibits as an invited guest at the BMW Technology Office China start-up area; no BMW collaboration or endorsement is implied.
- A live demo shows a simulated robot arm adjusting its speed to a visitor's distance.
One model for many machines
According to Algorized, perception systems for people have typically been built separately for each task: robot arms, mobile robots, buildings and connected devices each have their own. The company's approach is to build the understanding of people once, as a foundation model that every new product can start from.
The company says its research began with a search-and-rescue question: whether a wireless signal can find a heartbeat in earthquake rubble. That work grew into a foundation model for everyday settings.
AFM works from standard Wi-Fi, UWB or mmWave sensors and runs at the edge, on the device itself. Algorized says it runs across sensor types and silicon platforms, so robot, machine and device makers can build on one perception foundation.
How AFM works and where it applies
The sensor sends out low-power wireless signals that pass through clothing and reflect off the body. From those reflections, Algorized says, AFM tells a person from an object, locates and tracks each person in real time, and picks up the sub-millimeter chest movements that come with each breath and heartbeat. The company adds that it perceives people through darkness, smoke and dust.
Because the model works from wireless signals alone and understands people by how they move and breathe, Algorized says every person stays anonymous: "privacy preserved by physics, not by policy."
- Robotics and production: a robot arm or mobile robot becomes aware of nearby people and can adapt its speed and path while workers share the floor.
- Smart buildings: the model senses presence and people count, so heating, ventilation and cooling can follow actual occupancy.
- Connected devices: breathing and heart rate become non-diagnostic well-being features.
The release points to the scale of the robotics market in China, which installed 354,000 industrial robots in 2025, 59% of all installations worldwide, according to the International Federation of Robotics (IFR) World Robotics 2026 report.
Live demo and the BMW connection
On the stand, AFM runs live on data from one standard wireless sensor. When a visitor walks up, AFM classifies them as a person and tracks them in real time. A simulated robot arm adjusts its speed to the visitor's distance, waits when they come close and resumes its task as they step back. A visitor who stands still can see their breathing and heart rate on screen.
Algorized takes part as an invited guest in the start-up exhibition area hosted by BMW Technology Office China, which identified the company through outside-in scouting because of its offering in human-aware perception for Physical AI. The release states that participation alone does not mean any collaboration with BMW exists or that BMW endorses or validates Algorized's technology. AFM and all statements about its capabilities are Algorized's own.
Natalya Lopareva, CEO and founder of Algorized, said: "Everyone assumes we built five models. We built one, and every machine that works around people can build on it."
What it means for the market
Algorized describes itself as building the edge-native nervous system for Physical AI. It says the model ships as two products: a foundation model stack embedded on partner silicon across automotive, spaces and industrial applications, and a quality-managed safety stack for human-robot collaboration. The company says it is backed by the Amazon Industrial Innovation Fund and partners with silicon manufacturers and industrial OEMs.
For device and robot makers, the pitch is a single perception layer based on standard wireless sensors instead of separate systems per product. The capabilities remain the company's own claims, and the CIIE demo gives visitors and press a chance to see them live. Meetings can be booked through Algorized's media contact.
