NVIDIA GTC Berlin 2026: A Look Back - Recap Video

Why do humanoid robots need functional safety? We asked this question to visitors at NVIDIA GTC Berlin. The answer was clear: If safety is important when people work together, then it's even more important when a humanoid robot works alongside a human.

On October 21 and 22, 2026, Synapticon exhibited at NVIDIA GTC Berlin at STATION-Berlin, at our partner NexCOBOT's booth (#7040). Our recap video summarizes the two days in the words of the people who were at the booth.

THE DEMO: SAFE HUMAN DETECTION ON NVIDIA IGX THOR

We showcased Safe Human Detection, a component of POSITRON Safety AI®, live on NVIDIA IGX Thor. The setup was intentionally simple: a robot head with a camera, a monitor, and the computing hardware. Anyone who stepped in front of the camera could see in real time how the system detects people and how a simulated humanoid reacts when a person gets too close. What looks like a camera image is actually part of the security layer.

Dr. Florian Weißhardt, Head of Synapticon Intelligence, and Dieter Volpert, Senior Functional Safety Engineer, spoke at the booth about certification pathways and how AI and safety can run on a shared computing target.

WHAT WE HEARD AT THE BOOTH

For the video, we asked visitors what it would take to bring Physical AI into the real world. Four ideas emerged from the conversations:

Hardware and software go hand in hand. Robots don't think for themselves; they are controlled by software. It is precisely this software that needs guidelines, and those guidelines must work in tandem with the hardware.

Trust is essential. People must be able to feel confident that a new technology will not harm their health or their livelihoods.

Small gaps can have a big impact. Even minor security flaws can undermine trust in an entire ecosystem.

Safety comes first. One interviewee summed it up perfectly: It is the most important issue when humanoid robots become part of our everyday lives.

OUR PERSPECTIVE

A humanoid robot must be designed safely before it is allowed to work alongside humans. Functional safety is no longer an afterthought in Physical AI; it is one of the industry's core concerns. With humanoid robots, simply turning them off is not enough: an actively balanced robot that is cut off from power will not come to a stop-it will fall. Safety must therefore be built into the entire system, from the actuators to the sensors.

Thank you to NexCOBOT, who welcomed us to their booth as part of the NVIDIA Halo ecosystem, and to everyone who stopped by and shared their thoughts with us.

FREQUENTLY ASKED QUESTIONS

What did Synapticon showcase at NVIDIA GTC Berlin 2026?

A live demo of Safe Human Detection on NVIDIA IGX Thor at the NexCOBOT booth (#7040): a robot head with a camera, a monitor, and a simulated humanoid that reacted to visitors in real time.

What is Safe Human Detection?

Safe Human Detection is the camera-based perception layer of POSITRON Safety AI. It detects people in a robot's vicinity so that the robot's protective response becomes part of the safety system, rather than relying on safety fences or light curtains.

Why is safety different when it comes to humanoid robots?

A humanoid actively maintains its balance. If you cut off its power, it won't come to a safe stop-it will fall. That's why safety must be built into the entire system, from the actuators to the sensors.

Would you like to discuss safety for your robot? Email us at [email protected]. Learn more about the platform: POSITRON Safety AI