Robots are not rare. Flexible deployment still is.
Tobias Schmidt (Head of Partnerships, Wandelbots) started from a plain constraint. Industrial robots have been around for decades, and millions are already at work. Mass deployment still stalls because programming them is specialist work, and that specialism is per OEM: a different controller, a different language, a different skill for KUKA, ABB, Yaskawa, and the rest. If every new job is an engineering project, plants cannot keep up with how fast products and lines change.
Wandelbots’ answer is software-defined robotics. Schmidt described a Dresden company of about 140 people, and a hardware-agnostic, API-first platform called NOVA. He put almost 400 robot models on it already, with more than 100 customers and ecosystem partners. The same layer is meant to cover stationary industrial arms, mobile robots, and legged or humanoid form factors - including Boston Dynamics Spot - so one intelligence layer can execute across bodies.
Two shifts make that plausible now, in his telling. Compute has made simulation more realistic than the digital twins of twenty years ago, and AI changes what you can do once a twin exists. NOVA sits on NVIDIA Omniverse as the simulation-first path, with a NOVA extension inside that scene.
Schaeffler: layout, virtual controllers, then the same stack on the floor
Marwin Kunz (Strategic Alliances) took the customer path Wandelbots has been on with Schaeffler, including a GTC announcement that Schaeffler uses NOVA as a physical-AI platform for factories. The journey is plan, simulate, deploy - and the runtime has to match, or the model trained in sim is not the model that moves the arm.
