FLUX-mimic is a Video-Action Model (VAM) developed by mimic robotics together with Black Forest Labs. It combines the FLUX 3 generative video model with an action decoder: a model that, thanks to large-scale video pre-training, already understands the dynamics and behavior of the world is used to predict robot actions.
Unlike Vision-Language-Action models trained on static images, building on a generative video model lets a robot be fine-tuned for a specific task with as little as ~30 minutes of robot data, instead of the 30 or more hours previously required — dramatically speeding up industrial robot learning.
FLUX-mimic is being deployed at, among others, the Audi Production Lab, where it supports soft-body manipulation tasks in automotive manufacturing that were previously done manually.
The robot learning role describes software for training a robot's control policies and manipulation/locomotion skills using machine learning methods. It covers: reinforcement learning in simulation with sim-to-real transfer, imitation learning and learning from demonstration, training Vision-Language-Action (VLA) models, and fine-tuning robotics foundation models. It typically uses massively parallelized simulation environments (Isaac Lab, MuJoCo) to generate training data, then deploys the trained policies on a physical robot.
Robot Control denotes the role of software responsible for motion control, command execution, coordination of actuating elements and the direct operational logic of the robot.
Audi Production Lab — soft-body manipulation in automotive manufacturing.
