The German company FMC³ Robotics has introduced Doceo, a set of four technologies that collect data from factory workers as they do their usual jobs so that it can be used to teach industrial robots.
Doceo RealCap records visual, depth, movement and touch data with smart camera headsets, external cameras and optional gloves. Doceo Hub synchronises and annotates the data, Doceo Cloud handles labelling, model training, simulation and testing using technology from Vector Informatik, and Doceo Driver links the resulting skills to different robots.
Joachim Fetzer, the company's president for Europe, said the shortage is not of data in general but of qualified physical-world data showing how a task is really done. FMC³ is testing the approach in two proof-of-concept projects: a wheeled robot that moves bins to production lines at a consumer electronics plant, and a robot that handles electrode sheets at a battery cell factory.
FMC³ is also planning a German joint venture with Vector Informatik and PEM Motion to set up standard data infrastructure for robots that learn in industry.
Background
Doceo
Robot training data system
A set of FMC³ Robotics technologies that collect data from factory workers to train industrial robots.
