LG demonstrates CLOiD robot tasks in a manufacturing environment
South Korea’s LG Electronics and NVIDIA demonstrated CLOiD performing manipulation and manufacturing tasks in an actual manufacturing environment. NVIDIA describes a development workflow connecting real-world data collection, augmented training data, robot-model training and simulation-based evaluation. The demonstration feeds a repeatable process for developing and validating the robot’s capabilities, while the partners have disclosed no fleet deployment count.
Artificial Intelligence··Midday
CLOiD performs tasks in an actual manufacturing setting
South Korea’s LG Electronics and NVIDIA demonstrated robotic manipulation and manufacturing tasks using CLOiD, LG’s robot, in an actual manufacturing environment. Manipulation refers to a robot handling objects as part of a task. NVIDIA describes the demonstration as part of its joint development work with LG. The project centres on training and validating the robot’s capabilities for manufacturing, with a workflow intended to be reused as development continues.[1], [2]
Collected data feeds robot-model training
The workflow starts with real-world data collection. NVIDIA’s Cosmos tools then augment the data used in development, before training the robot model with Isaac GR00T, the company’s robot-learning technology. NVIDIA describes these connected stages as a physical AI data flywheel: real-world information feeds the preparation of training material and the model’s development. The sequence links the information collected outside simulation to the robot’s learned capabilities.[1]
Simulation supports repeated evaluation
The development arrangement also connects onboard computing with simulation-based evaluation. Onboard computing runs on the robot itself, while simulation supplies a virtual setting for evaluating capabilities. NVIDIA says the teams are using the combined workflow to refine a repeatable process for training, evaluating and improving manufacturing robots. The announced milestone is the demonstration in an actual manufacturing environment; the partners have not disclosed how many CLOiD robots they plan to deploy.[1]
