Researchers have developed a novel way to train robotic machines. Now, they are using AI-generated images to create training data for robots.
Called Genima, the new system fine-tunes Stable Diffusion to draw robots’ movements. It actually helps to guide robots in simulations and the real world.
Developed by researchers from Stephen James’s Robot Learning Lab in London, the system could reportedly make it easier to train different types of robots to complete tasks—machines ranging from mechanical arms to humanoid robots and driverless cars.
Researchers maintained that GENIMA is a behavior-cloning agent that fine-tunes Stable Diffusion to “draw joint-actions” as targets on RGB images. These images are fed into a controller that maps the visual targets into a sequence of joint-positions.