New technique helps robots pack objects into a tight space

MIT researchers are utilizing generative AI models to enhance robots’ ability to solve complex object manipulation problems. These problems include tasks like packing a box with different objects. Traditional methods typically tackle these problems sequentially, which can be time-consuming. However, the researchers have developed a diffusion model, a form of generative AI, that allows robots to generate global solutions to the packing problem by considering all constraints at once. This method produces faster and more successful solutions, even for problems with novel combinations of constraints and larger numbers of objects. The researchers believe that this technique can be applied to various complex tasks in different environments, such as order fulfillment in warehouses or organizing bookshelves in homes.



Source link