Thanks to Figure’s new Helix model, folding laundry may soon be off your to-do list. In a new video demonstration shared by the company, the system powers a humanoid robot as it carefully folds towels one by one, stacking them neatly in a basket. Helix is a generalist Vision-Language-Action (VLA) model that merges perception, language comprehension, and learned control, tackling several long-standing challenges in robotics. According to the Figure company, as a series of firsts, it represents a major leap toward robots that can seamlessly interpret their surroundings, understand instructions, and perform complex tasks with human-like adaptability. In March 2025, a video showcased how using reinforcement learning, Figure achieved smoother, more natural humanoid walking, which replaced the typical rigid, mechanical gait with fluid movement. In the video, Figure’s 02 humanoid robot is tasked with folding towels. It approaches the pile with care, patiently picking up one towel at a time. The robot carefully handles each towel, roughly smoothing the fabric and folding it into a reasonably neat shape. Its movements are steady but not perfect; occasionally, it makes a fold adjustment before continuing. It puts the towels into the basket one at a time in a generally neat stack, demonstrating a focus on finishing the task rather than exactitude. The humanoid is powered by Figure’s Helix AI model, which offers a novel method of controlling upper-body manipulations. It can move its wrists, torso, head, and individual fingers at a high pace continuously. This makes it possible for more sophisticated, human-like movements and interactions.

Full report : Figure’s Helix uses VLMs to help robots follow spoken commands and adapt actions, avoiding heavy programming or repeated demonstrations.

For more see the OODA Company Profile on Figure.ai.