Figure AI Inc. demonstrated the capabilities of its Helix VLA model in household chores, with robots visually evaluating scenes and cooperating to put away groceries.
The robots work independently until they need to hand off items to each other within reach, showcasing a 'telepathic' interaction without verbal communication.
The supervisory AI architecture breaks down overall goals into subtasks while controlling each robot independently.
Figure's humanoid robots successfully handled tasks like closing drawers, refrigerator doors, and placing objects intuitively without specific instructions.
The Helix system's VLA model is highlighted as a key technology for humanoid robot manufacturers, enabling new skills to be specified using language.
Figure demonstrated groundbreaking advancements with the Helix VLA model, featuring full upper-body control, multi-robot collaboration, and 'pick up anything' capability.
The Helix model utilizes natural language prompts, a unified neural network for all behaviors, and onboard processing for reduced latency and commercial readiness.
Production trials are underway, with Figure aiming to ship 100,000 humanoid robots in the next four years and securing funding to support their commercial deployment.
The Robotics Summit & Expo will showcase humanoids, with insights from industry leaders discussing technical and business challenges in humanoid development.
The event will bring together developers to explore enabling technologies, best practices, and emerging trends in robotics across various industries.