OpenAI-Powered Robot with new born startup Figure

Figure, in collaboration with OpenAI, released Figure 01, a humanoid robot utilizing a multimodal model for real-time interactions and task execution. The model processes visual and textual data, enabling tasks like object identification and garbage collection. “Our robot can describe its visual experience, plan future actions, reflect on its memory, and explain its reasoning verbally,”

Robotics company Figure and OpenAI, in just two weeks of their partnership, are achieving remarkable results. Figure’s robot, Figure 01, can now engage in full conversations with people, describe its visual experiences, plan future actions, use common sense reasoning, reflect on its memory, and explain its reasoning verbally. This progress is attributed to the integration of Figure 01 with a large pre-trained multimodal model.

In a recent update, Figure demonstrated Figure 01’s capabilities in a video where the robot identifies objects in its surroundings, offers an apple when requested for food, and explains its decision-making process. It also performs tasks such as putting away dishes and cleaning up trash, all while maintaining a natural tone of voice. Notably, these behaviors are learned and not teleoperated.

The integration of specialized AI models into Figure 01, powered by onboard cameras, allows the robot to process visual and language inputs, enabling it to respond appropriately and perform tasks based on learned behaviors. This approach is expected to accelerate the commercial timeline for Figure 01, which is still in development.

Figure’s partnership with OpenAI holds great promise, especially for clients like BMW, which plans to introduce general-purpose robots on its production lines. If Figure 01 continues to advance, BMW employees could soon have robot coworkers capable of handling tasks efficiently while also providing companionship.