Brett Adcock’s Figure Robotics has introduced an innovative “watch-and-learn” feature for their autonomous robots. This technology empowers humanoid robots to observe and understand human tasks, enabling them to autonomously execute a diverse range of actions and adapt to dynamic work environments. This breakthrough is poised to transform the capabilities of humanoid robots across multiple industries.
Figure Robotics aims to develop adaptable and autonomous full-body humanoid robots capable of seamlessly fitting into diverse workplaces and independently mastering various tasks. Unlike conventional robotic programming, these robots are engineered to learn through observation, heralding a new era of automation with reduced reliance on extensive programming.
About Figure 01
Figure 01 is a revolutionary autonomous humanoid robot ushering in a new era of robotics. Figure 01 is the first humanoid robot that can work on its own and is available for commercial use. It stands at 5’6″ tall, weighs 60kg, and can carry up to 20kg of weight.
With a runtime of 5 hours and a speed of 1.2m/s, it surpasses the capabilities of traditional single-function robots. Designed in a human form factor, Figure 01 combines the skill of human form with cutting-edge AI, making it a groundbreaking innovation in the field.
Efficient Learning Mechanism
Central to Figure’s humanoid robotics is its rapid learning and adaptability. In a recent demonstration, a Figure robot showcased its proficiency by operating a Keurig coffee machine and responding to simple verbal commands. Despite the human operator and machine still playing a significant role, the robot’s accomplishment is noteworthy. Similar to Figure’s humanoid robots, Google set to launch ‘Bard Advanced,’ emphasizing swift learning and adaptability.
The Figure robot acquired the necessary skills with just 10 hours of video observation, facilitating knowledge transfer to other robots through swarm learning. This learning process holds immense potential, suggesting humanoid robots can undertake diverse tasks with minimal human intervention.
Figure 01 Making Coffee
Yesterday, Brett Adcock, the creator of Figure Technologies, took to Twitter to announce the imminent arrival of the “ChatGPT moment for robotics.”
Today, on Twitter, he shared some exciting news. Figure 01, their clever robot, can now make coffee by watching people. He called this a “ChatGPT moment for robotics,” emphasizing the robot’s ability to learn this skill from human actions.
Figure 01 has learned to make coffee ☕️— Figure (@Figure_robot) January 7, 2024
End-to-end AI system, trained in 10 hours, just by watching humans make coffee
Our neural networks are taking video in, trajectories out
Join us to ship a fleet of AI robots: https://t.co/7fGEbRbeol pic.twitter.com/DNAZWnaqUv
What Is The Coffee-Making Process by Figure 01?
Figure 01, an autonomous humanoid robot, masters coffee making through an innovative process. Here’s how it works:
1. Data Collection Module
Figure 01 employs a camera as its data collection module. The camera records videos of humans making coffee, capturing various actions and details.
2. Learning Module
The learning module, a deep neural network, processes the video frames. It identifies key steps, actions, and parameters required for coffee making.
3. Recognition and Understanding
Using computer vision, Figure 01 recognizes objects like the coffee machine, mug, button, and pouring actions. Natural language processing helps it understand verbal instructions and feedback from humans.
4. Feature Extraction and Optimization
Machine learning extracts relevant features and patterns from the data. Deep learning optimizes the neural network, enhancing performance over time.
5. AI Models and Adaptability
State-of-the-art AI models, including CGCNN, PHYSNet, SCHNet, MPNN, MPNN Transformer, and TorchMDNet, enable Figure 01 to handle complex physical and chemical interactions. These models allow adaptation to different environments, considering variables like temperature, pressure, water and coffee flow, mug characteristics, coffee type, and user preferences.
6. Reinforcement Learning
Figure 01 employs reinforcement learning, a technique where it learns from its actions and feedback. It tries different actions, receives rewards or penalties based on outcomes, and optimizes its actions to maximize rewards.
Figure Robotics is moving fast, just like the OpenAI GPT Store. They both want to bring advanced AI solutions to the market quickly. Figure 01 stands out as the first commercially viable humanoid robot. Figure’s “watch-and-learn” technology goes beyond making coffee, promising broader applications.
A Focus on Figure Robotics and Industry Commitment
The swift advancements by Figure Robotics, alongside notable companies like Tesla and Agility, emphasize the imperative of introducing autonomous humanoid robots to the market. These entities are not solely engaged in research but are firmly dedicated to providing tangible solutions that can promptly enhance the workforce.
Figure’s commitment is exemplified by the swift development of the 01 robot, which achieved the remarkable feat of walking within just 12 months. This accomplishment underscores their dedication to pushing the technological boundaries in the field of humanoid robotics.
What’s The Public Reaction On Reddit?
On Reddit, people have different opinions about Figure 01’s coffee-making skills and its comparison to the “ChatGPT moment for robotics.” One person doubts the comparison, saying there’s a need for proof and evidence, dismissing it as marketing hype without solid proof.
Another person compares Figure 01’s current demonstration to early Boston Dynamics videos, suggesting it might only show its abilities in specific conditions, similar to the early stages of other robot developments.
In response to the “ChatGPT moment,” someone else suggests calling it a “transformer moment,” pointing out that Figure 01’s success in pre-recorded tasks isn’t as significant as the ChatGPT moment.
Some argue that a true ChatGPT moment in robotics happens only when the product is successfully shipped, regardless of the timeline. Concerns about the demo video’s lack of information on other robot capabilities and the potential for remote control by a person make the discussion more complex on Reddit.
The Future of Humanoid Robots
While Figure 01 may not replace baristas, it signifies a collaborative future where AI enhances tasks, pushing the boundaries of machine capabilities. In envisioning a new era, Figure Technologies aims to create robots that work alongside humans, boosting productivity.
Time will tell if these optimistic visions come true, but ongoing progress suggests that humanoid robots might soon change how we work worldwide. If successful, it could reshape our relationship with technology and impact economies for generations to come. The era of versatile robot coworkers may be just around the corner.
The ongoing advancements in this technology hint at a future where these robots become integral to various industries, automating tasks that traditionally required human expertise. Entering a new era, humanoid robots stand poised to redefine our daily existence. Recent strides in AI and robotics signal a technological and societal shift comparable to historical revolutions.
Despite concerns about job distribution and societal changes, the inevitable progress of humanoid robots will bring both positive and challenging implications as they become part of our rapidly evolving world. The future is promising, and Figure Robotics leads this transformative journey.