{"id":150824,"date":"2025-06-17T19:27:03","date_gmt":"2025-06-17T19:27:03","guid":{"rendered":"https:\/\/teknomers.com\/en\/figure-02-has-worked-for-only-one-hour-whats-unsettling-is-that-their-brain-already-remembers-and-their-hands-feel\/"},"modified":"2025-06-17T19:27:05","modified_gmt":"2025-06-17T19:27:05","slug":"figure-02-has-worked-for-only-one-hour-whats-unsettling-is-that-their-brain-already-remembers-and-their-hands-feel","status":"publish","type":"post","link":"https:\/\/teknomers.com\/en\/figure-02-has-worked-for-only-one-hour-whats-unsettling-is-that-their-brain-already-remembers-and-their-hands-feel\/","title":{"rendered":"Figure 02 has worked for only one hour. What’s unsettling is that their brain already remembers, and their hands “feel.”"},"content":{"rendered":"\n

Imagine a world where \u00a0humanoid robots\u00a0 coexist alongside humans, not just in controlled environments, but as integral parts of our daily lives. As technology advances and costs decrease, the discussion surrounding these robots will shift from technical capabilities to \u00a0ethical considerations\u00a0. Questions will arise about their roles: How far can they go? Who will determine their limits?<\/p>\n

Picture this: robots deployed as soldiers in conflict zones, working alongside human troops; robots serving as emotional companions for the elderly or anyone living alone; robots performing mundane tasks like cleaning and cooking; and robots tirelessly operating in factories, taking on roles once held by humans.<\/p>\n

Surprisingly, many of these scenarios are no longer mere figments of our imagination. The robot \u00a0Figure 02\u00a0, employed at a BMW factory since last year, is a prime example. This humanoid can operate autonomously, classify objects accurately, and execute tasks with a combination of \u00a0\u201ctouch\u201d\u00a0 and \u00a0\u201cshort-term memory\u201d\u00a0, as its developers describe.<\/p>\n

What once seemed like the realm of science fiction, as depicted in the 2018 video game \u2018Detroit: Become Human\u2019, is coming to life. Even its creator, \u00a0David Cage\u00a0, could hardly have predicted that by 2025 we would witness such advancements in real life. But here we are, and what\u2019s truly fascinating is that \u00a0Figure 02\u00a0 does not operate in isolation. It is powered by a sophisticated neural network that is quietly revolutionizing robotics.<\/p>\n

Helix: The Brain Behind Figure 02<\/h2>\n

This neural network, known as \u00a0Helix\u00a0, serves as the brain of the robot. It enables its movements to appear increasingly natural. Within just three months of its implementation in a logistics environment, Helix has demonstrated remarkable capabilities. It has learned not only to manipulate objects skillfully but also to \u00a0understand context\u00a0 and adapt its actions based on previous experiences. Over time, it improves its speed and accuracy.<\/p>\n

\n
<\/div>\n<\/div>\n

In its early days, Helix was tasked with simpler objects. As it has evolved, it now manipulates a wide variety of items, ranging from rigid boxes to deformable plastic bags\u2014objects that are inherently challenging to grasp. Many of these bags crumple or slide easily, making them a true test of a robot’s adaptability.<\/p>\n

Yet, Helix adjusts its grip based on the specific characteristics of each object. It modifies its strategy if the package is flatter or softer and knows when to switch to \u00a0a more precise gripping tool\u00a0. This adjustment happens in real-time, without explicit instructions about the type of package in front of it.<\/p>\n

Rather than being programmed line by line, this skill was acquired through observation. Helix underwent training via \u00a060 hours of human demonstrations\u00a0. With each new example, its understanding of task execution deepened, leading to a notable decrease in the average time taken to process packages.<\/p>\n

Helix isn\u2019t just quick\u2014it\u2019s also highly accurate. The orientation of labels, which previously led to mistakes, is now correct \u00a094.4%\u00a0 of the time. How? By learning to \u00a0smooth wrinkled envelopes\u00a0 before scanning. A gentle press on the plastic is sufficient for the barcode to become visible\u2014a small yet significant adjustment.<\/p>\n

\n
\n \"Figure\n <\/div>\n<\/div>\n

Integral to this learning process is memory. Helix is equipped with a vision system that not only analyzes its immediate environment but also retains visual information from moments prior. This \u00a0short-term visual memory\u00a0 allows it to make more informed decisions. If it detects a package already rotated to a certain angle, it doesn\u2019t repeat the action. If it initially fails to locate a label, it remembers its previous attempt and retries in that direction.<\/p>\n

An additional breakthrough has involved tracking its own movements. Previously, each action stood alone: see, act, reassess. Now, Helix \u00a0remembers the positioning of its arms\u00a0, torso, or head moments ago, facilitating smoother movements. If something goes awry\u2014like a package slipping\u2014Helix can adjust its actions without starting from scratch.<\/p>\n

The latest advancement includes \u00a0a sense of touch\u00a0. Not human-like, but a method of determining how much pressure it exerts on an object. This feedback lets Helix pause a movement if it meets resistance or adjust its motion based on the object’s weight. As a result, it can handle items more delicately, accommodating variations in weight and rigidity.<\/p>\n

Forget theory; \u00a0Figure 02\u00a0 has already been put through its paces in real-world conditions. During a one-hour session, \u00a0the robot operated without interruptions\u00a0, autonomously classifying packages while leveraging all its capabilities: tactile sensitivity, short-term visual memory, and real-time error correction. These flawless demonstrations illustrate remarkable fluidity and precision in action.<\/p>\n

\n
\n
\n
\n \"Kickboxing\n <\/div>\n<\/p><\/div>\n<\/p><\/div>\n<\/div>\n

Footage from this session reveals not only a robot functioning well but a neural network successfully navigating the complexities of the physical world. It highlights \u00a0Figure 02\u00a0 making decisions, adapting, and acting with a level of autonomy that would have sounded mythical just a short while ago.<\/p>\n

As companies like Tesla with \u00a0Optimus\u00a0, Boston Dynamics with \u00a0Atlas\u00a0, and Agility Robotics with \u00a0Digit\u00a0 vie to dominate the humanoid robot market, \u00a0Figure 02\u00a0 progresses steadily along its own path\u2014quiet yet effective. As time unfolds, we will be watching closely to see how this technology evolves.<\/p>\n

Images | Figure AI<\/a><\/p>\n

For further insights, explore how Google is advancing robotics through \u00a0Gemini Robotics\u00a0 for real-world applications, or learn about NVIDIA’s multi-million-dollar \u00a0AI supercomputer\u00a0 that operates using extensive cabling.<\/p>\n


\n
General News – 2<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

Imagine a world where \u00a0humanoid robots\u00a0 coexist alongside humans, not just in controlled environments, but as integral parts of our daily lives. As technology advances and costs decrease, the discussion surrounding these robots will shift from technical capabilities to \u00a0ethical considerations\u00a0. Questions will arise about their roles: How far can they go? Who will determine […]<\/p>\n","protected":false},"author":1,"featured_media":150825,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[36399],"tags":[2144,3878,11807,2667,1476,6098,37197,9889,4315],"class_list":["post-150824","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-technology","tag-brain","tag-feel","tag-figure","tag-hands","tag-hour","tag-remembers","tag-unsettling","tag-whats","tag-worked"],"_links":{"self":[{"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/posts\/150824","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/comments?post=150824"}],"version-history":[{"count":0,"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/posts\/150824\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/media\/150825"}],"wp:attachment":[{"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/media?parent=150824"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/categories?post=150824"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/teknomers.com\/en\/wp-json\/wp\/v2\/tags?post=150824"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}