Gemini 4 Argon for Robotics
https://deepmind.google/blog/gemini-4-argon-for-robotics/📌 【Google DeepMind】Gemini 4 Argon 讓機器人學會「自己練習再上場」
TL;DR：Argon 在空間推理基準 ERQA 拿下 87%，並被用來串起從技能學習到模擬資料生成的整個機器人工作流程。
一個語言模型要操控機械手臂、理解接觸力學、還要在模擬環境裡自我排練，這聽起來像是分屬不同領域的三件事。但 Google DeepMind 展示的案例顯示，Gemini 4 Argon 正把這些環節串成同一條工作流。
🤔 從「看得懂」到「做得到」
DeepMind 表示 Argon 在物理理解、程式碼生成與視覺語意推理上出現明顯躍進，朝「physical AGI」邁進一步。在測試模型理解 3D 環境中物理與空間問題能力的 Embodied Reasoning（ERQA）基準上，Argon 純推理拿下 87%，優於其他前沿模型。
🧩 四大應用場景：學技能、建分身、下指令、評表現
DeepMind 展示了 Argon 在機器人生命週期中的代表性用法：
- 學習新技能：操控 Boston Dynamics 的 Spot 四足機器人（12 自由度腿部＋6 自由度手臂），在踩踏板丟垃圾前，先用 MuJoCo 打造帶鉸鏈的垃圾桶數位分身反覆練習動作。
- 視覺閉環控制：透過 Gemini Omni World Model 操控機器人，將打散的 Scrabble 字母磚重新排列成單字，過程中 Argon 會寫出帶時間戳的動作提示、檢視模擬關鍵影格，再反覆修正指令，而非下一次性的靜態指令。
- 建立數位分身：觀看一段真實機器人遠端操作錄影，自動建立外觀與接觸行為都相符的模擬資產，並以視覺推理迭代調校場景與物理參數。
- 人類動作重定向：從第一人稱操作影片中辨識關鍵物件，把簡單的邊界框換成可處理凹面容器的凸分解網格，再把人類手部軌跡重定向到一個 76 自由度、雙手各 22 自由度的雙手人形機器人上，於 MuJoCo 接觸物理中運作。
📊 用模擬資料把真實成功率從 56% 拉到 75%
Argon 也被用來當作生成式視覺-語言-動作（VLA）模型的資料來源：只靠一張真實桌面的 RGB 照片，Argon 就能自動建出校準過的 MuJoCo 模擬分身、定義新任務，並合成閉環動作產生器，收集超過 4,500 筆領域隨機化（domain-randomized）模擬情節，最終把真實機器人的任務成功率從 56% 提升到 75%，完成時間則從 35 秒縮短到 24 秒。
💡 不只是控制器，還是評分員
DeepMind 也示範 Argon 作為機器人任務的「通用價值函數」：分析一段 60 秒的雙手環形分揀操作錄影，Argon 能同步檢視頭部與左右夾爪的攝影機畫面，在接觸轉換點前後產生任務進度預測，並把整段軌跡拆成 28 個正、中性、負進度片段。另一個案例中，針對一段 136 秒的雙手摺紙影片，Argon 產出四層時間對齊的事後標註：整體目標、6 個語意子任務、17 個接觸級原子動作，以及能在操作失誤時讓任務進度持平、不會誤判的品質控制重加權機制。此外，給定一張紙本手繪草圖，Argon 也能自動建構 GPU 加速的 mujoco_warp 場景、合成 76 自由度全身動作軌跡，並驗證物理閉環中的結果。
🎯 實務啟示
對機器人團隊而言，比起單一能力提升，更值得關注的是 Argon 把「建模擬環境、生成訓練資料、評估操作品質」這幾個通常需要不同工具鏈的步驟，收斂成同一個模型的自然語言工作流程，這可能顯著降低 sim-to-real 資料生產的門檻。
🔗 來源
- 標題：Gemini 4 Argon for Robotics
- 作者／機構：Google DeepMind
- 連結：https://deepmind.google/blog/gemini-4-argon-for-robotics/
#GeminiArgon #Robotics #GoogleDeepMind #PhysicalAGI #MuJoCo #VLA #EmbodiedAI #Sim2Real #RoboticsAI #ERQA
原始資料 Google DeepMind · 收集於 2026-10-13
摘要原文
Gemini 4 Argon marks a step change in physical understanding, code generation, and visual-semantic reasoning—unlocking new workflows across the robotics stack and accelerating progress toward physical AGI. You can see this leap in capability on the Embodied Reasoning (ERQA) benchmark , which tests a model's ability to understand and solve physical, spatial problems in 3D environments. Using pure reasoning, Argon scores 87%, outperforming every other frontier model. As a general-purpose physical reasoning model, Argon can be applied across the entire robotics lifecycle. In this post, we share representative prompts for Argon across four key areas: Argon controls a Boston Dynamics Spot® quadruped (12-DoF legs + 6-DoF arm) to learn a new skill: stepping on a foot-pedal trash bin to throw away soda cans. Argon creates an articulated MuJoCo digital twin of a foot-pedal trash bin to practice its moves before stepping on the physical pedal and dropping soda cans into the open bin. Sorry, your browser doesn't support playback for this video Argon controls a simple robotic arm to place a fallen mug upright and put the crumpled napkin inside. Argon reasons with a visual coordinate harness and action primitives to perform this task over several turns. Sorry, your browser doesn't support playback for this video Argon steers robots in the Gemini Omni World Model to rearrange five scattered Scrabble tiles ( G, R, A, N, O ) into a clean row spelling A - R - G - O - N . Instead of specifying one static prompt, Argon steers in a closed visual loop. It writes timestamped action prompts, inspecting intermediate rollout keyframes, and refining its prompts. Sorry, your browser doesn't support playback for this video Argon views a real robot teleoperation episode and develops a digital twin; it obtains or creates simulation assets that look and behave like the real objects upon contact with the dexterous hands. Argon tunes the assets and scenes to match sim visuals and contact events with those in the real recorded data in an iterative visual-reasoning loop. Sorry, your browser doesn't support playback for this video Given an egocentric human video of a manipulation task, Argon identifies the task-relevant objects, replaces crude bounding boxes with compound convex-decomposed meshes for concave containers, and retargets the human hand trajectory onto a 76-DoF bimanual humanoid with 22-DoF five-fingered hands in MuJoCo contact physics. Sorry, your browser doesn't support playback for this video With Argon, a developer can snap a photo of a rough pen-and-paper sketch—showing an Apollo 2 humanoid throwing a ball along a parabolic arc to topple a 3-2-1 stack of boxes on a table—and have Argon automatically construct the GPU-accelerated mujoco_warp scene, synthesize a balanced 76-DoF whole-body and five-finger throw trajectory, verify 6/6 target knockdown in closed-loop physics, and export a Blender 4.5 Cycles render. Sorry, your browser doesn't support playback for this video Argon has precise physical understanding of dexterous manipulation and can provide detailed segmentation and text labeling from video. Argon is able to identify poor manipulation segments such as pauses, unstable grasps and fumbles. Given an overhead video of a bimanual ALOHA 2 robot folding a square sheet of paper into an origami fox over 136 seconds, Argon produces a time-aligned 4-level hindsight annotation hierarchy: Episode Goal (G1), 6 contiguous Semantic Subtasks (G2), 17 contact-level Atomic Primitives (G3), and automated Quality Control (QC) reweighting that holds task progress flat during fumbles and retries. Sorry, your browser doesn't support playback for this video To recreate the classic 12-ball balance scale puzzle with robots, Argon builds a bimanual Franka FR3 MuJoCo environment with 12 visually identical pool balls (one secretly heavier or lighter) and an articulated Roberval two-pan balance scale. It then writes a vision-only closed-loop solver that triangulates the balls from stereo RGB cameras, visually reads the scale tilt after each weighing, and isolates the odd ball into the Heavy (Red) or Light (Blue) tray within 3 weighings. Sorry, your browser doesn't support playback for this video Argon is a general value function for robotics tasks. Evaluated on a 60-second bimanual ring-sorting episode from a human UMI dataset, Argon iteratively inspects synchronized head, left-gripper, and right-gripper camera feeds around contact transitions to generate dense value function / task progress predictions and decompose the trajectory into 28 segments with positive, neutral, and negative progress slopes. Sorry, your browser doesn't support playback for this video Argon can generate new data in simulation to improve performance of generalist vision-language-action (VLA) models. Starting from a single RGB photo of the real tabletop, Argon autonomously builds a calibrated MuJoCo simulation twin, defines new tasks, and synthesizes a closed-loop motion generator to collect 4,500+ domain-randomized episodes in simulation—boosting real-world physical robot Success Rate from 56% to 75% and reducing completion time from 35s to 24s . Sorry, your browser doesn't support playback for this video On a simple 6-DoF arm, Argon iteratively self-improves both execution speed and placement accuracy on a pick and place task via trial and self-reflection. Sorry, your browser doesn't support playback for this video With a single prompt per paper, Argon reads a robotics control manuscript and any released checkpoints, ports the physics model and closed-loop controller (from 7-network hierarchical locomotion in DeepLoco and 50 Hz PPO actors in HUSKY to Web Worker iLQR in Whole-Body MPC ) into self-contained MuJoCo WebAssembly browser labs, and verifies them in headless Chromium. Sorry, your browser doesn't support playback for this video Argon bridges coding, visual perception, physics simulation, and embodied robot control. By reasoning natively over multi-view video, 3D geometry, contact mechanics, and proprioceptive telemetry, Argon understands not just what objects are in a scene, but how they physically interact with agents to accomplish whole-body dexterous tasks. Gemini 4 Argon marks an important milestone accelerating the path toward physical AGI. By building this core intelligence, our goal is to enable robotics that can work alongside humans to solve complex challenges. You can start building with Gemini 4 Argon today across Google’s developer platforms:
由 claude-code/sonnet 自動生成