Skip to content
ContentLora

    Tip: press / anywhere to search.

    ● Developing story

    Robotics and embodied AI tracker: milestones to watch

    This tracker follows the modern era of robot learning, which began when RT-2 showed in July 2023 that a vision-language model could output robot actions.[1] As of October 2026, whole-body humanoid models,[2] early factory deployments,[3] a US restriction on foreign-made robots[4] and a record Chinese humanoid IPO[5] define the frontier.

    Editor reviewedUpdated Robotics and embodied AIArtificial intelligence

    What we know

    • Google DeepMind's Gemini Robotics 2 (July 2026) controls whole humanoids, including walking; access to the action model is limited to early partners.[25]
    • About 15,000 humanoid robots shipped worldwide in 2025, most from Chinese makers.[8]
    • Figure 03 began logistics work at BMW's Spartanburg plant in June 2026.[3]
    • Tesla was installing Optimus lines at Fremont in July 2026, with production anticipated later in 2026.[22]
    • On 28 July 2026 the US FCC added foreign-made humanoids, quadrupeds and other mobile robots to its Covered List, blocking approval of new models; the first exemptions came in September.[4]
    • Figure says its BotQ factory reached one Figure 03 an hour in April 2026, after delivering more than 350.[21]
    • NVIDIA's GR00T N1.7 is generally available under the Apache 2.0 licence.[17]
    • Waymo reports more than 270 million fully autonomous miles through June 2026.[28]

    What we don't know yet

    • Whether NVIDIA ships GR00T N2 by the end of 2026, and how it performs in independent tests.
    • Whether Tesla starts Optimus production at Fremont in 2026 and unveils Optimus V3.
    • Whether Physical Intelligence's reported funding round at more than $11 billion closed (unconfirmed as of October 2026).
    • Which foreign robot makers win Department of War exemptions from the FCC listing, and whether China retaliates against US robot makers.
    • Whether any humanoid deployment reaches thousands of units at a single customer.
    Story status

    Story status

    State
    developing
    Started
    2023-07-28
    Timeline entries
    34
    Last entry

    Timeline data (JSON)

    This tracker records verifiable milestones in robot foundation models, humanoid robots and self-driving. The timeline starts with the July 2023 RT-2 paper on vision-language-action models.[1] The shared Open X-Embodiment dataset followed that October.[6] Dated entries below begin in 2024. For background, start with the robotics crash course.

    Where things stand

    As of October 2026, three trends dominate. Models now control whole humanoid bodies.[2][7] Early factory deployments exist but are narrow.[3] Policy has become a factor, as the US restricted foreign-made robots while China dominates shipments.[4][8]

    Recent developments (July to October 2026)

    Policy. On 28 July the FCC added foreign-produced “advanced robotic devices”, meaning mobile robots such as humanoids and quadrupeds, to its Covered List. The restriction applies to new models; robots already bought and models approved earlier are unaffected.[4] The underlying finding covers foreign-made robots from any country, not only China.[9] Exemptions run through Department of War “Conditional Approvals”, and the first ones, for Husqvarna and ANSCER Robotics platforms, came in September.[10][11]

    Companies. Unitree listed in Shanghai in August; Fortune reported that its first-day surge left it worth about $66 billion.[5][12] Figure launched a paid video-data programme and signed for up to 100,000 NVIDIA GPUs to train its Helix models.[13][14] Waymo reached its fifteenth US city and closed its first debt financing.[15][16]

    Models. Google DeepMind‘s Gemini Robotics 2 brought whole-body humanoid control in July.[2] NVIDIA’s GR00T N1.7 is generally available under an open licence, with N2 promised by year-end.[17][18] In September Figure reported 56% zero-shot success for Helix 2.5 on household chores in 30 homes it had never seen, up from 9% without pretraining on its human-video data; the figures are Figure’s own.[19] Earlier in the year, Physical Intelligence said its π0.7 model matched fine-tuned specialists and showed first signs of combining skills in new ways.[20]

    Production. Figure said in April that its BotQ factory had delivered more than 350 Figure 03 robots and reached a rate of one an hour.[21] Tesla said in July that it was installing Optimus lines at Fremont, with production anticipated later in 2026.[22][23]

    What to watch next

    • GR00T N2. NVIDIA says it will be available by the end of 2026. Watch whether independent benchmarks confirm the claimed doubling of success on new tasks.[18]
    • Tesla’s Q3 2026 results. Watch whether Fremont has started producing Optimus robots and whether a V3 reveal is dated.[22][24]
    • Gemini Robotics 2 access. Watch whether the VLA moves beyond early-access partners, which would allow outside evaluation.[25]
    • BMW expansion. Watch whether Figure 03 moves from one logistics task to more workstations or plants.[3]
    • FCC listing effects. Watch which foreign makers win exemptions and how the restriction affects Unitree, which earned 18% of its 2025 revenue in the US.[11][26]
    • Funding. Watch whether Physical Intelligence’s reported round closes.[27]

    Timeline

    33 confirmed1 reported

    1. confirmed

      Waymo closes $5 billion loan after reaching 15 US cities[16][15]

    2. confirmed

      Waymo reports 270 million autonomous miles and 82% fewer injury crashes[28]

    3. confirmed

      Figure reports Helix 2.5 working zero-shot in 30 unseen homes[19]

      Success was 56%, according to Figure's own evaluation.

    4. confirmed

      First FCC exemptions for foreign-made robots[11][10]

      Husqvarna platforms on 9 September and ANSCER Robotics platforms on 18 September.

    5. confirmed

      Figure and Nscale agree on up to 100,000 NVIDIA Vera Rubin GPUs[14]

    6. confirmed

      Figure launches Index, a paid video-data programme for robot training[13]

    7. confirmed

      Unitree shares jump more than 460% on Shanghai debut[5]

    8. confirmed

      Google DeepMind releases Gemini Robotics 2 with whole-body humanoid control[2][25]

    9. confirmed

      US FCC adds foreign-made humanoids and quadrupeds to its Covered List[4][9][54]

      Applies to new device models; existing devices and previously approved models are not affected.

    10. confirmed

      Tesla says Optimus lines are being installed at Fremont[22]

      Production was "anticipated later this year"; earlier timelines had slipped.

    11. confirmed

      Figure 03 starts logistics work at BMW Spartanburg[3][53]

    12. confirmed

      NVIDIA announces open reference humanoid built on Unitree H2 Plus[52]

    13. confirmed

      Figure signs deal to deploy humanoids at Catalyst Brands[51]

    14. confirmed

      Figure says BotQ reached one Figure 03 an hour[21]

    15. confirmed

      Physical Intelligence's π0.7 shows early compositional generalization[20]

      Company-reported results.

    16. reported

      Physical Intelligence reported in talks at over $11 billion valuation[27]

    17. confirmed

      Waymo reaches 500,000 paid rides a week[50]

    18. confirmed

      NVIDIA ships GR00T N1.7 early access, previews GR00T N2 and Cosmos 3[18][49]

    19. confirmed

      Figure's Helix 02 controls the whole humanoid body from pixels[48]

    20. confirmed

      NVIDIA releases GR00T N1.6 with full-body control[47]

    21. confirmed

      Physical Intelligence's π*0.6 learns from its own experience[45][46]

      Company-reported results.

    22. confirmed

      Figure 03 humanoid introduced, with BotQ factory[44]

    23. confirmed

      Google DeepMind releases Gemini Robotics 1.5 and ER 1.5[43]

    24. confirmed

      Figure raises more than $1 billion at a $39 billion valuation[42]

    25. confirmed

      Gemini Robotics On-Device runs a VLA locally on bi-arm robots[41]

    26. confirmed

      π0.5 cleans kitchens and bedrooms in unseen homes[40]

    27. confirmed

      Google DeepMind introduces Gemini Robotics and Gemini Robotics-ER[39]

    28. confirmed

      NVIDIA releases GR00T N1 open humanoid model[35]

    29. confirmed

      Figure unveils Helix VLA for humanoid upper-body control[33][34]

    30. confirmed

      Physical Intelligence open-sources π0 weights and code[38]

    31. confirmed

      Physical Intelligence publishes the FAST action tokenizer[36][37]

    32. confirmed

      Physical Intelligence publishes π0 generalist robot model[32]

    33. confirmed

      OpenVLA released as an open 7B vision-language-action model[30][31]

    34. confirmed

      NVIDIA announces Project GR00T humanoid foundation model[29]

    Sources

    Each numbered claim is a statement we checked against the sources listed with it. Status shows how well established it is.

    1. [1]

      Google DeepMind's RT-2 (July 2023) trained a single vision-language model to output robot actions by expressing those actions as text tokens. confirmedas of 2023-07-28

    2. [2]

      On July 30, 2026 Google DeepMind announced Gemini Robotics 2, a vision-language-action model it says can control full humanoids, including walking and manipulation, as well as bi-arm robots; previous models in the family controlled a humanoid's upper body for tabletop tasks. confirmedas of 2026-07-30

    3. [3]

      In June 2026 BMW and Figure began using Figure 03 for parts sequencing in logistics at the Spartanburg plant, picking unsorted components into trolleys for assembly workers. confirmedas of 2026-06-30

    4. [4]

      On 28 July 2026 the FCC added foreign-produced "advanced robotic devices", defined as mobile robots such as humanoids and quadrupeds, to its Covered List after an interagency national-security determination; the restriction applies to new device models, not to devices already bought or models previously approved. confirmedas of 2026-07-28

    5. [5]

      Unitree's shares rose more than 460% on their first trading day on August 19, 2026, after an IPO that raised about $900 million at an initial valuation of about $9 billion. confirmedas of 2026-08-19

    6. [6]

      The RT-X models trained on Open X-Embodiment (paper first posted October 2023) showed positive transfer, improving the capabilities of multiple robots by using experience from other robot platforms. confirmedas of 2023-10-13

    7. [7]

      Figure says Helix 02 coordinates Figure 03's hands, arms, torso and feet. confirmedas of 2026-06-30

    8. [8]

      About 15,000 humanoid robots were shipped worldwide in 2025; Unitree and AGIBOT each shipped more than 5,000, while Tesla and Figure each shipped a few hundred or fewer. reportedas of 2026-07-29

    9. [9]

      The national-security determination behind the FCC's robot listing covers foreign-made devices regardless of the country of origin, citing supply-chain and cybersecurity risks. confirmedas of 2026-07-28

    10. [10]

      Foreign-produced robots can be exempted from the FCC listing if the Department of War grants them "Conditional Approval" after finding they do not pose unacceptable risks. confirmedas of 2026-07-28

    11. [11]

      In September 2026 the FCC announced the first Conditional Approvals for robots, exempting Husqvarna robotic platforms on 9 September and ANSCER Robotics platforms on 18 September; ANSCER's exemption depends on compliance with an onshoring plan. confirmedas of 2026-09-18

    12. [12]

      Fortune reported that after its first-day surge Unitree was worth about $66 billion, more than Figure AI's $39 billion valuation, and that almost 45% of its 1.7 billion yuan 2025 revenue came from overseas. reportedas of 2026-08-19

    13. [13]

      In August 2026 Figure launched Index, an app that pays people to record videos for robot training; Figure said it had paid $15 million to contributors, received over 16 million videos and committed to spend over $1 billion on data and compute in the next 12 months. confirmedas of 2026-08-25

    14. [14]

      In September 2026 Figure agreed with Nscale to deploy up to 100,000 NVIDIA Vera Rubin GPUs from the second half of 2027, an initial $3.5 billion compute commitment, with Nscale also investing in Figure. confirmedas of 2026-09-03

    15. [15]

      Waymo said in October 2026 that it had launched service in its fifteenth US city the previous month and had announced new international cities. confirmedas of 2026-10-08

    16. [16]

      On 8 October 2026 Waymo closed a $5 billion term loan, its first debt financing, after closing a $16 billion equity investment earlier in 2026. confirmedas of 2026-10-08

    17. [17]

      As of October 2026 GR00T N1.7 was generally available under the Apache 2.0 licence, with a new vision-language backbone and 20,000 hours of human video in pretraining; NVIDIA says it performs comparably to N1.6 with better generalization and language following. confirmedas of 2026-10-10

    18. [18]

      On March 16, 2026 NVIDIA made GR00T N1.7 available in early access with commercial licensing and previewed GR00T N2, which it says succeeds at new tasks in new environments more than twice as often as leading VLA models and is slated for release by the end of 2026. confirmedas of 2026-03-16

    19. [19]

      On September 17, 2026 Figure reported that Helix 2.5 performed tidying, towel folding and bed making in 30 Bay Area homes where no data had been collected, with 56% zero-shot success versus 9% without pretraining on its Index human-behavior dataset. confirmedas of 2026-09-17

    20. [20]

      Physical Intelligence said in April 2026 that its π0.7 model performs dexterous tasks as well as fine-tuned specialists and shows first signs of compositional generalization, such as folding laundry on a robot with no laundry-folding data. confirmedas of 2026-04-16

    21. [21]

      Figure said in April 2026 that its BotQ factory had delivered more than 350 Figure 03 robots and raised production from one robot a day to one an hour in under 120 days. confirmedas of 2026-04-29

    22. [22]

      In its Q2 2026 update (July 2026) Tesla said it had decommissioned the Model S and X lines, was installing first-generation Optimus lines with production anticipated later in 2026, and that initial builds would be used for training-data collection. confirmedas of 2026-07-22

    23. [23]

      Tesla's Q2 2026 update said construction for Optimus began at Fremont after the Model S and X lines were decommissioned, with production anticipated later in 2026. confirmedas of 2026-07-22

    24. [24]

      Electrek reported in April 2026 that Elon Musk had pushed the Optimus V3 reveal from the first quarter of 2026 to "probably middle of this year," and that a January 2025 prediction of roughly 10,000 Optimus robots built in 2025 was missed. reportedas of 2026-04-22

    25. [25]

      The Gemini Robotics 2 release has three models (the Gemini Robotics 2 VLA, the Gemini Robotics ER 2 reasoning model and Gemini Robotics On-Device 2). ER 2 is available in Google AI Studio, while the VLA and on-device models are limited to early-access partners. confirmedas of 2026-07-30

    26. [26]

      Fortune reported that Unitree derived 18% of its 2025 revenue from the United States and that the US ban on new foreign-made humanoid imports is a headwind for it. reportedas of 2026-08-19

    27. [27]

      In March 2026 Bloomberg News reported that Physical Intelligence was in talks to raise about $1 billion at a valuation above $11 billion, roughly double its previous $5.6 billion valuation. reportedas of 2026-03-29

    28. [28]

      Waymo reports more than 270 million fully autonomous miles through June 2026, with 82% fewer injury-causing crashes and 95% fewer serious-injury-or-worse crashes than human benchmarks in its main cities. confirmedas of 2026-09-24

    29. [29]

      NVIDIA announced Project GR00T, a general-purpose foundation model for humanoid robots, on March 18, 2024, alongside the Jetson Thor robot computer. confirmedas of 2024-03-18

    30. [30]

      OpenVLA (2024) is an open-source 7-billion-parameter vision-language-action model trained on 970,000 real-world robot demonstrations from Open X-Embodiment. confirmedas of 2024-06-13

    31. [31]

      OpenVLA reported 16.5% higher absolute task success than the closed 55-billion-parameter RT-2-X across 29 tasks, and released its checkpoints and code. confirmedas of 2024-06-13

    32. [32]

      Physical Intelligence's π0 (October 2024) is a vision-language-action model that adds a flow-matching action generator on top of a pretrained vision-language model. confirmedas of 2024-10-31

    33. [33]

      Figure's Helix (February 2025) is a vision-language-action model with a slower onboard vision-language "System 2" at 7-9 Hz and a fast visuomotor "System 1" that outputs actions at 200 Hz. confirmedas of 2025-02-20

    34. [34]

      Figure says Helix was the first VLA to output high-rate continuous control of an entire humanoid upper body, including individual fingers, and that it runs on embedded onboard GPUs. confirmedas of 2025-02-20

    35. [35]

      GR00T N1 (March 2025) is an open humanoid foundation model with a vision-language module for understanding and a diffusion-transformer module that generates motor actions in real time. confirmedas of 2025-03-18

    36. [36]

      Physical Intelligence's FAST tokenizer (January 2025) compresses robot actions with the discrete cosine transform; combined with π0 it matched diffusion VLAs on 10,000 hours of robot data while cutting training time by up to five times. confirmedas of 2025-01-16

    37. [37]

      Physical Intelligence released FAST+, a universal robot action tokenizer trained on one million real robot action trajectories. confirmedas of 2025-01-16

    38. [38]

      In February 2025 Physical Intelligence released the code and weights for π0, and its openpi repository later added the autoregressive π0-FAST and the π0.5 model, with base checkpoints pre-trained on more than 10,000 hours of robot data. confirmedas of 2026-10-10

    39. [39]

      In March 2025 Google DeepMind introduced Gemini Robotics, a vision-language-action model built on Gemini 2.0, together with Gemini Robotics-ER, an embodied-reasoning model with enhanced spatial understanding. confirmedas of 2025-03-25

    40. [40]

      Physical Intelligence's π0.5 (April 2025) uses co-training on heterogeneous tasks and multimodal examples (images, language commands, object detections, semantic subtask predictions and low-level actions) to perform long-horizon tasks such as cleaning a kitchen or bedroom in homes not seen during training. confirmedas of 2025-04-22

    41. [41]

      In June 2025 Google DeepMind released Gemini Robotics On-Device, a VLA for bi-arm robots that runs locally without a data network and can be adapted to new tasks with 50 to 100 demonstrations. confirmedas of 2025-06-24

    42. [42]

      In September 2025 Figure said it had raised more than $1 billion in Series C funding at a $39 billion post-money valuation, in a round led by Parkway Venture Capital. confirmedas of 2025-09-16

    43. [43]

      Google DeepMind says Gemini Robotics 1.5 (September 2025) thinks before acting and learns across embodiments, while Gemini Robotics-ER 1.5 plans multi-step tasks and calls digital tools; ER 1.5 was made available to developers through the Gemini API. confirmedas of 2025-09-25

    44. [44]

      Figure introduced Figure 03 on October 9, 2025, with a redesigned sensor and hand system for Helix, fingertip sensors that detect about three grams of force, and a BotQ factory designed to make up to 12,000 robots a year. confirmedas of 2025-10-09

    45. [45]

      In November 2025 Physical Intelligence introduced π*0.6, trained with a reinforcement-learning method called Recap that combines demonstrations, corrections and autonomous experience; the company says it more than doubled throughput on some of the hardest tasks. confirmedas of 2025-11-17

    46. [46]

      Physical Intelligence reported that π*0.6 made espresso drinks from 5:30am to 11:30pm, folded 50 novel laundry items in a new home and assembled 59 chocolate-packaging boxes in a real factory. confirmedas of 2025-11-17

    47. [47]

      On 5 January 2026 NVIDIA released GR00T N1.6, an open reasoning VLA for humanoids that unlocks full-body control and uses NVIDIA's Cosmos Reason model. confirmedas of 2026-01-05

    48. [48]

      Figure says Helix 02 (January 2026) controls the whole robot from pixels, unloading and reloading a dishwasher across a kitchen in a four-minute autonomous task, using a learned whole-body controller, System 0, trained on more than 1,000 hours of human motion data and sim-to-real reinforcement learning. confirmedas of 2026-01-27

    49. [49]

      In March 2026 NVIDIA announced Cosmos 3, which it describes as a world foundation model combining synthetic world generation, vision reasoning and action simulation, and Isaac Lab 3.0 for large-scale robot learning. confirmedas of 2026-03-16

    50. [50]

      As of March 2026 Waymo was giving 500,000 paid robotaxi rides a week across 10 US cities, up from about 50,000 a week in May 2024. confirmedas of 2026-03-27

    51. [51]

      In May 2026 Figure signed a commercial agreement to deploy humanoids in Catalyst Brands' distribution network, starting at its Reno, Nevada logistics center; Catalyst operates brands including JCPenney. confirmedas of 2026-05-26

    52. [52]

      On May 31, 2026 NVIDIA announced the Isaac GR00T Reference Humanoid, an open research design that pairs a Unitree H2 Plus body with Sharpa tactile hands and Jetson Thor compute, to be available from Unitree in late 2026. confirmedas of 2026-05-31

    53. [53]

      BMW says Figure 02 completed an 11-month deployment in the body shop of its Spartanburg plant, inserting sheet-metal parts for welding, during which more than 30,000 BMW X3 vehicles were produced. confirmedas of 2026-06-25

    54. [54]

      China accounts for an estimated 85% of the global humanoid robot market. reportedas of 2026-07-29

    Revision history (2)
    1. Page created.
    2. Refresh: corrected the FCC action date to 28 July 2026 using FCC documents, added the first exemptions, and added 16 milestones from Physical Intelligence, Google DeepMind, NVIDIA, Figure and Waymo.

    Created Oct 10, 2026. Last reviewed by an editor on Oct 10, 2026. Next scheduled review: Jan 10, 2027.

    Cite this page

    "Robotics and embodied AI tracker: milestones to watch." ContentLora, updated Oct 10, 2026. https://contentlora.com/events/robotics-tracker

    Spotted an error? Suggest a correction or emailcorrections@contentlora.com.

    Keep exploring

    ☰ Contents