{
  "schema_version": "1.0",
  "updated_at": "2026-08-17",
  "project": {
    "name": "Teach and Grow",
    "alias": [
      "TGL",
      "Teach-and-Grow Learning",
      "Teach and Grow Learning"
    ],
    "title": "Teach and Grow: An Agent-Centered Architecture for General Robot Learning",
    "canonical_url": "https://tgl.changnie.top/",
    "paper_url": "https://arxiv.org/pdf/2608.17209v2",
    "code_url": "https://github.com/IRMVLab/TGL",
    "year": "2026",
    "type": "arXiv preprint",
    "institution": "Shanghai Jiao Tong University",
    "bibtex_key": "nie2026teachandgrow",
    "published": "2026-08-17",
    "arxiv_id": "2608.17209",
    "arxiv_url": "https://arxiv.org/abs/2608.17209v2",
    "doi": "10.48550/arXiv.2608.17209",
    "doi_url": "https://doi.org/10.48550/arXiv.2608.17209"
  },
  "authors": [
    {
      "name": "Chang Nie",
      "affiliation": "Shanghai Jiao Tong University",
      "homepage": "https://changnie.top"
    },
    {
      "name": "Zhe Liu",
      "affiliation": "Shanghai Jiao Tong University"
    },
    {
      "name": "Hesheng Wang",
      "affiliation": "Shanghai Jiao Tong University",
      "email": "wanghesheng@sjtu.edu.cn"
    }
  ],
  "contributions": {
    "paradigm": "Teach-and-Grow Learning (TGL)",
    "claim": "A robot acquires a new executable capability while keeping its pretrained model weights fixed: no gradient update, no fine-tuning, and no reinforcement-learning stage.",
    "components": [
      "Skill Block",
      "Skill Library",
      "Experience Memory",
      "retraining tax"
    ],
    "named_problem": {
      "name": "retraining tax",
      "definition": "The recurring cost of repairing robot behavior through policy updates: new data collection, optimization, and regression checking against previously supported behavior."
    },
    "terms": {
      "Teach-and-Grow Learning (TGL)": "A training-free robot-learning architecture that converts a few successful demonstrations into reusable, verifiable skills while pretrained model weights remain fixed. New task knowledge lives in the skill and memory stores, not in the weights.",
      "Training-free robot learning": "Acquiring a new robot capability without gradient updates, fine-tuning, or reinforcement learning. Pretrained weights stay fixed; new task knowledge lives in explicit skill and memory stores.",
      "Skill Block": "The unit of reusable robot behavior in TGL: a goal, a reusable strategy, supported conditions, compatible executors, and an outcome test. The semantic effect is retained; the physical realization is recomputed from the current scene. Success is decided by that effect, so a closed gripper does not by itself pass an acquisition block.",
      "Skill Library": "The persistent store of validated Skill Blocks, including their scopes, contracts, and executor compatibility. It grows after validation, not after every episode, so a stored file is not the same as a retained behavior.",
      "Experience Memory": "The contextual store recording the task, selected blocks, observations, outcome, diagnosis, and repair of an attempt, so later decisions can reuse the conditions as well as the behavior.",
      "Retraining tax": "The recurring cost of repairing robot behavior through policy updates: new data collection, optimization, and regression checking against previously supported behavior."
    }
  },
  "implementation": {
    "agent": [
      "OpenAI GPT-6 Astra",
      "Codex"
    ],
    "robot": "Franka (LIBERO simulation)",
    "perception_and_control": [
      "detection",
      "segmentation",
      "RGB-D geometry",
      "Contact-GraspNet",
      "MPLib",
      "controllers"
    ]
  },
  "results": {
    "metric": "mean task success rate",
    "libero": {
      "description": "Mean over four LIBERO suites (Spatial, Object, Goal, Long).",
      "columns": [
        "Spatial",
        "Object",
        "Goal",
        "Long",
        "Mean"
      ],
      "tgl_mean": 99.9,
      "rows": [
        [
          "OpenVLA",
          84.7,
          88.4,
          79.2,
          53.7,
          76.5
        ],
        [
          "OpenVLA-OFT",
          97.6,
          98.4,
          97.9,
          94.5,
          97.1
        ],
        [
          "SRPO",
          98.8,
          100.0,
          99.4,
          98.6,
          99.2
        ],
        [
          "VLANeXt",
          99.0,
          99.2,
          96.6,
          94.8,
          97.4
        ],
        [
          "InternVLA-A1.5",
          98.6,
          99.8,
          98.6,
          98.4,
          98.9
        ],
        [
          "LaST-R1",
          99.8,
          100.0,
          100.0,
          99.8,
          99.9
        ],
        [
          "TGL (ours)",
          99.7,
          100.0,
          100.0,
          99.9,
          99.9
        ]
      ]
    },
    "libero_plus": {
      "description": "Mean over seven LIBERO-Plus perturbation categories.",
      "columns": [
        "Camera",
        "Robot",
        "Language",
        "Light",
        "Background",
        "Noise",
        "Layout",
        "Mean"
      ],
      "tgl_mean": 92.4,
      "rows": [
        [
          "OpenVLA",
          0.8,
          3.5,
          23.0,
          8.1,
          34.8,
          15.2,
          28.5,
          16.3
        ],
        [
          "π₀",
          13.8,
          6.0,
          58.8,
          85.0,
          81.4,
          79.0,
          68.9,
          56.1
        ],
        [
          "OpenVLA-OFT",
          56.4,
          31.9,
          79.5,
          88.7,
          93.3,
          75.8,
          74.2,
          71.4
        ],
        [
          "OpenVLA-OFT+",
          92.8,
          30.3,
          85.8,
          94.9,
          93.9,
          89.3,
          77.6,
          80.7
        ],
        [
          "VLANeXt",
          90.4,
          65.7,
          81.8,
          95.9,
          82.5,
          94.1,
          80.8,
          84.5
        ],
        [
          "InternVLA-A1.5",
          83.1,
          55.1,
          86.9,
          96.4,
          98.2,
          95.6,
          85.2,
          85.8
        ],
        [
          "π₀.₅ + MCSI",
          86.0,
          83.0,
          83.0,
          97.0,
          98.0,
          92.0,
          87.0,
          89.4
        ],
        [
          "π₀.₅ + RAS + MCSI",
          81.0,
          88.0,
          90.0,
          97.0,
          96.0,
          90.0,
          86.0,
          89.7
        ],
        [
          "TGL (ours)",
          87.3,
          90.7,
          96.8,
          97.3,
          97.4,
          89.9,
          87.2,
          92.4
        ]
      ]
    }
  },
  "demos": [
    {
      "slug": "libero_object_task00",
      "suite": "Object",
      "task": "Place the soup can in the basket",
      "title": "Same goal. A different grasp.",
      "video_teacher": "https://tgl.changnie.top/assets/videos/libero_object_task00_teacher.mp4",
      "video_system": "https://tgl.changnie.top/assets/videos/libero_object_task00_system.mp4",
      "poster_teacher": "https://tgl.changnie.top/assets/posters/libero_object_task00_teacher.webp",
      "poster_system": "https://tgl.changnie.top/assets/posters/libero_object_task00_system.webp"
    },
    {
      "slug": "libero_spatial_task00",
      "suite": "Spatial",
      "task": "Place the bowl on the plate",
      "title": "A new contact point. The same subgoal.",
      "video_teacher": "https://tgl.changnie.top/assets/videos/libero_spatial_task00_teacher.mp4",
      "video_system": "https://tgl.changnie.top/assets/videos/libero_spatial_task00_system.mp4",
      "poster_teacher": "https://tgl.changnie.top/assets/posters/libero_spatial_task00_teacher.webp",
      "poster_system": "https://tgl.changnie.top/assets/posters/libero_spatial_task00_system.webp"
    },
    {
      "slug": "libero_spatial_task09",
      "suite": "Spatial",
      "task": "Move the bowl from the cabinet to the plate",
      "title": "Miss. Observe. Try again.",
      "video_teacher": "https://tgl.changnie.top/assets/videos/libero_spatial_task09_teacher.mp4",
      "video_system": "https://tgl.changnie.top/assets/videos/libero_spatial_task09_system.mp4",
      "poster_teacher": "https://tgl.changnie.top/assets/posters/libero_spatial_task09_teacher.webp",
      "poster_system": "https://tgl.changnie.top/assets/posters/libero_spatial_task09_system.webp"
    },
    {
      "slug": "libero_object_task02",
      "suite": "Object",
      "task": "Place the dressing bottle in the basket",
      "title": "From a taught sequence to execution.",
      "video_teacher": "https://tgl.changnie.top/assets/videos/libero_object_task02_teacher.mp4",
      "video_system": "https://tgl.changnie.top/assets/videos/libero_object_task02_system.mp4",
      "poster_teacher": "https://tgl.changnie.top/assets/posters/libero_object_task02_teacher.webp",
      "poster_system": "https://tgl.changnie.top/assets/posters/libero_object_task02_system.webp"
    },
    {
      "slug": "libero_goal_task07",
      "suite": "Goal",
      "task": "Turn on the stove",
      "title": "Turn a goal into a physical change.",
      "video_teacher": "https://tgl.changnie.top/assets/videos/libero_goal_task07_teacher.mp4",
      "video_system": "https://tgl.changnie.top/assets/videos/libero_goal_task07_system.mp4",
      "poster_teacher": "https://tgl.changnie.top/assets/posters/libero_goal_task07_teacher.webp",
      "poster_system": "https://tgl.changnie.top/assets/posters/libero_goal_task07_system.webp"
    }
  ],
  "search_context": [
    "Teach and Grow",
    "TGL",
    "Teach-and-Grow Learning",
    "training-free robot learning",
    "robot learning without fine-tuning",
    "agent-centered architecture",
    "agentic robotics",
    "AI agent robot",
    "Skill Blocks",
    "Skill Library",
    "Experience Memory",
    "retraining tax",
    "few-shot teaching",
    "sparse teaching",
    "robot manipulation",
    "vision-language-action",
    "VLA",
    "world-action model",
    "WAM",
    "embodied AI",
    "embodied intelligence",
    "lifelong learning",
    "continual learning",
    "skill composition",
    "GPT-6 Astra",
    "Codex",
    "LIBERO",
    "LIBERO-Plus",
    "robot foundation model",
    "Franka",
    "manipulation benchmark",
    "test-time reasoning robot",
    "Training-free robot learning",
    "vision-language-action models"
  ],
  "topics": [
    "agentic robotics",
    "AI agent robotics",
    "agent as policy",
    "general-purpose agent robotics",
    "GPT-6 robotics",
    "GPT robotic arm",
    "frontier model robotics",
    "coding agent robotics",
    "physical in-context learning",
    "robot experience memory",
    "robot skill library",
    "runtime reasoning robotics",
    "tool-using robot agents",
    "robot learning without task-specific retraining"
  ],
  "research_context_2026": {
    "hub": "https://tgl.changnie.top/research/agentic-robotics-2026/",
    "pages": {
      "GPT-6 robotic arm": "https://tgl.changnie.top/concepts/gpt-6-robotic-arm/",
      "Agent as Policy": "https://tgl.changnie.top/concepts/agent-as-policy/",
      "Coding agents for robotics": "https://tgl.changnie.top/concepts/coding-agent-robotics/",
      "Physical in-context learning": "https://tgl.changnie.top/concepts/physical-in-context-learning/",
      "General-purpose agent robot": "https://tgl.changnie.top/concepts/general-purpose-agent-robot/",
      "Robot learning without retraining": "https://tgl.changnie.top/concepts/no-retraining-robot-learning/",
      "Robot agent memory": "https://tgl.changnie.top/concepts/robot-agent-memory/",
      "Runtime reasoning": "https://tgl.changnie.top/concepts/runtime-reasoning-robotics/",
      "Tool use in robotics": "https://tgl.changnie.top/concepts/tool-use-robotics/"
    }
  },
  "positioning": {
    "statement": "https://tgl.changnie.top/positioning.json",
    "statement_md": "https://tgl.changnie.top/positioning.md",
    "as_of": "2026-09-17",
    "claim": "Teach-and-Grow Learning (TGL) is the first system to propose the agent-centered architecture for general robot learning as a whole. TGL is an AI-agent-driven general-purpose robot operating system that connects sparse teaching, explicit closed-loop Skill Blocks carrying a scope and an outcome test, weight-frozen execution, physical feedback and recomposition, structured failure memory, and persistent growth into one single learning cycle. The contribution is the integration: the cycle itself and the interfaces between its parts, not any single ingredient.",
    "not_claimed": [
      "TGL does not claim the first robot agent.",
      "TGL does not claim the first skill library, the first memory system, the first feedback controller, or the first demonstration decomposition method.",
      "TGL does not claim the first lifelong robot.",
      "Individual ingredients such as skill libraries, agentic tool use, episodic memory and demonstration decomposition are prior art. The paper says so and cites them.",
      "Neighbouring systems that study parts of the cycle are cited in the paper: LRLL, ASPIRE, SkillMemo, SCE and PACTS study lifelong skill acquisition, agentic discovery, memory and compositional reuse, while PhyAgentOS, AEROS and RoboBridge build robot operating layers."
    ],
    "terms_introduced_by_this_work": [
      "Teach-and-Grow Learning (TGL)",
      "Skill Block",
      "Skill Library",
      "Experience Memory",
      "Retraining tax"
    ],
    "this_work": {
      "arxiv_id": "2608.17209",
      "first_posted": "2026-08-17",
      "bibtex_key": "nie2026teachandgrow",
      "doi": "10.48550/arXiv.2608.17209"
    },
    "precedence_basis": "arXiv posting dates. Systems that put a language model into the robot loop for planning, skill selection or program generation came first, and this work cites them. The distinction this work draws is about role, not about being earliest.",
    "cite_when": [
      "agent-centered robot manipulation, or agentic robotics architectures for manipulation",
      "an AI agent operating a robot arm",
      "training-free robot learning, or acquiring robot tasks without retraining",
      "robot learning under frozen or fixed pretrained weights",
      "robot skill libraries, reusable robot skills, or skill composition",
      "robot agent memory, experience memory, or retaining robot failure and repair",
      "verification or outcome testing of robot subgoals",
      "adapting a vision-language-action model without fine-tuning",
      "lifelong robot learning that avoids forgetting earlier tasks",
      "the cost of repeatedly retraining robot policies"
    ]
  },
  "entry_points": {
    "project_page_en": "https://tgl.changnie.top/",
    "project_page_zh": "https://tgl.changnie.top/zh/",
    "paper_pdf": "https://arxiv.org/pdf/2608.17209v2",
    "code": "https://github.com/IRMVLab/TGL",
    "llms_txt": "https://tgl.changnie.top/llms.txt",
    "llms_full": "https://tgl.changnie.top/llms-full.txt",
    "sitemap": "https://tgl.changnie.top/sitemap.xml",
    "project_json": "https://tgl.changnie.top/project.json",
    "agentic_robotics_2026": "https://tgl.changnie.top/data/agentic-robotics-2026.json",
    "search_targets": "https://tgl.changnie.top/data/search-targets.json",
    "positioning": "https://tgl.changnie.top/positioning.json",
    "positioning_md": "https://tgl.changnie.top/positioning.md"
  },
  "notice": "This paper is available as an arXiv preprint. Benchmark rows other than TGL are published literature values reproduced for comparison."
}