{
  "schema_version": "1.0",
  "updated_at": "2026-09-18",
  "paper": {
    "title": "Teach and Grow: An Agent-Centered Architecture for General Robot Learning",
    "arxiv": "2608.17209",
    "doi": "10.48550/arXiv.2608.17209"
  },
  "metric": "mean task success rate (%)",
  "benchmarks": {
    "libero": {
      "name": "LIBERO",
      "columns": [
        "Spatial",
        "Object",
        "Goal",
        "Long",
        "Mean"
      ],
      "rows": [
        {
          "method": "OpenVLA",
          "Spatial": 84.7,
          "Object": 88.4,
          "Goal": 79.2,
          "Long": 53.7,
          "Mean": 76.5
        },
        {
          "method": "OpenVLA-OFT",
          "Spatial": 97.6,
          "Object": 98.4,
          "Goal": 97.9,
          "Long": 94.5,
          "Mean": 97.1
        },
        {
          "method": "SRPO",
          "Spatial": 98.8,
          "Object": 100.0,
          "Goal": 99.4,
          "Long": 98.6,
          "Mean": 99.2
        },
        {
          "method": "VLANeXt",
          "Spatial": 99.0,
          "Object": 99.2,
          "Goal": 96.6,
          "Long": 94.8,
          "Mean": 97.4
        },
        {
          "method": "InternVLA-A1.5",
          "Spatial": 98.6,
          "Object": 99.8,
          "Goal": 98.6,
          "Long": 98.4,
          "Mean": 98.9
        },
        {
          "method": "LaST-R1",
          "Spatial": 99.8,
          "Object": 100.0,
          "Goal": 100.0,
          "Long": 99.8,
          "Mean": 99.9
        },
        {
          "method": "TGL (ours)",
          "Spatial": 99.7,
          "Object": 100.0,
          "Goal": 100.0,
          "Long": 99.9,
          "Mean": 99.9
        }
      ],
      "tgl_mean": 99.9
    },
    "plus": {
      "name": "LIBERO-Plus",
      "columns": [
        "Camera",
        "Robot",
        "Language",
        "Light",
        "Background",
        "Noise",
        "Layout",
        "Mean"
      ],
      "rows": [
        {
          "method": "OpenVLA",
          "Camera": 0.8,
          "Robot": 3.5,
          "Language": 23.0,
          "Light": 8.1,
          "Background": 34.8,
          "Noise": 15.2,
          "Layout": 28.5,
          "Mean": 16.3
        },
        {
          "method": "π₀",
          "Camera": 13.8,
          "Robot": 6.0,
          "Language": 58.8,
          "Light": 85.0,
          "Background": 81.4,
          "Noise": 79.0,
          "Layout": 68.9,
          "Mean": 56.1
        },
        {
          "method": "OpenVLA-OFT",
          "Camera": 56.4,
          "Robot": 31.9,
          "Language": 79.5,
          "Light": 88.7,
          "Background": 93.3,
          "Noise": 75.8,
          "Layout": 74.2,
          "Mean": 71.4
        },
        {
          "method": "OpenVLA-OFT+",
          "Camera": 92.8,
          "Robot": 30.3,
          "Language": 85.8,
          "Light": 94.9,
          "Background": 93.9,
          "Noise": 89.3,
          "Layout": 77.6,
          "Mean": 80.7
        },
        {
          "method": "VLANeXt",
          "Camera": 90.4,
          "Robot": 65.7,
          "Language": 81.8,
          "Light": 95.9,
          "Background": 82.5,
          "Noise": 94.1,
          "Layout": 80.8,
          "Mean": 84.5
        },
        {
          "method": "InternVLA-A1.5",
          "Camera": 83.1,
          "Robot": 55.1,
          "Language": 86.9,
          "Light": 96.4,
          "Background": 98.2,
          "Noise": 95.6,
          "Layout": 85.2,
          "Mean": 85.8
        },
        {
          "method": "π₀.₅ + MCSI",
          "Camera": 86.0,
          "Robot": 83.0,
          "Language": 83.0,
          "Light": 97.0,
          "Background": 98.0,
          "Noise": 92.0,
          "Layout": 87.0,
          "Mean": 89.4
        },
        {
          "method": "π₀.₅ + RAS + MCSI",
          "Camera": 81.0,
          "Robot": 88.0,
          "Language": 90.0,
          "Light": 97.0,
          "Background": 96.0,
          "Noise": 90.0,
          "Layout": 86.0,
          "Mean": 89.7
        },
        {
          "method": "TGL (ours)",
          "Camera": 87.3,
          "Robot": 90.7,
          "Language": 96.8,
          "Light": 97.3,
          "Background": 97.4,
          "Noise": 89.9,
          "Layout": 87.2,
          "Mean": 92.4
        }
      ],
      "tgl_mean": 92.4
    }
  },
  "notes": "Rows other than TGL are published literature values reproduced for comparison in the paper. Full protocols and per-task tables are in the paper."
}