captions_abot_style/caption_samples_v2.md

# ABot-Style Caption Samples

## rt_000000 / pick_place / place_a2b_left

**source_instruction:** Place the pink toycar shaped like real car directly to the left of the palm-sized phone.
**roles:** active_arm=unknown | verb=place | A=the pink toycar shaped like real car | target=the palm-sized phone | relation=left of | confidence=0.95
**flags:** template_exact

**caption_short:** A dual-arm Aloha-AgileX robot places the pink toycar shaped like real car to the left of the palm-sized phone.

**caption_worldarena_style:** In a fixed robotic workspace, generate a rigid, physically consistent embodied robotic arm. The arm maintains high stability with no deformation and enters the frame to place the pink toycar shaped like real car to the left of the palm-sized phone.

**Stage 1 JSON:**
```json
{
  "high_level_task": {
    "task_family": "pick_place",
    "robotwin_task_name": "place_a2b_left",
    "intent": "place the pink toycar shaped like real car to the left of the palm-sized phone",
    "source_instruction": "Place the pink toycar shaped like real car directly to the left of the palm-sized phone."
  },
  "role_parse": {
    "active_arm": "unknown",
    "main_verb": "place",
    "manipulated_object": "the pink toycar shaped like real car",
    "target_object": "the palm-sized phone",
    "target_receptacle": "",
    "spatial_relation": "left of",
    "secondary_action": "",
    "primary_action": "place_relative",
    "pre_action": "",
    "compound_instruction": "false",
    "unresolved_pronoun": "false",
    "target_pollution_fixed": "false",
    "parse_confidence": 0.95
  },
  "robot_initial_state": {
    "morphology": "Aloha-AgileX-style dual-arm gripper robot",
    "visible_assumption": "robot arms may enter the fixed camera view from outside the frame",
    "initial_contact": "no contact is assumed before approach unless visible in the video"
  },
  "scene_context": {
    "location_type": "fixed tabletop robotic workspace",
    "camera_view": "static head-camera view, 640x480 video rendered from a fixed workspace camera",
    "objects": [
      "the pink toycar shaped like real car"
    ],
    "static_objects": [
      "the palm-sized phone"
    ]
  },
  "skill_sequence": [
    {
      "skill_index": 1,
      "verb": "approach",
      "active_arm": "unknown",
      "object": "the pink toycar shaped like real car",
      "target": "workspace contact region",
      "description": "The active gripper approaches the pink toycar shaped like real car while staying in the fixed tabletop workspace."
    },
    {
      "skill_index": 2,
      "verb": "contact_or_grasp",
      "active_arm": "unknown",
      "object": "the pink toycar shaped like real car",
      "target": "the pink toycar shaped like real car",
      "description": "The gripper or tool should make visible or physically plausible contact with the pink toycar shaped like real car before the main motion."
    },
    {
      "skill_index": 3,
      "verb": "place_relative",
      "active_arm": "unknown",
      "object": "the pink toycar shaped like real car",
      "target": "the palm-sized phone",
      "description": "The robot performs the action: place the pink toycar shaped like real car to the left of the palm-sized phone. Parsed relation: left of."
    },
    {
      "skill_index": 4,
      "verb": "finish",
      "active_arm": "unknown",
      "object": "the pink toycar shaped like real car",
      "target": "the palm-sized phone",
      "description": "The motion ends with the pink toycar shaped like real car in a stable state relative to the palm-sized phone."
    }
  ],
  "key_state_changes": [
    "the pink toycar shaped like real car changes pose or state according to the task intent",
    "the gripper state may open or close near the contact moment",
    "the final state should preserve the relation between the pink toycar shaped like real car and the palm-sized phone"
  ],
  "physical_relations": {
    "contact": "motion of the pink toycar shaped like real car should be explained by gripper, end-effector, or tool contact",
    "support": "the pink toycar shaped like real car should remain supported by the gripper, tabletop, receptacle, or target structure",
    "containment": "not required unless visible",
    "violations_to_avoid": [
      "object teleportation",
      "unexplained floating motion",
      "severe penetration through rigid objects",
      "non-rigid deformation of the robot or target object"
    ]
  },
  "task_result_note": "Use visible evidence from the clip; if an object or contact is not visible, keep the description generic.",
  "caption_quality_flags": [
    "template_exact"
  ]
}
```

**Stage 2 Dense Caption:**

[Phase 1] Scene Setup
A fixed head-camera view observes an Aloha-AgileX-style dual-arm gripper robot in a tabletop workspace. The parsed manipulated object is the pink toycar shaped like real car. The parsed target or reference is the palm-sized phone, and the intended spatial relation is left of. The task family is pick_place, so the description focuses on visible rigid-body manipulation rather than appearance embellishment.

[Phase 2] Action Detail
The robot uses an active gripper to perform the action: place the pink toycar shaped like real car to the left of the palm-sized phone. The motion should begin with an approach toward the pink toycar shaped like real car, followed by contact, grasping, pressing, or tool interaction as appropriate for the task. During the main action, the pink toycar shaped like real car should move with the gripper or tool rather than independently, and the trajectory should remain continuous in the fixed camera view.

[Phase 3] State Transition
The expected state transition is that the pink toycar shaped like real car moves from its initial tabletop pose to left of the palm-sized phone. The final arrangement should be stable and physically supported by the tabletop, the gripper, the receptacle, or the referenced object. The clip should not show object teleportation, unexplained floating, severe penetration, or deformation of rigid objects.

[Phase 4] Camera Summary
The camera remains static and documents the manipulation in an observation-style view. White tabletop background, mild simulator grain, and partial robot visibility are normal for this domain. The caption avoids claiming contact or object identity beyond what is parsed from the instruction and task metadata.

## rt_000001 / pick_place / place_a2b_left

**source_instruction:** With the left arm, place the blue and white playingcards packaging left of the soap bar with curved edges
**roles:** active_arm=left | verb=place | A=the blue and white playingcards packaging | target=the soap bar with curved edges | relation=left of | confidence=0.95
**flags:** template_exact

**caption_short:** A dual-arm Aloha-AgileX robot uses its left arm to place the blue and white playingcards packaging to the left of the soap bar with curved edges.

**caption_worldarena_style:** In a fixed robotic workspace, generate a rigid, physically consistent embodied robotic arm. The arm maintains high stability with no deformation and enters the frame to place the blue and white playingcards packaging to the left of the soap bar with curved edges.

**Stage 1 JSON:**
```json
{
  "high_level_task": {
    "task_family": "pick_place",
    "robotwin_task_name": "place_a2b_left",
    "intent": "place the blue and white playingcards packaging to the left of the soap bar with curved edges",
    "source_instruction": "With the left arm, place the blue and white playingcards packaging left of the soap bar with curved edges"
  },
  "role_parse": {
    "active_arm": "left",
    "main_verb": "place",
    "manipulated_object": "the blue and white playingcards packaging",
    "target_object": "the soap bar with curved edges",
    "target_receptacle": "",
    "spatial_relation": "left of",
    "secondary_action": "",
    "primary_action": "place_relative",
    "pre_action": "",
    "compound_instruction": "false",
    "unresolved_pronoun": "false",
    "target_pollution_fixed": "false",
    "parse_confidence": 0.95
  },
  "robot_initial_state": {
    "morphology": "Aloha-AgileX-style dual-arm gripper robot",
    "visible_assumption": "robot arms may enter the fixed camera view from outside the frame",
    "initial_contact": "no contact is assumed before approach unless visible in the video"
  },
  "scene_context": {
    "location_type": "fixed tabletop robotic workspace",
    "camera_view": "static head-camera view, 640x480 video rendered from a fixed workspace camera",
    "objects": [
      "the blue and white playingcards packaging"
    ],
    "static_objects": [
      "the soap bar with curved edges"
    ]
  },
  "skill_sequence": [
    {
      "skill_index": 1,
      "verb": "approach",
      "active_arm": "left",
      "object": "the blue and white playingcards packaging",
      "target": "workspace contact region",
      "description": "The active gripper approaches the blue and white playingcards packaging while staying in the fixed tabletop workspace."
    },
    {
      "skill_index": 2,
      "verb": "contact_or_grasp",
      "active_arm": "left",
      "object": "the blue and white playingcards packaging",
      "target": "the blue and white playingcards packaging",
      "description": "The gripper or tool should make visible or physically plausible contact with the blue and white playingcards packaging before the main motion."
    },
    {
      "skill_index": 3,
      "verb": "place_relative",
      "active_arm": "left",
      "object": "the blue and white playingcards packaging",
      "target": "the soap bar with curved edges",
      "description": "The robot performs the action: place the blue and white playingcards packaging to the left of the soap bar with curved edges. Parsed relation: left of."
    },
    {
      "skill_index": 4,
      "verb": "finish",
      "active_arm": "left",
      "object": "the blue and white playingcards packaging",
      "target": "the soap bar with curved edges",
      "description": "The motion ends with the blue and white playingcards packaging in a stable state relative to the soap bar with curved edges."
    }
  ],
  "key_state_changes": [
    "the blue and white playingcards packaging changes pose or state according to the task intent",
    "the gripper state may open or close near the contact moment",
    "the final state should preserve the relation between the blue and white playingcards packaging and the soap bar with curved edges"
  ],
  "physical_relations": {
    "contact": "motion of the blue and white playingcards packaging should be explained by gripper, end-effector, or tool contact",
    "support": "the blue and white playingcards packaging should remain supported by the gripper, tabletop, receptacle, or target structure",
    "containment": "not required unless visible",
    "violations_to_avoid": [
      "object teleportation",
      "unexplained floating motion",
      "severe penetration through rigid objects",
      "non-rigid deformation of the robot or target object"
    ]
  },
  "task_result_note": "Use visible evidence from the clip; if an object or contact is not visible, keep the description generic.",
  "caption_quality_flags": [
    "template_exact"
  ]
}
```

**Stage 2 Dense Caption:**

[Phase 1] Scene Setup
A fixed head-camera view observes an Aloha-AgileX-style dual-arm gripper robot in a tabletop workspace. The parsed manipulated object is the blue and white playingcards packaging. The parsed target or reference is the soap bar with curved edges, and the intended spatial relation is left of. The task family is pick_place, so the description focuses on visible rigid-body manipulation rather than appearance embellishment.

[Phase 2] Action Detail
The robot uses the left arm to perform the action: place the blue and white playingcards packaging to the left of the soap bar with curved edges. The motion should begin with an approach toward the blue and white playingcards packaging, followed by contact, grasping, pressing, or tool interaction as appropriate for the task. During the main action, the blue and white playingcards packaging should move with the gripper or tool rather than independently, and the trajectory should remain continuous in the fixed camera view.

[Phase 3] State Transition
The expected state transition is that the blue and white playingcards packaging moves from its initial tabletop pose to left of the soap bar with curved edges. The final arrangement should be stable and physically supported by the tabletop, the gripper, the receptacle, or the referenced object. The clip should not show object teleportation, unexplained floating, severe penetration, or deformation of rigid objects.

[Phase 4] Camera Summary
The camera remains static and documents the manipulation in an observation-style view. White tabletop background, mild simulator grain, and partial robot visibility are normal for this domain. The caption avoids claiming contact or object identity beyond what is parsed from the instruction and task metadata.

## rt_000002 / pick_place / place_a2b_left

**source_instruction:** Use the left arm, move the dark brown phone with smooth surface left of the bread loaf with bumpy surface
**roles:** active_arm=left | verb=place | A=the dark brown phone with smooth surface | target=the bread loaf with bumpy surface | relation=left of | confidence=0.95
**flags:** template_exact

**caption_short:** A dual-arm Aloha-AgileX robot uses its left arm to place the dark brown phone with smooth surface to the left of the bread loaf with bumpy surface.

**caption_worldarena_style:** In a fixed robotic workspace, generate a rigid, physically consistent embodied robotic arm. The arm maintains high stability with no deformation and enters the frame to place the dark brown phone with smooth surface to the left of the bread loaf with bumpy surface.

**Stage 1 JSON:**
```json
{
  "high_level_task": {
    "task_family": "pick_place",
    "robotwin_task_name": "place_a2b_left",
    "intent": "place the dark brown phone with smooth surface to the left of the bread loaf with bumpy surface",
    "source_instruction": "Use the left arm, move the dark brown phone with smooth surface left of the bread loaf with bumpy surface"
  },
  "role_parse": {
    "active_arm": "left",
    "main_verb": "place",
    "manipulated_object": "the dark brown phone with smooth surface",
    "target_object": "the bread loaf with bumpy surface",
    "target_receptacle": "",
    "spatial_relation": "left of",
    "secondary_action": "",
    "primary_action": "place_relative",
    "pre_action": "",
    "compound_instruction": "false",
    "unresolved_pronoun": "false",
    "target_pollution_fixed": "false",
    "parse_confidence": 0.95
  },
  "robot_initial_state": {
    "morphology": "Aloha-AgileX-style dual-arm gripper robot",
    "visible_assumption": "robot arms may enter the fixed camera view from outside the frame",
    "initial_contact": "no contact is assumed before approach unless visible in the video"
  },
  "scene_context": {
    "location_type": "fixed tabletop robotic workspace",
    "camera_view": "static head-camera view, 640x480 video rendered from a fixed workspace camera",
    "objects": [
      "the dark brown phone with smooth surface"
    ],
    "static_objects": [
      "the bread loaf with bumpy surface"
    ]
  },
  "skill_sequence": [
    {
      "skill_index": 1,
      "verb": "approach",
      "active_arm": "left",
      "object": "the dark brown phone with smooth surface",
      "target": "workspace contact region",
      "description": "The active gripper approaches the dark brown phone with smooth surface while staying in the fixed tabletop workspace."
    },
    {
      "skill_index": 2,
      "verb": "contact_or_grasp",
      "active_arm": "left",
      "object": "the dark brown phone with smooth surface",
      "target": "the dark brown phone with smooth surface",
      "description": "The gripper or tool should make visible or physically plausible contact with the dark brown phone with smooth surface before the main motion."
    },
    {
      "skill_index": 3,
      "verb": "place_relative",
      "active_arm": "left",
      "object": "the dark brown phone with smooth surface",
      "target": "the bread loaf with bumpy surface",
      "description": "The robot performs the action: place the dark brown phone with smooth surface to the left of the bread loaf with bumpy surface. Parsed relation: left of."
    },
    {
      "skill_index": 4,
      "verb": "finish",
      "active_arm": "left",
      "object": "the dark brown phone with smooth surface",
      "target": "the bread loaf with bumpy surface",
      "description": "The motion ends with the dark brown phone with smooth surface in a stable state relative to the bread loaf with bumpy surface."
    }
  ],
  "key_state_changes": [
    "the dark brown phone with smooth surface changes pose or state according to the task intent",
    "the gripper state may open or close near the contact moment",
    "the final state should preserve the relation between the dark brown phone with smooth surface and the bread loaf with bumpy surface"
  ],
  "physical_relations": {
    "contact": "motion of the dark brown phone with smooth surface should be explained by gripper, end-effector, or tool contact",
    "support": "the dark brown phone with smooth surface should remain supported by the gripper, tabletop, receptacle, or target structure",
    "containment": "not required unless visible",
    "violations_to_avoid": [
      "object teleportation",
      "unexplained floating motion",
      "severe penetration through rigid objects",
      "non-rigid deformation of the robot or target object"
    ]
  },
  "task_result_note": "Use visible evidence from the clip; if an object or contact is not visible, keep the description generic.",
  "caption_quality_flags": [
    "template_exact"
  ]
}
```

**Stage 2 Dense Caption:**

[Phase 1] Scene Setup
A fixed head-camera view observes an Aloha-AgileX-style dual-arm gripper robot in a tabletop workspace. The parsed manipulated object is the dark brown phone with smooth surface. The parsed target or reference is the bread loaf with bumpy surface, and the intended spatial relation is left of. The task family is pick_place, so the description focuses on visible rigid-body manipulation rather than appearance embellishment.

[Phase 2] Action Detail
The robot uses the left arm to perform the action: place the dark brown phone with smooth surface to the left of the bread loaf with bumpy surface. The motion should begin with an approach toward the dark brown phone with smooth surface, followed by contact, grasping, pressing, or tool interaction as appropriate for the task. During the main action, the dark brown phone with smooth surface should move with the gripper or tool rather than independently, and the trajectory should remain continuous in the fixed camera view.

[Phase 3] State Transition
The expected state transition is that the dark brown phone with smooth surface moves from its initial tabletop pose to left of the bread loaf with bumpy surface. The final arrangement should be stable and physically supported by the tabletop, the gripper, the receptacle, or the referenced object. The clip should not show object teleportation, unexplained floating, severe penetration, or deformation of rigid objects.

[Phase 4] Camera Summary
The camera remains static and documents the manipulation in an observation-style view. White tabletop background, mild simulator grain, and partial robot visibility are normal for this domain. The caption avoids claiming contact or object identity beyond what is parsed from the instruction and task metadata.

## rt_000003 / pick_place / place_a2b_left

**source_instruction:** Set the glossy blue toycar directly to the left of the playingcards box with white symbols
**roles:** active_arm=unknown | verb=place | A=the glossy blue toycar | target=the playingcards box with white symbols | relation=left of | confidence=0.95
**flags:** template_exact

**caption_short:** A dual-arm Aloha-AgileX robot places the glossy blue toycar to the left of the playingcards box with white symbols.

**caption_worldarena_style:** In a fixed robotic workspace, generate a rigid, physically consistent embodied robotic arm. The arm maintains high stability with no deformation and enters the frame to place the glossy blue toycar to the left of the playingcards box with white symbols.

**Stage 1 JSON:**
```json
{
  "high_level_task": {
    "task_family": "pick_place",
    "robotwin_task_name": "place_a2b_left",
    "intent": "place the glossy blue toycar to the left of the playingcards box with white symbols",
    "source_instruction": "Set the glossy blue toycar directly to the left of the playingcards box with white symbols"
  },
  "role_parse": {
    "active_arm": "unknown",
    "main_verb": "place",
    "manipulated_object": "the glossy blue toycar",
    "target_object": "the playingcards box with white symbols",
    "target_receptacle": "",
    "spatial_relation": "left of",
    "secondary_action": "",
    "primary_action": "place_relative",
    "pre_action": "",
    "compound_instruction": "false",
    "unresolved_pronoun": "false",
    "target_pollution_fixed": "false",
    "parse_confidence": 0.95
  },
  "robot_initial_state": {
    "morphology": "Aloha-AgileX-style dual-arm gripper robot",
    "visible_assumption": "robot arms may enter the fixed camera view from outside the frame",
    "initial_contact": "no contact is assumed before approach unless visible in the video"
  },
  "scene_context": {
    "location_type": "fixed tabletop robotic workspace",
    "camera_view": "static head-camera view, 640x480 video rendered from a fixed workspace camera",
    "objects": [
      "the glossy blue toycar"
    ],
    "static_objects": [
      "the playingcards box with white symbols"
    ]
  },
  "skill_sequence": [
    {
      "skill_index": 1,
      "verb": "approach",
      "active_arm": "unknown",
      "object": "the glossy blue toycar",
      "target": "workspace contact region",
      "description": "The active gripper approaches the glossy blue toycar while staying in the fixed tabletop workspace."
    },
    {
      "skill_index": 2,
      "verb": "contact_or_grasp",
      "active_arm": "unknown",
      "object": "the glossy blue toycar",
      "target": "the glossy blue toycar",
      "description": "The gripper or tool should make visible or physically plausible contact with the glossy blue toycar before the main motion."
    },
    {
      "skill_index": 3,
      "verb": "place_relative",
      "active_arm": "unknown",
      "object": "the glossy blue toycar",
      "target": "the playingcards box with white symbols",
      "description": "The robot performs the action: place the glossy blue toycar to the left of the playingcards box with white symbols. Parsed relation: left of."
    },
    {
      "skill_index": 4,
      "verb": "finish",
      "active_arm": "unknown",
      "object": "the glossy blue toycar",
      "target": "the playingcards box with white symbols",
      "description": "The motion ends with the glossy blue toycar in a stable state relative to the playingcards box with white symbols."
    }
  ],
  "key_state_changes": [
    "the glossy blue toycar changes pose or state according to the task intent",
    "the gripper state may open or close near the contact moment",
    "the final state should preserve the relation between the glossy blue toycar and the playingcards box with white symbols"
  ],
  "physical_relations": {
    "contact": "motion of the glossy blue toycar should be explained by gripper, end-effector, or tool contact",
    "support": "the glossy blue toycar should remain supported by the gripper, tabletop, receptacle, or target structure",
    "containment": "not required unless visible",
    "violations_to_avoid": [
      "object teleportation",
      "unexplained floating motion",
      "severe penetration through rigid objects",
      "non-rigid deformation of the robot or target object"
    ]
  },
  "task_result_note": "Use visible evidence from the clip; if an object or contact is not visible, keep the description generic.",
  "caption_quality_flags": [
    "template_exact"
  ]
}
```

**Stage 2 Dense Caption:**

[Phase 1] Scene Setup
A fixed head-camera view observes an Aloha-AgileX-style dual-arm gripper robot in a tabletop workspace. The parsed manipulated object is the glossy blue toycar. The parsed target or reference is the playingcards box with white symbols, and the intended spatial relation is left of. The task family is pick_place, so the description focuses on visible rigid-body manipulation rather than appearance embellishment.

[Phase 2] Action Detail
The robot uses an active gripper to perform the action: place the glossy blue toycar to the left of the playingcards box with white symbols. The motion should begin with an approach toward the glossy blue toycar, followed by contact, grasping, pressing, or tool interaction as appropriate for the task. During the main action, the glossy blue toycar should move with the gripper or tool rather than independently, and the trajectory should remain continuous in the fixed camera view.

[Phase 3] State Transition
The expected state transition is that the glossy blue toycar moves from its initial tabletop pose to left of the playingcards box with white symbols. The final arrangement should be stable and physically supported by the tabletop, the gripper, the receptacle, or the referenced object. The clip should not show object teleportation, unexplained floating, severe penetration, or deformation of rigid objects.

[Phase 4] Camera Summary
The camera remains static and documents the manipulation in an observation-style view. White tabletop background, mild simulator grain, and partial robot visibility are normal for this domain. The caption avoids claiming contact or object identity beyond what is parsed from the instruction and task metadata.

## rt_000004 / pick_place / place_a2b_left

**source_instruction:** Grab the golden brown bread using the right arm and place left of the coffee-box with cup image on top.
**roles:** active_arm=right | verb=place | A=the golden brown bread | target=the coffee-box with cup image on top | relation=left of | confidence=0.95
**flags:** template_exact

**caption_short:** A dual-arm Aloha-AgileX robot uses its right arm to place the golden brown bread to the left of the coffee-box with cup image on top.

**caption_worldarena_style:** In a fixed robotic workspace, generate a rigid, physically consistent embodied robotic arm. The arm maintains high stability with no deformation and enters the frame to place the golden brown bread to the left of the coffee-box with cup image on top.

**Stage 1 JSON:**
```json
{
  "high_level_task": {
    "task_family": "pick_place",
    "robotwin_task_name": "place_a2b_left",
    "intent": "place the golden brown bread to the left of the coffee-box with cup image on top",
    "source_instruction": "Grab the golden brown bread using the right arm and place left of the coffee-box with cup image on top."
  },
  "role_parse": {
    "active_arm": "right",
    "main_verb": "place",
    "manipulated_object": "the golden brown bread",
    "target_object": "the coffee-box with cup image on top",
    "target_receptacle": "",
    "spatial_relation": "left of",
    "secondary_action": "",
    "primary_action": "place_relative",
    "pre_action": "lift_or_grasp",
    "compound_instruction": "false",
    "unresolved_pronoun": "false",
    "target_pollution_fixed": "false",
    "parse_confidence": 0.95
  },
  "robot_initial_state": {
    "morphology": "Aloha-AgileX-style dual-arm gripper robot",
    "visible_assumption": "robot arms may enter the fixed camera view from outside the frame",
    "initial_contact": "no contact is assumed before approach unless visible in the video"
  },
  "scene_context": {
    "location_type": "fixed tabletop robotic workspace",
    "camera_view": "static head-camera view, 640x480 video rendered from a fixed workspace camera",
    "objects": [
      "the golden brown bread"
    ],
    "static_objects": [
      "the coffee-box with cup image on top"
    ]
  },
  "skill_sequence": [
    {
      "skill_index": 1,
      "verb": "lift_or_grasp",
      "active_arm": "right",
      "object": "the golden brown bread",
      "target": "the golden brown bread",
      "description": "The robot first performs a pre-action such as lifting or grasping the golden brown bread before the placement step."
    },
    {
      "skill_index": 1,
      "verb": "approach",
      "active_arm": "right",
      "object": "the golden brown bread",
      "target": "workspace contact region",
      "description": "The active gripper approaches the golden brown bread while staying in the fixed tabletop workspace."
    },
    {
      "skill_index": 2,
      "verb": "contact_or_grasp",
      "active_arm": "right",
      "object": "the golden brown bread",
      "target": "the golden brown bread",
      "description": "The gripper or tool should make visible or physically plausible contact with the golden brown bread before the main motion."
    },
    {
      "skill_index": 3,
      "verb": "place_relative",
      "active_arm": "right",
      "object": "the golden brown bread",
      "target": "the coffee-box with cup image on top",
      "description": "The robot performs the action: place the golden brown bread to the left of the coffee-box with cup image on top. Parsed relation: left of."
    },
    {
      "skill_index": 4,
      "verb": "finish",
      "active_arm": "right",
      "object": "the golden brown bread",
      "target": "the coffee-box with cup image on top",
      "description": "The motion ends with the golden brown bread in a stable state relative to the coffee-box with cup image on top."
    }
  ],
  "key_state_changes": [
    "the golden brown bread changes pose or state according to the task intent",
    "the gripper state may open or close near the contact moment",
    "the final state should preserve the relation between the golden brown bread and the coffee-box with cup image on top"
  ],
  "physical_relations": {
    "contact": "motion of the golden brown bread should be explained by gripper, end-effector, or tool contact",
    "support": "the golden brown bread should remain supported by the gripper, tabletop, receptacle, or target structure",
    "containment": "not required unless visible",
    "violations_to_avoid": [
      "object teleportation",
      "unexplained floating motion",
      "severe penetration through rigid objects",
      "non-rigid deformation of the robot or target object"
    ]
  },
  "task_result_note": "Use visible evidence from the clip; if an object or contact is not visible, keep the description generic.",
  "caption_quality_flags": [
    "template_exact"
  ]
}
```

**Stage 2 Dense Caption:**

[Phase 1] Scene Setup
A fixed head-camera view observes an Aloha-AgileX-style dual-arm gripper robot in a tabletop workspace. The parsed manipulated object is the golden brown bread. The parsed target or reference is the coffee-box with cup image on top, and the intended spatial relation is left of. The task family is pick_place, so the description focuses on visible rigid-body manipulation rather than appearance embellishment.

[Phase 2] Action Detail
The robot uses the right arm to perform the action: place the golden brown bread to the left of the coffee-box with cup image on top. The sequence should first lift or grasp the golden brown bread, establishing contact before the placement step. The motion should begin with an approach toward the golden brown bread, followed by contact, grasping, pressing, or tool interaction as appropriate for the task. During the main action, the golden brown bread should move with the gripper or tool rather than independently, and the trajectory should remain continuous in the fixed camera view.

[Phase 3] State Transition
The expected state transition is that the golden brown bread moves from its initial tabletop pose to left of the coffee-box with cup image on top. The final arrangement should be stable and physically supported by the tabletop, the gripper, the receptacle, or the referenced object. The clip should not show object teleportation, unexplained floating, severe penetration, or deformation of rigid objects.

[Phase 4] Camera Summary
The camera remains static and documents the manipulation in an observation-style view. White tabletop background, mild simulator grain, and partial robot visibility are normal for this domain. The caption avoids claiming contact or object identity beyond what is parsed from the instruction and task metadata.

## rt_000005 / pick_place / place_a2b_right

**source_instruction:** Grab the compact stapler with smooth surface using the right arm, then move it to the woodenblock medium size smooth texture's right.
**roles:** active_arm=right | verb=place | A=object A | target=object B | relation=right of | confidence=0.95
**flags:** template_exact;compound_instruction;secondary_action_parsed

**caption_short:** A dual-arm Aloha-AgileX robot uses its right arm to place object A to the right of object B.

**caption_worldarena_style:** In a fixed robotic workspace, generate a rigid, physically consistent embodied robotic arm. The arm maintains high stability with no deformation and enters the frame to place object A to the right of object B.

**Stage 1 JSON:**
```json
{
  "high_level_task": {
    "task_family": "pick_place",
    "robotwin_task_name": "place_a2b_right",
    "intent": "place object A to the right of object B",
    "source_instruction": "Grab the compact stapler with smooth surface using the right arm, then move it to the woodenblock medium size smooth texture's right."
  },
  "role_parse": {
    "active_arm": "right",
    "main_verb": "place",
    "manipulated_object": "object A",
    "target_object": "object B",
    "target_receptacle": "",
    "spatial_relation": "right of",
    "secondary_action": "move it to the woodenblock medium size smooth texture's right",
    "primary_action": "",
    "pre_action": "",
    "compound_instruction": "true",
    "unresolved_pronoun": "false",
    "target_pollution_fixed": "false",
    "parse_confidence": 0.95
  },
  "robot_initial_state": {
    "morphology": "Aloha-AgileX-style dual-arm gripper robot",
    "visible_assumption": "robot arms may enter the fixed camera view from outside the frame",
    "initial_contact": "no contact is assumed before approach unless visible in the video"
  },
  "scene_context": {
    "location_type": "fixed tabletop robotic workspace",
    "camera_view": "static head-camera view, 640x480 video rendered from a fixed workspace camera",
    "objects": [
      "object A"
    ],
    "static_objects": [
      "object B"
    ]
  },
  "skill_sequence": [
    {
      "skill_index": 1,
      "verb": "approach",
      "active_arm": "right",
      "object": "object A",
      "target": "workspace contact region",
      "description": "The active gripper approaches object A while staying in the fixed tabletop workspace."
    },
    {
      "skill_index": 2,
      "verb": "contact_or_grasp",
      "active_arm": "right",
      "object": "object A",
      "target": "object A",
      "description": "The gripper or tool should make visible or physically plausible contact with object A before the main motion."
    },
    {
      "skill_index": 3,
      "verb": "place",
      "active_arm": "right",
      "object": "object A",
      "target": "object B",
      "description": "The robot performs the action: place object A to the right of object B. Parsed relation: right of."
    },
    {
      "skill_index": 4,
      "verb": "secondary_action",
      "active_arm": "right",
      "object": "object A",
      "target": "object B",
      "description": "move it to the woodenblock medium size smooth texture's right"
    },
    {
      "skill_index": 5,
      "verb": "finish",
      "active_arm": "right",
      "object": "object A",
      "target": "object B",
      "description": "The motion ends with object A in a stable state relative to object B."
    }
  ],
  "key_state_changes": [
    "object A changes pose or state according to the task intent",
    "the gripper state may open or close near the contact moment",
    "the final state should preserve the relation between object A and object B"
  ],
  "physical_relations": {
    "contact": "motion of object A should be explained by gripper, end-effector, or tool contact",
    "support": "object A should remain supported by the gripper, tabletop, receptacle, or target structure",
    "containment": "not required unless visible",
    "violations_to_avoid": [
      "object teleportation",
      "unexplained floating motion",
      "severe penetration through rigid objects",
      "non-rigid deformation of the robot or target object"
    ]
  },
  "task_result_note": "Use visible evidence from the clip; if an object or contact is not visible, keep the description generic.",
  "caption_quality_flags": [
    "template_exact",
    "compound_instruction",
    "secondary_action_parsed"
  ]
}
```

**Stage 2 Dense Caption:**

[Phase 1] Scene Setup
A fixed head-camera view observes an Aloha-AgileX-style dual-arm gripper robot in a tabletop workspace. The parsed manipulated object is object A. The parsed target or reference is object B, and the intended spatial relation is right of. The task family is pick_place, so the description focuses on visible rigid-body manipulation rather than appearance embellishment.

[Phase 2] Action Detail
The robot uses the right arm to perform the action: place object A to the right of object B. The motion should begin with an approach toward object A, followed by contact, grasping, pressing, or tool interaction as appropriate for the task. During the main action, object A should move with the gripper or tool rather than independently, and the trajectory should remain continuous in the fixed camera view. After the primary action, the instruction also includes a secondary action: move it to the woodenblock medium size smooth texture's right.

[Phase 3] State Transition
The expected state transition is that object A moves from its initial tabletop pose to right of object B. The final arrangement should be stable and physically supported by the tabletop, the gripper, the receptacle, or the referenced object. The clip should not show object teleportation, unexplained floating, severe penetration, or deformation of rigid objects.

[Phase 4] Camera Summary
The camera remains static and documents the manipulation in an observation-style view. White tabletop background, mild simulator grain, and partial robot visibility are normal for this domain. The caption avoids claiming contact or object identity beyond what is parsed from the instruction and task metadata.

## rt_000006 / pick_place / place_a2b_right

**source_instruction:** Place the gray mouse with smooth surface on the right of the office stapler with plastic and metal parts.
**roles:** active_arm=unknown | verb=place | A=the gray mouse with smooth surface on the | target=the office stapler with plastic and metal parts | relation=right of | confidence=0.95
**flags:** template_exact

**caption_short:** A dual-arm Aloha-AgileX robot places the gray mouse with smooth surface on the to the right of the office stapler with plastic and metal parts.

**caption_worldarena_style:** In a fixed robotic workspace, generate a rigid, physically consistent embodied robotic arm. The arm maintains high stability with no deformation and enters the frame to place the gray mouse with smooth surface on the to the right of the office stapler with plastic and metal parts.

**Stage 1 JSON:**
```json
{
  "high_level_task": {
    "task_family": "pick_place",
    "robotwin_task_name": "place_a2b_right",
    "intent": "place the gray mouse with smooth surface on the to the right of the office stapler with plastic and metal parts",
    "source_instruction": "Place the gray mouse with smooth surface on the right of the office stapler with plastic and metal parts."
  },
  "role_parse": {
    "active_arm": "unknown",
    "main_verb": "place",
    "manipulated_object": "the gray mouse with smooth surface on the",
    "target_object": "the office stapler with plastic and metal parts",
    "target_receptacle": "",
    "spatial_relation": "right of",
    "secondary_action": "",
    "primary_action": "place_relative",
    "pre_action": "",
    "compound_instruction": "false",
    "unresolved_pronoun": "false",
    "target_pollution_fixed": "false",
    "parse_confidence": 0.95
  },
  "robot_initial_state": {
    "morphology": "Aloha-AgileX-style dual-arm gripper robot",
    "visible_assumption": "robot arms may enter the fixed camera view from outside the frame",
    "initial_contact": "no contact is assumed before approach unless visible in the video"
  },
  "scene_context": {
    "location_type": "fixed tabletop robotic workspace",
    "camera_view": "static head-camera view, 640x480 video rendered from a fixed workspace camera",
    "objects": [
      "the gray mouse with smooth surface on the"
    ],
    "static_objects": [
      "the office stapler with plastic and metal parts"
    ]
  },
  "skill_sequence": [
    {
      "skill_index": 1,
      "verb": "approach",
      "active_arm": "unknown",
      "object": "the gray mouse with smooth surface on the",
      "target": "workspace contact region",
      "description": "The active gripper approaches the gray mouse with smooth surface on the while staying in the fixed tabletop workspace."
    },
    {
      "skill_index": 2,
      "verb": "contact_or_grasp",
      "active_arm": "unknown",
      "object": "the gray mouse with smooth surface on the",
      "target": "the gray mouse with smooth surface on the",
      "description": "The gripper or tool should make visible or physically plausible contact with the gray mouse with smooth surface on the before the main motion."
    },
    {
      "skill_index": 3,
      "verb": "place_relative",
      "active_arm": "unknown",
      "object": "the gray mouse with smooth surface on the",
      "target": "the office stapler with plastic and metal parts",
      "description": "The robot performs the action: place the gray mouse with smooth surface on the to the right of the office stapler with plastic and metal parts. Parsed relation: right of."
    },
    {
      "skill_index": 4,
      "verb": "finish",
      "active_arm": "unknown",
      "object": "the gray mouse with smooth surface on the",
      "target": "the office stapler with plastic and metal parts",
      "description": "The motion ends with the gray mouse with smooth surface on the in a stable state relative to the office stapler with plastic and metal parts."
    }
  ],
  "key_state_changes": [
    "the gray mouse with smooth surface on the changes pose or state according to the task intent",
    "the gripper state may open or close near the contact moment",
    "the final state should preserve the relation between the gray mouse with smooth surface on the and the office stapler with plastic and metal parts"
  ],
  "physical_relations": {
    "contact": "motion of the gray mouse with smooth surface on the should be explained by gripper, end-effector, or tool contact",
    "support": "the gray mouse with smooth surface on the should remain supported by the gripper, tabletop, receptacle, or target structure",
    "containment": "not required unless visible",
    "violations_to_avoid": [
      "object teleportation",
      "unexplained floating motion",
      "severe penetration through rigid objects",
      "non-rigid deformation of the robot or target object"
    ]
  },
  "task_result_note": "Use visible evidence from the clip; if an object or contact is not visible, keep the description generic.",
  "caption_quality_flags": [
    "template_exact"
  ]
}
```

**Stage 2 Dense Caption:**

[Phase 1] Scene Setup
A fixed head-camera view observes an Aloha-AgileX-style dual-arm gripper robot in a tabletop workspace. The parsed manipulated object is the gray mouse with smooth surface on the. The parsed target or reference is the office stapler with plastic and metal parts, and the intended spatial relation is right of. The task family is pick_place, so the description focuses on visible rigid-body manipulation rather than appearance embellishment.

[Phase 2] Action Detail
The robot uses an active gripper to perform the action: place the gray mouse with smooth surface on the to the right of the office stapler with plastic and metal parts. The motion should begin with an approach toward the gray mouse with smooth surface on the, followed by contact, grasping, pressing, or tool interaction as appropriate for the task. During the main action, the gray mouse with smooth surface on the should move with the gripper or tool rather than independently, and the trajectory should remain continuous in the fixed camera view.

[Phase 3] State Transition
The expected state transition is that the gray mouse with smooth surface on the moves from its initial tabletop pose to right of the office stapler with plastic and metal parts. The final arrangement should be stable and physically supported by the tabletop, the gripper, the receptacle, or the referenced object. The clip should not show object teleportation, unexplained floating, severe penetration, or deformation of rigid objects.

[Phase 4] Camera Summary
The camera remains static and documents the manipulation in an observation-style view. White tabletop background, mild simulator grain, and partial robot visibility are normal for this domain. The caption avoids claiming contact or object identity beyond what is parsed from the instruction and task metadata.

## rt_000007 / pick_place / place_a2b_right

**source_instruction:** Use the right arm to set the soap with rounded corners right of the plastic mouse with smooth finish.
**roles:** active_arm=right | verb=place | A=the soap with rounded corners | target=the plastic mouse with smooth finish | relation=right of | confidence=0.95
**flags:** template_exact

**caption_short:** A dual-arm Aloha-AgileX robot uses its right arm to place the soap with rounded corners to the right of the plastic mouse with smooth finish.

**caption_worldarena_style:** In a fixed robotic workspace, generate a rigid, physically consistent embodied robotic arm. The arm maintains high stability with no deformation and enters the frame to place the soap with rounded corners to the right of the plastic mouse with smooth finish.

**Stage 1 JSON:**
```json
{
  "high_level_task": {
    "task_family": "pick_place",
    "robotwin_task_name": "place_a2b_right",
    "intent": "place the soap with rounded corners to the right of the plastic mouse with smooth finish",
    "source_instruction": "Use the right arm to set the soap with rounded corners right of the plastic mouse with smooth finish."
  },
  "role_parse": {
    "active_arm": "right",
    "main_verb": "place",
    "manipulated_object": "the soap with rounded corners",
    "target_object": "the plastic mouse with smooth finish",
    "target_receptacle": "",
    "spatial_relation": "right of",
    "secondary_action": "",
    "primary_action": "place_relative",
    "pre_action": "",
    "compound_instruction": "false",
    "unresolved_pronoun": "false",
    "target_pollution_fixed": "false",
    "parse_confidence": 0.95
  },
  "robot_initial_state": {
    "morphology": "Aloha-AgileX-style dual-arm gripper robot",
    "visible_assumption": "robot arms may enter the fixed camera view from outside the frame",
    "initial_contact": "no contact is assumed before approach unless visible in the video"
  },
  "scene_context": {
    "location_type": "fixed tabletop robotic workspace",
    "camera_view": "static head-camera view, 640x480 video rendered from a fixed workspace camera",
    "objects": [
      "the soap with rounded corners"
    ],
    "static_objects": [
      "the plastic mouse with smooth finish"
    ]
  },
  "skill_sequence": [
    {
      "skill_index": 1,
      "verb": "approach",
      "active_arm": "right",
      "object": "the soap with rounded corners",
      "target": "workspace contact region",
      "description": "The active gripper approaches the soap with rounded corners while staying in the fixed tabletop workspace."
    },
    {
      "skill_index": 2,
      "verb": "contact_or_grasp",
      "active_arm": "right",
      "object": "the soap with rounded corners",
      "target": "the soap with rounded corners",
      "description": "The gripper or tool should make visible or physically plausible contact with the soap with rounded corners before the main motion."
    },
    {
      "skill_index": 3,
      "verb": "place_relative",
      "active_arm": "right",
      "object": "the soap with rounded corners",
      "target": "the plastic mouse with smooth finish",
      "description": "The robot performs the action: place the soap with rounded corners to the right of the plastic mouse with smooth finish. Parsed relation: right of."
    },
    {
      "skill_index": 4,
      "verb": "finish",
      "active_arm": "right",
      "object": "the soap with rounded corners",
      "target": "the plastic mouse with smooth finish",
      "description": "The motion ends with the soap with rounded corners in a stable state relative to the plastic mouse with smooth finish."
    }
  ],
  "key_state_changes": [
    "the soap with rounded corners changes pose or state according to the task intent",
    "the gripper state may open or close near the contact moment",
    "the final state should preserve the relation between the soap with rounded corners and the plastic mouse with smooth finish"
  ],
  "physical_relations": {
    "contact": "motion of the soap with rounded corners should be explained by gripper, end-effector, or tool contact",
    "support": "the soap with rounded corners should remain supported by the gripper, tabletop, receptacle, or target structure",
    "containment": "not required unless visible",
    "violations_to_avoid": [
      "object teleportation",
      "unexplained floating motion",
      "severe penetration through rigid objects",
      "non-rigid deformation of the robot or target object"
    ]
  },
  "task_result_note": "Use visible evidence from the clip; if an object or contact is not visible, keep the description generic.",
  "caption_quality_flags": [
    "template_exact"
  ]
}
```

**Stage 2 Dense Caption:**

[Phase 1] Scene Setup
A fixed head-camera view observes an Aloha-AgileX-style dual-arm gripper robot in a tabletop workspace. The parsed manipulated object is the soap with rounded corners. The parsed target or reference is the plastic mouse with smooth finish, and the intended spatial relation is right of. The task family is pick_place, so the description focuses on visible rigid-body manipulation rather than appearance embellishment.

[Phase 2] Action Detail
The robot uses the right arm to perform the action: place the soap with rounded corners to the right of the plastic mouse with smooth finish. The motion should begin with an approach toward the soap with rounded corners, followed by contact, grasping, pressing, or tool interaction as appropriate for the task. During the main action, the soap with rounded corners should move with the gripper or tool rather than independently, and the trajectory should remain continuous in the fixed camera view.

[Phase 3] State Transition
The expected state transition is that the soap with rounded corners moves from its initial tabletop pose to right of the plastic mouse with smooth finish. The final arrangement should be stable and physically supported by the tabletop, the gripper, the receptacle, or the referenced object. The clip should not show object teleportation, unexplained floating, severe penetration, or deformation of rigid objects.

[Phase 4] Camera Summary
The camera remains static and documents the manipulation in an observation-style view. White tabletop background, mild simulator grain, and partial robot visibility are normal for this domain. The caption avoids claiming contact or object identity beyond what is parsed from the instruction and task metadata.

## rt_000008 / pick_place / place_a2b_right

**source_instruction:** Use the right arm to place the medium-sized handheld phone right of the decorative woodenblock.
**roles:** active_arm=right | verb=place | A=the medium-sized handheld phone | target=the decorative woodenblock | relation=right of | confidence=0.95
**flags:** template_exact

**caption_short:** A dual-arm Aloha-AgileX robot uses its right arm to place the medium-sized handheld phone to the right of the decorative woodenblock.

**caption_worldarena_style:** In a fixed robotic workspace, generate a rigid, physically consistent embodied robotic arm. The arm maintains high stability with no deformation and enters the frame to place the medium-sized handheld phone to the right of the decorative woodenblock.

**Stage 1 JSON:**
```json
{
  "high_level_task": {
    "task_family": "pick_place",
    "robotwin_task_name": "place_a2b_right",
    "intent": "place the medium-sized handheld phone to the right of the decorative woodenblock",
    "source_instruction": "Use the right arm to place the medium-sized handheld phone right of the decorative woodenblock."
  },
  "role_parse": {
    "active_arm": "right",
    "main_verb": "place",
    "manipulated_object": "the medium-sized handheld phone",
    "target_object": "the decorative woodenblock",
    "target_receptacle": "",
    "spatial_relation": "right of",
    "secondary_action": "",
    "primary_action": "place_relative",
    "pre_action": "",
    "compound_instruction": "false",
    "unresolved_pronoun": "false",
    "target_pollution_fixed": "false",
    "parse_confidence": 0.95
  },
  "robot_initial_state": {
    "morphology": "Aloha-AgileX-style dual-arm gripper robot",
    "visible_assumption": "robot arms may enter the fixed camera view from outside the frame",
    "initial_contact": "no contact is assumed before approach unless visible in the video"
  },
  "scene_context": {
    "location_type": "fixed tabletop robotic workspace",
    "camera_view": "static head-camera view, 640x480 video rendered from a fixed workspace camera",
    "objects": [
      "the medium-sized handheld phone"
    ],
    "static_objects": [
      "the decorative woodenblock"
    ]
  },
  "skill_sequence": [
    {
      "skill_index": 1,
      "verb": "approach",
      "active_arm": "right",
      "object": "the medium-sized handheld phone",
      "target": "workspace contact region",
      "description": "The active gripper approaches the medium-sized handheld phone while staying in the fixed tabletop workspace."
    },
    {
      "skill_index": 2,
      "verb": "contact_or_grasp",
      "active_arm": "right",
      "object": "the medium-sized handheld phone",
      "target": "the medium-sized handheld phone",
      "description": "The gripper or tool should make visible or physically plausible contact with the medium-sized handheld phone before the main motion."
    },
    {
      "skill_index": 3,
      "verb": "place_relative",
      "active_arm": "right",
      "object": "the medium-sized handheld phone",
      "target": "the decorative woodenblock",
      "description": "The robot performs the action: place the medium-sized handheld phone to the right of the decorative woodenblock. Parsed relation: right of."
    },
    {
      "skill_index": 4,
      "verb": "finish",
      "active_arm": "right",
      "object": "the medium-sized handheld phone",
      "target": "the decorative woodenblock",
      "description": "The motion ends with the medium-sized handheld phone in a stable state relative to the decorative woodenblock."
    }
  ],
  "key_state_changes": [
    "the medium-sized handheld phone changes pose or state according to the task intent",
    "the gripper state may open or close near the contact moment",
    "the final state should preserve the relation between the medium-sized handheld phone and the decorative woodenblock"
  ],
  "physical_relations": {
    "contact": "motion of the medium-sized handheld phone should be explained by gripper, end-effector, or tool contact",
    "support": "the medium-sized handheld phone should remain supported by the gripper, tabletop, receptacle, or target structure",
    "containment": "not required unless visible",
    "violations_to_avoid": [
      "object teleportation",
      "unexplained floating motion",
      "severe penetration through rigid objects",
      "non-rigid deformation of the robot or target object"
    ]
  },
  "task_result_note": "Use visible evidence from the clip; if an object or contact is not visible, keep the description generic.",
  "caption_quality_flags": [
    "template_exact"
  ]
}
```

**Stage 2 Dense Caption:**

[Phase 1] Scene Setup
A fixed head-camera view observes an Aloha-AgileX-style dual-arm gripper robot in a tabletop workspace. The parsed manipulated object is the medium-sized handheld phone. The parsed target or reference is the decorative woodenblock, and the intended spatial relation is right of. The task family is pick_place, so the description focuses on visible rigid-body manipulation rather than appearance embellishment.

[Phase 2] Action Detail
The robot uses the right arm to perform the action: place the medium-sized handheld phone to the right of the decorative woodenblock. The motion should begin with an approach toward the medium-sized handheld phone, followed by contact, grasping, pressing, or tool interaction as appropriate for the task. During the main action, the medium-sized handheld phone should move with the gripper or tool rather than independently, and the trajectory should remain continuous in the fixed camera view.

[Phase 3] State Transition
The expected state transition is that the medium-sized handheld phone moves from its initial tabletop pose to right of the decorative woodenblock. The final arrangement should be stable and physically supported by the tabletop, the gripper, the receptacle, or the referenced object. The clip should not show object teleportation, unexplained floating, severe penetration, or deformation of rigid objects.

[Phase 4] Camera Summary
The camera remains static and documents the manipulation in an observation-style view. White tabletop background, mild simulator grain, and partial robot visibility are normal for this domain. The caption avoids claiming contact or object identity beyond what is parsed from the instruction and task metadata.

## rt_000009 / pick_place / place_a2b_right

**source_instruction:** Move the smooth loaf with rounded edges right of the white body stapler with accents with the right arm
**roles:** active_arm=right | verb=place | A=the smooth loaf with rounded edges | target=the white body stapler with accents | relation=right of | confidence=0.95
**flags:** template_exact

**caption_short:** A dual-arm Aloha-AgileX robot uses its right arm to place the smooth loaf with rounded edges to the right of the white body stapler with accents.

**caption_worldarena_style:** In a fixed robotic workspace, generate a rigid, physically consistent embodied robotic arm. The arm maintains high stability with no deformation and enters the frame to place the smooth loaf with rounded edges to the right of the white body stapler with accents.

**Stage 1 JSON:**
```json
{
  "high_level_task": {
    "task_family": "pick_place",
    "robotwin_task_name": "place_a2b_right",
    "intent": "place the smooth loaf with rounded edges to the right of the white body stapler with accents",
    "source_instruction": "Move the smooth loaf with rounded edges right of the white body stapler with accents with the right arm"
  },
  "role_parse": {
    "active_arm": "right",
    "main_verb": "place",
    "manipulated_object": "the smooth loaf with rounded edges",
    "target_object": "the white body stapler with accents",
    "target_receptacle": "",
    "spatial_relation": "right of",
    "secondary_action": "",
    "primary_action": "place_relative",
    "pre_action": "",
    "compound_instruction": "false",
    "unresolved_pronoun": "false",
    "target_pollution_fixed": "false",
    "parse_confidence": 0.95
  },
  "robot_initial_state": {
    "morphology": "Aloha-AgileX-style dual-arm gripper robot",
    "visible_assumption": "robot arms may enter the fixed camera view from outside the frame",
    "initial_contact": "no contact is assumed before approach unless visible in the video"
  },
  "scene_context": {
    "location_type": "fixed tabletop robotic workspace",
    "camera_view": "static head-camera view, 640x480 video rendered from a fixed workspace camera",
    "objects": [
      "the smooth loaf with rounded edges"
    ],
    "static_objects": [
      "the white body stapler with accents"
    ]
  },
  "skill_sequence": [
    {
      "skill_index": 1,
      "verb": "approach",
      "active_arm": "right",
      "object": "the smooth loaf with rounded edges",
      "target": "workspace contact region",
      "description": "The active gripper approaches the smooth loaf with rounded edges while staying in the fixed tabletop workspace."
    },
    {
      "skill_index": 2,
      "verb": "contact_or_grasp",
      "active_arm": "right",
      "object": "the smooth loaf with rounded edges",
      "target": "the smooth loaf with rounded edges",
      "description": "The gripper or tool should make visible or physically plausible contact with the smooth loaf with rounded edges before the main motion."
    },
    {
      "skill_index": 3,
      "verb": "place_relative",
      "active_arm": "right",
      "object": "the smooth loaf with rounded edges",
      "target": "the white body stapler with accents",
      "description": "The robot performs the action: place the smooth loaf with rounded edges to the right of the white body stapler with accents. Parsed relation: right of."
    },
    {
      "skill_index": 4,
      "verb": "finish",
      "active_arm": "right",
      "object": "the smooth loaf with rounded edges",
      "target": "the white body stapler with accents",
      "description": "The motion ends with the smooth loaf with rounded edges in a stable state relative to the white body stapler with accents."
    }
  ],
  "key_state_changes": [
    "the smooth loaf with rounded edges changes pose or state according to the task intent",
    "the gripper state may open or close near the contact moment",
    "the final state should preserve the relation between the smooth loaf with rounded edges and the white body stapler with accents"
  ],
  "physical_relations": {
    "contact": "motion of the smooth loaf with rounded edges should be explained by gripper, end-effector, or tool contact",
    "support": "the smooth loaf with rounded edges should remain supported by the gripper, tabletop, receptacle, or target structure",
    "containment": "not required unless visible",
    "violations_to_avoid": [
      "object teleportation",
      "unexplained floating motion",
      "severe penetration through rigid objects",
      "non-rigid deformation of the robot or target object"
    ]
  },
  "task_result_note": "Use visible evidence from the clip; if an object or contact is not visible, keep the description generic.",
  "caption_quality_flags": [
    "template_exact"
  ]
}
```

**Stage 2 Dense Caption:**

[Phase 1] Scene Setup
A fixed head-camera view observes an Aloha-AgileX-style dual-arm gripper robot in a tabletop workspace. The parsed manipulated object is the smooth loaf with rounded edges. The parsed target or reference is the white body stapler with accents, and the intended spatial relation is right of. The task family is pick_place, so the description focuses on visible rigid-body manipulation rather than appearance embellishment.

[Phase 2] Action Detail
The robot uses the right arm to perform the action: place the smooth loaf with rounded edges to the right of the white body stapler with accents. The motion should begin with an approach toward the smooth loaf with rounded edges, followed by contact, grasping, pressing, or tool interaction as appropriate for the task. During the main action, the smooth loaf with rounded edges should move with the gripper or tool rather than independently, and the trajectory should remain continuous in the fixed camera view.

[Phase 3] State Transition
The expected state transition is that the smooth loaf with rounded edges moves from its initial tabletop pose to right of the white body stapler with accents. The final arrangement should be stable and physically supported by the tabletop, the gripper, the receptacle, or the referenced object. The clip should not show object teleportation, unexplained floating, severe penetration, or deformation of rigid objects.

[Phase 4] Camera Summary
The camera remains static and documents the manipulation in an observation-style view. White tabletop background, mild simulator grain, and partial robot visibility are normal for this domain. The caption avoids claiming contact or object identity beyond what is parsed from the instruction and task metadata.