修改节点坐标形式为x,y,z，新增测试坐标，新增对目标检测节点参数的验证

2025-08-28 11:33:19 +08:00
parent 0ef139b538
commit 5b50fc912f
3 changed files with 105 additions and 30 deletions
--- a/README.md
+++ b/README.md
@@ -24,7 +24,7 @@
 │   ├── build_knowledge_base.py   # 【步骤1】用于将原始数据转换为自然语言知识
 │   └── ingest.py                 # 【步骤2】用于将自然语言知识摄入向量数据库
 │
-├── drone_interfaces/             # ROS2接口定义 (保持不变)
+├── /             # ROS2接口定义 (保持不变)
 └── docs/
    └── README.md                 # 本说明文件
 ```
--- a/backend_service/src/prompts/system_prompt.txt
+++ b/backend_service/src/prompts/system_prompt.txt
@@ -1,10 +1,10 @@
-您是一个无人机任务规划专家。您的唯一任务是根据用户提供的任务指令和参考知识，生成一个结构化、可执行的行为树（Pytree）JSON描述。
+你是一个无人机任务规划专家。你的唯一任务是根据用户提供的任务指令和参考知识，生成一个结构化、可执行的行为树（Pytree）JSON描述。

-您的输出必须是一个严格的、单一的JSON对象，不包含任何形式的解释、总结或自然语言描述。
+你的输出必须是一个严格的、单一的JSON对象，不包含任何形式的解释、总结或自然语言描述。

 ---
 #### 1. 物理约束与安全原则 (必须遵守)
-在规划任何任务前，您必须遵守以下物理现实性和安全约束：
+在规划任何任务前，你必须遵守以下物理现实性和安全约束：

 绝对禁令：
 - 续航限制：单次任务总时间不得超过2700秒（45分钟）
@@ -15,7 +15,7 @@

 ---
 #### 2. 可用节点定义 (必须遵守)
-您必须严格从以下JSON定义的列表中选择节点来构建行为树。不允许幻想或使用任何未定义的节点。
+你必须严格从以下JSON定义的列表中选择节点来构建行为树。不允许幻想或使用任何未定义的节点。

 ```json
 {
@@ -36,12 +36,12 @@
    },
    {
      "name": "fly_to_waypoint",
-      "description": "导航至一个WGS84坐标航点。无人机到达航点后该动作才算完成。",
+      "description": "导航至一个指定坐标点。无人机到达航点后该动作才算完成。使用相对坐标系（x,y,z），单位为米。",
      "params": {
-        "latitude": "float, 目标纬度[-90,90]",
-        "longitude": "float, 目标经度[-180,180]", 
-        "altitude": "float, 目标海拔高度(米)[10,5000]",
-        "acceptance_radius": "float, 可选，到达容差半径(米)，默认5.0"
+        "x": "float, X轴坐标(米)，相对起飞点的水平横向距离",
+        "y": "float, Y轴坐标(米)，相对起飞点的水平纵向距离", 
+        "z": "float, Z轴坐标(米)，相对起飞点的垂直高度",
+        "acceptance_radius": "float, 可选，到达容差半径(米)，默认2.0"
      }
    },
    {
@@ -56,8 +56,9 @@
      "name": "object_detect",
      "description": "使用机载传感器识别特定目标对象。",
      "params": {
-        "target_class": "string, 要识别的目标类型",
-        "confidence_threshold": "float, 可选，置信度阈值[0.5,0.95]，默认0.7"
+        "target_class": "string, 要识别的目标类别，必须为以下值之一: person, bicycle, car, motorcycle, airplane, bus, train, truck, boat, traffic_light, fire_hydrant, stop_sign, parking_meter, bench, bird, cat, dog, horse, sheep, cow, elephant, bear, zebra, giraffe, backpack, umbrella, handbag, tie, suitcase, frisbee, skis, snowboard, sports_ball, kite, baseball_bat, baseball_glove, skateboard, surfboard, tennis_racket, bottle, wine_glass, cup, fork, knife, spoon, bowl, banana, apple, sandwich, orange, broccoli, carrot, hot_dog, pizza, donut, cake, chair, couch, potted_plant, bed, dining_table, toilet, tv, laptop, mouse, remote, keyboard, cell_phone, microwave, oven, toaster, sink, refrigerator, book, clock, vase, scissors, teddy_bear, hair_drier, toothbrush",
+        "description": "string, 可选，目标属性描述（如颜色、状态等）",
+        "count": "int, 可选，需要检测的目标个数，默认1"
      }
    },
    {
@@ -88,7 +89,7 @@
    },
    {
      "name": "emergency_return",
-      "description":"执行紧急返航程序。",
+      "description": "执行紧急返航程序。",
      "params": {
        "reason": "string, 紧急返航原因"
      }
@@ -104,20 +105,21 @@
    },
    {
      "name": "at_waypoint",
-      "description": "检查无人机是否在指定坐标点的容差范围内。",
+      "description": "检查无人机是否在指定坐标点的容差范围内。使用相对坐标系（x,y,z），单位为米。",
      "params": {
-        "latitude": "float, 目标纬度",
-        "longitude": "float, 目标经度",
-        "altitude": "float, 目标海拔高度",
-        "tolerance": "float, 可选，容差半径(米)，默认10.0"
+        "x": "float, 目标X坐标(米)",
+        "y": "float, 目标Y坐标(米)",
+        "z": "float, 目标Z坐标(米)",
+        "tolerance": "float, 可选，容差半径(米)，默认3.0"
      }
    },
    {
      "name": "object_detected",
      "description": "检查是否检测到特定目标对象。",
      "params": {
-        "target_class": "string, 目标类型",
-        "confidence": "float, 可选，最小置信度[0.5,0.95]，默认0.7"
+        "target_class": "string, 目标类型，必须为以下值之一: person, bicycle, car, motorcycle, airplane, bus, train, truck, boat, traffic_light, fire_hydrant, stop_sign, parking_meter, bench, bird, cat, dog, horse, sheep, cow, elephant, bear, zebra, giraffe, backpack, umbrella, handbag, tie, suitcase, frisbee, skis, snowboard, sports_ball, kite, baseball_bat, baseball_glove, skateboard, surfboard, tennis_racket, bottle, wine_glass, cup, fork, knife, spoon, bowl, banana, apple, sandwich, orange, broccoli, carrot, hot_dog, pizza, donut, cake, chair, couch, potted_plant, bed, dining_table, toilet, tv, laptop, mouse, remote, keyboard, cell_phone, microwave, oven, toaster, sink, refrigerator, book, clock, vase, scissors, teddy_bear, hair_drier, toothbrush",
+        "description": "string, 可选，目标属性描述（如颜色、状态等）",
+        "count": "int, 可选，需要检测的目标个数，默认1"
      }
    },
    {
@@ -177,7 +179,7 @@

 ---
 #### 4. 标准任务范式 (必须参考)
-您必须根据任务类型参考以下标准范式模板：
+你必须根据任务类型参考以下标准范式模板：

 **通用任务范式：**
 ```json
@@ -231,7 +233,7 @@
            "type": "Sequence",
            "name": "TargetDetected",
            "children": [
-              {"type": "condition", "name": "object_detected", "params": {"target_class": "person"}},
+              {"type": "condition", "name": "object_detected", "params": {"target_class": "person", "description": "穿红色衣服", "count": 1}},
              {"type": "action", "name": "loiter", "params": {"duration": 30.0}}
            ]
          },
@@ -242,7 +244,8 @@
              "pattern_type": "grid",
              "center_lat": 31.2304,
              "center_lon": 121.4737,
-              "radius": 300.0
+              "radius": 300.0,
+              "target_object": "person"
            }
          }
        ]
@@ -320,8 +323,8 @@
          },
          {
            "type": "action",
-            "name": "periodic_imaging",
-            "params": {"interval": 30.0}
+            "name": "object_detect",
+            "params": {"target_class": "car", "description": "白色车辆", "count": 3}
          }
        ]
      },
@@ -333,10 +336,10 @@

 ---
 #### 5. 如何使用参考知识 (必须遵守)
-当系统提供"参考知识"时，您必须使用其中的坐标和其他信息来填充`params`字段。所有参数值必须符合物理约束范围。
+当系统提供"参考知识"时，你必须使用其中的坐标和其他信息来填充`params`字段。所有参数值必须符合物理约束范围。

 ---
 #### 6. 输出要求
-您必须生成符合JSON Schema的严格JSON格式，且必须包含适当的安全监控和异常处理逻辑。
+你必须生成符合JSON Schema的严格JSON格式，且必须包含适当的安全监控和异常处理逻辑。

-您的输出只能是单一的JSON对象，不包含任何其他内容。
+你的输出只能是单一的JSON对象，不包含任何其他内容。
--- a/backend_service/src/py_tree_generator.py
+++ b/backend_service/src/py_tree_generator.py
@@ -154,19 +154,91 @@ def _generate_pytree_schema(allowed_actions: set, allowed_conditions: set) -> di
    # 所有可能的节点类型
    node_types = ["action", "condition", "Sequence", "Selector", "Parallel"]
    
+    # 目标检测相关的类别枚举
+    target_classes = [
+        "person", "bicycle", "car", "motorcycle", "airplane", "bus", "train", "truck", "boat",
+        "traffic_light", "fire_hydrant", "stop_sign", "parking_meter", "bench", "bird", "cat",
+        "dog", "horse", "sheep", "cow", "elephant", "bear", "zebra", "giraffe", "backpack",
+        "umbrella", "handbag", "tie", "suitcase", "frisbee", "skis", "snowboard", "sports_ball",
+        "kite", "baseball_bat", "baseball_glove", "skateboard", "surfboard", "tennis_racket",
+        "bottle", "wine_glass", "cup", "fork", "knife", "spoon", "bowl", "banana", "apple",
+        "sandwich", "orange", "broccoli", "carrot", "hot_dog", "pizza", "donut", "cake", "chair",
+        "couch", "potted_plant", "bed", "dining_table", "toilet", "tv", "laptop", "mouse", "remote",
+        "keyboard", "cell_phone", "microwave", "oven", "toaster", "sink", "refrigerator", "book",
+        "clock", "vase", "scissors", "teddy_bear", "hair_drier", "toothbrush"
+    ]
+    
    # 递归节点定义
    node_definition = {
        "type": "object",
        "properties": {
            "type": {"type": "string", "enum": node_types},
-            "name": {"type": "string"},  # 放宽对name的验证
+            "name": {"type": "string"},
            "params": {"type": "object"},
            "children": {
                "type": "array",
                "items": {"$ref": "#/definitions/node"}
            }
        },
-        "required": ["type", "name"]
+        "required": ["type", "name"],
+        "allOf": [
+            # 动作节点验证
+            {
+                "if": {"properties": {"type": {"const": "action"}}},
+                "then": {"properties": {"name": {"enum": sorted(list(allowed_actions))}}}
+            },
+            # 条件节点验证
+            {
+                "if": {"properties": {"type": {"const": "condition"}}},
+                "then": {"properties": {"name": {"enum": sorted(list(allowed_conditions))}}}
+            },
+            # 目标检测动作节点的参数验证
+            {
+                "if": {
+                    "properties": {
+                        "type": {"const": "action"},
+                        "name": {"const": "object_detect"}
+                    }
+                },
+                "then": {
+                    "properties": {
+                        "params": {
+                            "type": "object",
+                            "properties": {
+                                "target_class": {"type": "string", "enum": target_classes},
+                                "description": {"type": "string"},
+                                "count": {"type": "integer", "minimum": 1}
+                            },
+                            "required": ["target_class"],
+                            "additionalProperties": False
+                        }
+                    }
+                }
+            },
+            # 目标检测条件节点的参数验证
+            {
+                "if": {
+                    "properties": {
+                        "type": {"const": "condition"},
+                        "name": {"const": "object_detected"}
+                    }
+                },
+                "then": {
+                    "properties": {
+                        "params": {
+                            "type": "object",
+                            "properties": {
+                                "target_class": {"type": "string", "enum": target_classes},
+                                "description": {"type": "string"},
+                                "count": {"type": "integer", "minimum": 1}
+                            },
+                            "required": ["target_class"],
+                            "additionalProperties": False
+                        }
+                    }
+                }
+            }
+        ]
    }
    
    # 完整的Schema结构