nUR MCP Server

Instructions

比较多个机器人的性能

参数:
- robot_ids: 机器人ID列表
- metric_columns: 比较指标列
- start_time: 开始时间戳
- end_time: 结束时间戳

返回:
- 比较结果

Implementation Reference

• src/nonead_universal_robots_mcp/server.py:1185-1214 (handler)

  MCP tool handler for compare_robots_performance. Decorated with @mcp.tool() for automatic registration. Delegates core logic to AdvancedDataAnalyzer.compare_robots() after initialization check.

  @mcp.tool()
  def compare_robots_performance(robot_ids: list, metric_columns: list, start_time: float = None, end_time: float = None):
      """
      比较多个机器人的性能
      
      参数:
      - robot_ids: 机器人ID列表
      - metric_columns: 比较指标列
      - start_time: 开始时间戳
      - end_time: 结束时间戳
      
      返回:
      - 比较结果
      """
      try:
          if advanced_data_analyzer is None:
              return return_msg("高级数据分析器未初始化")
          
          # 执行比较
          comparison = advanced_data_analyzer.compare_robots(
              robot_ids=robot_ids,
              metric_columns=metric_columns,
              start_time=start_time,
              end_time=end_time
          )
          
          return return_msg({"comparison": comparison})
      except Exception as e:
          logger.error(f"比较机器人性能失败: {str(e)}")
          return return_msg(f"比较机器人性能失败: {str(e)}")

• URBasic/advanced_data_analyzer.py:1859-1932 (helper)

  Core helper method implementing the robot comparison logic. Loads data for each robot_id in the time range, computes statistics (mean, median, std, min, max) for each metric_column, and summarizes the best performer per metric (maximizing or minimizing based on metric name).

  def compare_robots(self, 
                    robot_ids: List[str],
                    start_time: Optional[float] = None,
                    end_time: Optional[float] = None,
                    metric_columns: Optional[List[str]] = None) -> Dict[str, Any]:
      """
      比较多个机器人的性能
      
      Args:
          robot_ids: 机器人ID列表
          start_time: 开始时间
          end_time: 结束时间
          metric_columns: 要比较的指标列
          
      Returns:
          Dict[str, Any]: 比较结果
      """
      comparison_results = {
          'robots': {},
          'summary': {}
      }
      
      all_data = []
      
      # 收集每个机器人的数据
      for robot_id in robot_ids:
          df = self.load_data(
              start_time=start_time,
              end_time=end_time,
              robot_id=robot_id
          )
          
          if not df.empty:
              comparison_results['robots'][robot_id] = {
                  'data_points': len(df),
                  'metrics': {}
              }
              
              # 计算指标
              if metric_columns:
                  for col in metric_columns:
                      if col in df.columns:
                          comparison_results['robots'][robot_id]['metrics'][col] = {
                              'mean': df[col].mean(),
                              'median': df[col].median(),
                              'std': df[col].std(),
                              'min': df[col].min(),
                              'max': df[col].max()
                          }
              
              all_data.append((robot_id, df))
      
      # 生成比较摘要
      if metric_columns and all_data:
          for col in metric_columns:
              values = []
              valid_robots = []
              
              for robot_id, df in all_data:
                  if col in df.columns:
                      values.append(df[col].mean())
                      valid_robots.append(robot_id)
              
              if values:
                  best_idx = np.argmin(values) if 'error' in col.lower() or 'temperature' in col.lower() else np.argmax(values)
                  comparison_results['summary'][col] = {
                      'best_robot': valid_robots[best_idx],
                      'best_value': values[best_idx],
                      'average_value': np.mean(values),
                      'range': {'min': min(values), 'max': max(values)},
                      'variation': np.std(values) / np.mean(values) if np.mean(values) > 0 else 0
                  }
      
      return comparison_results