【摘要】 研究目的：基于图像识别的钢筋交叉点定位是实现钢筋绑扎机器人自动化作业的关键。然而，当相机发生大幅度旋转时，相机坐标系与场地（世界）坐标系之间出现较大角度偏转，导致传统三维坐标转换模型（如布尔莎模型）因参数耦合严重而转换精度下降，难以满足高精度绑扎需求。为此，本研究针对大角度偏转条件下的坐标转换问题，改进传统布尔莎模型，提出一种适用于智能钢筋绑扎场景的高精度、低复杂度坐标转换方法。研究结论：（1）所提出的“二维+一维”坐标转换方法有效提升了相机大角度旋转工况下的转换精度，试验结果显示，小坐标值下精度提升约1%，大坐标值下精度提升达5%，满足钢筋绑扎的毫米级精度要求；（2）通过将三维转换解耦为平面与深度两个独立子问题，显著降低了模型求解维度与计算复杂度，避免了传统布尔莎模型中七参数联合迭代带来的收敛困难；（3）新方法扩大了相机的有效运动范围，增强了机器人对复杂姿态变化的适应能力，解决了大角度偏转引起的定位失准问题；（4）该成果可应用于钢筋绑扎机器人等建筑机器人系统，亦可推广至工业检测、室内外导航等需大角度视场变换的场景。更多还原

【Abstract】 Research purposes: The positioning of steel bar intersections based on image recognition is key to achieving automated operation for steel bar binding robots. However, when the camera undergoes significant rotation, a large angular deviation exists between the camera coordinate system and the field(world) coordinate system. This leads to accuracy degradation in traditional three-dimensional coordinate transformation models(such as the Bursa model) due to severe parameter coupling, making it difficult to meet high-precision binding requirements. Therefore, this study addresses the coordinate transformation problem under large angular deviations, improves the traditional Bursa model, and proposes a high-precision, low-complexity coordinate transformation method suitable for intelligent steel bar binding scenarios.Research conclusions:(1) The proposed "two-dimensional + one-dimensional" coordinate transformation method effectively improves conversion accuracy under large camera rotation angles. Experimental results show that precision is increased by approximately 1% for small coordinate values and up to 5% for large coordinate values, meeting the millimeter-level precision requirements for steel bar binding.(2) By decoupling the three-dimensional transformation into two independent sub-problems(plane and depth), the model’s solution dimensions and computational complexity are significantly reduced, avoiding the convergence difficulties caused by the joint iteration of multiple parameters in the traditional Bursa model.(3) The new method expands the effective movement range of the camera and enhances the robot’s adaptability to complex posture changes, solving the positioning deviation problem caused by large angular deviations.(4) This achievement can be applied to steel bar binding robots and other construction robotic systems, and can also be extended to scenarios requiring large-angle field-of-view transformation, such as industrial inspection and indoor/outdoor navigation.更多还原