如何用MediaPipe姿态估计3D坐标在Unity中实时驱动人形模型

• 现有基础：已实现Python程序运行MediaPipe姿态估计，通过UDP将3D姿态坐标数据传输至Unity，目前已能用球体和线条渲染出姿态结果。
• Unity坐标系下的人体关键点样本坐标：

[[-0.97, 11.58, -4.34], [-1.0, 11.87, -4.14], [-1.0, 11.88, -4.13], [-1.0, 11.88, -4.14], [-1.41, 11.81, -4.15], [-1.41, 11.81, -4.16], [-1.41, 11.82, -4.15], [-0.46, 11.92, -2.64], [-2.34, 11.65, -2.86], [-0.66, 11.46, -3.8], [-1.22, 11.37, -3.85], [1.19, 10.72, -1.23], [-3.12, 10.3, -2.47], [2.04, 9.17, -0.63], [-2.98, 8.71, -3.4], [3.43, 7.62, -1.72], [-0.01, 8.59, -5.06], [3.82, 7.15, -1.93], [1.07, 8.44, -5.46], [3.7, 7.07, -2.35], [1.22, 8.79, -5.46], [3.41, 7.48, -1.86], [0.33, 8.7, -5.08], [1.41, 7.27, 0.47], [-1.42, 7.13, -0.45], [0.99, 4.35, 2.16], [-1.27, 4.28, 0.96], [0.78, 1.9, 4.41], [-1.18, 1.94, 3.2], [0.89, 1.59, 4.51], [-1.15, 1.63, 3.3], [2.13, 1.21, 3.14], [-1.04, 1.1, 1.6]]

• 关键点对应关系：采用MediaPipe标准人体姿态关键点索引定义。
• 目标需求：利用上述实时传输的3D姿态坐标，驱动Unity中的人形角色模型完成动作同步。

内容的提问来源于stack exchange，提问作者073