针对仿人视觉系统目标跟踪过程中眼、颈的转动速度对跟踪精度的影响, 提出了一种基于雅可比矩阵的角度分解最优化方法。首先, 建立眼、颈2级四自由度空间坐标系, 搭建系统模型; 其次, 构建与眼、颈转动角度相关的雅可比矩阵, 综合考虑眼、颈转动角速度, 得到关注不同转动轴角速度变量的目标跟踪角度分解数学模型; 最后, 通过仿真和物理实验分析了各自由度转动角速度在最优化条件下对角度分解的影响, 得到了基于所述系统的目标跟踪角度分解最优化方案。实验结果表明: 在给定范围内, 眼、颈的转动角度分配比与眼、颈转动角速度的比值相同, 且与均分法相比较, 文中所述方法在时间效率上具有明显优势。

An optimized angle decomposition method based on Jacobi matrix was analyzed, considering the effect of rotational angular velocity for humanoid vision system. Firstly, the two-stages four-degrees′ coordinate was established, and the system model based on this coordinate was built. Secondly, considering the rotational angular velocity of eyes and neck, a target tracking mathematical model of angle decomposition, considering various rotation angular velocities, was built based on Jacobi matrix. Lastly, through experiments, the influence of rotation angular velocity in each degrees of freedom on rotation angle decomposition under the condition of optimization was studied. After that the optimized angle decomposition model based on the system mentioned above has been settled. By experimental data, within a given range, the ratio of decomposed eyes and neck rotation angle is equal to the ratio of rotational angular velocity of eyes and neck. And compared with split method, the method described in this article has remarkable advantages on time efficiency.