论文部分内容阅读
针对深空探测星际着陆过程中的自主导航问题,提出一种基于星表特征直线的着陆器位姿估计算法。该算法首先采用EDLine算法对着陆过程中所拍摄的图像进行特征直线提取;其次根据直线局部特征对特征直线进行匹配;之后利用至少3对已匹配的特征直线,建立关于着陆器位姿的几何约束方程;然后根据奇异值分解法得到着陆器位姿的候选解;最终通过最小二乘法从候选解中选取着陆器位置、姿态的唯一解。仿真结果表明:该算法可以快速估计着陆器位姿,且在高度为2 000 m时位置误差小于2 m、姿态误差小于0.10°。
In order to solve the problem of autonomous navigation during deep space exploration in the process of interplanetary landing, a pose estimation algorithm of lander based on the stellar feature line is proposed. In this algorithm, firstly, the feature extraction of the images captured by the landing process is performed by using the EDLine algorithm. Secondly, the feature lines are matched according to the local features of the lines. At least three pairs of matched feature lines are then used to establish geometric constraints on the pose of the lander Then, the candidate solution of the lander pose is obtained by using the singular value decomposition method. Finally, the least square method is used to select the unique solution of the lander position and attitude from the candidate solutions. Simulation results show that this algorithm can quickly estimate the position and attitude of the lander, and the position error is less than 2 m and the attitude error is less than 0.10 ° at a height of 2 000 m.