沙质环境中基于视觉的月球车定位方法

视觉里程计是月球车上重要的传感器之一，能够实现月球车在松软月面的定位，并对滑转率进行估计．针对沙质的月面模拟环境，为月球车开发了适应能力强的视觉里程算法．首先，研究基于形态学的图像增强算法以保证提取到足够多的特征点；其次，在对极线约束下实现角点的匹配，并且通过优化搜索窗口尺寸和预估搜索方向等方法提高匹配和跟踪的效率；再其次，采用随机抽样一致（RANSAC）算法进行运动估计，计算巡视器在6个自由度上的姿态变化；最后，在月面模拟场地对视觉里程算法进行试验，证明所提算法在特征稀少的环境中仍能保持较高的稳定性和定位精度．

Visual-based location for lunar rover in sandy environment

Visual odometry is an important sensor for lunar rover, which can realize vehicle location and slip check in soft lunar surface. A visual odometry algorithm with high adaptability to sandy simulated lunar surface is developed. Firstly, morphology-based image enhancement method is studied to guarantee enough corners been extracted. Secondly, corner matching is implemented under epipolar restriction and the efficiency of corner matching and tracking is enhanced by optimizing the search window size and estimated search direction. Thirdly, random sample consensus (RANSAC) algorithm is proposed to compute and update rover pose to the 6 degree of freedom. Finally, experiments under a simulated lunar surface have confirmed the stability and positioning accuracy of the proposed method, even while driving in an environment of scarce features.