Minglei Li, Zezhou Sun, Shaochuang Liu, Youqing Ma, Hao Ma, Changming Sun, and Yang Jia
Chang’e-3 system, lunar rover exploration, stereo vision mapping,visual odometry
In December 2013, China’s Chang’e-3 system successfully landed on the moon and started carrying out rover exploration. Stereo vision technologies have played important roles in terrain mapping and rover localization from beginning to end in this mission. This article mainly discusses two stereo vision implementations. First, using stereo images, we propose a novel solution for terrain reconstruction that combines both geometry and appearance information within a joint model. This model could provide accurate terrain maps for the rover navigation and virtual reality simulations of the mission. Then, we present a binocular visual odometry algorithm which leverages an image-network-based bundle adjustment to obtain rover’s position and orientation from cross-site images. In addition, both field experiments and onboard results are presented to verify the effectiveness of the proposed algorithms. These technologies have contributed substantially to the Chang’e-3 mission and help to lay a foundation for future autonomous missions.
Important Links:
Go Back
