A System Ensuring the Effectiveness of Lunar Rover in Highly Bumpy Environments

In the unstructured bumpy environment of the lunar surface, the perception algorithm of the lunar rover often leads to the distortion of sensor data due to the mutation of rover pose. This paper proposes a multi-modal fusion anti-turbulence framework, which solves the problem of sensor data offset and off-line calibration parameter failure by quantifying road turbulence degree, lidar motion compensation and multi-sensor spatio-temporal alignment technology. Based on the multi-source data of LiDAR, binocular camera, IMU and odometer, the system realized dynamic compensation and online calibration by complementary filtering and generalized Iterative Closest Point (GICP) algorithm. Experiments show that the point cloud distortion error is reduced by 36.2 % after motion compensation, and the spatio-temporal alignment improves the IMU-LiDAR trajectory coincidence by 89 %. The proposed system significantly improves the robustness of obstacle detection and slope estimation in bumpy environments.