Future lunar surface exploration needs to be carried out in areas with more complex terrain and richer scientific research value without prior information. Therefore, the patrol device needs to be able to autonomously detect terrain and obstacle avoidance when planning exploration tasks. However, Deep reinforcement learning has been successfully applied in various game-like environments, and it is a challenging task to apply deep reinforcement learning to the navigation and obstacle avoidance detection of the rover on lunar surface. We propose an autonomous decision-making method for lunar vehicle patrols based on deep reinforcement learning. The algorithm first uses lidar for simultaneous positioning and graphing. For the constructed static three-dimensional point cloud map of the lunar surface, it is described as a two-dimensional grid using octree-map projection The grid map is processed by the long-short time memory (LSTM) to process the two-dimensional grid obstacle information. Using dueling deep Q network (dueling DQN) with multi-step learning to train the planetary vehicle automatic path planning, and finally in the ROS static obstacle environment simulation verification experiment, the simulation results verify that the proposed method is effective and adaptable for the lunar rover in different terrain environments.
The freeze–thaw effect has a significant impact on the strength deterioration of tunnel-lining concrete in cold regions. Therefore, the strength deterioration characteristics of concrete in a tunnel were studied, and silicone coating materials were used to improve its frost resistance and durability under freeze–thaw cycles. Freeze–thaw cycle tests were conducted on concrete specimens with different coatings. The freeze–thaw damage phenomenon, dynamic elastic modulus, and mass loss of the specimens were used to evaluate the freeze–thaw durability of concrete strengthened with coatings. The results demonstrated that silicone coatings effectively prevented moisture and corrosive substances from infiltrating the concrete, thereby enhancing its durability; the silicone–polyether hybrid had the most significant frost resistance at 500 g/m2 and silane type III at 300 g/m2, with freezing resistance times of 175 and 300, respectively. During the freeze–thaw process, the strength reduction rate of specimens was much greater than the mass loss rate of concrete. Taking into account the water environment surrounding the lining concrete and the site temperature, an equivalent indoor freeze–thaw cycle conversion model was established. The results can provide an experimental basis for selecting better frost-resistant materials for tunnel concrete in cold regions.
Planetary rover is the most efficient way to explore planets. The exploration area of planetary rover is a complex and unfamiliar environment. Planetary rover needs to perceive surrounding terrain and obstacle information clearly so as to ensure that the planetary rover can effectively and accurately perform autonomous obstacle avoidance and path planning in the unfamiliar environment. There is a problem that the slope is misjudged as an obstacle in autonomous obstacle avoidance, which leads to wrong planning decision. Meanwhile the passable area is treated as restricted area. Therefore, it is necessary to detect and estimate slope and identify the slope passing ability of planetary rover. A method for slope detection of planetary surface in the autonomous obstacle avoidance of planetary rover is presented in this paper. The down-sampling of original point cloud is processed by voxel mesh filtering. Meanwhile the real time 3D point cloud of large-scale planetary surface is clipped and segmented. The ground and obstacle are separated by local ground plane fitting estimation. Then slopes are calculated and classified, through which the slope above the threshold is reserved as obstacle and the slope below the threshold is set as passable area. The method presented in this paper is able to detect and classify the slope accurately in real time, which is shown in the simulation results.