Adaptive cooperative detection method for unmanned planetary vehicles based on deep reinforcement learning

Unmanned planetary vehicles have important significance for human development and utilization of space resources through the detection of extraterrestrial objects. This paper proposes an adaptive cooperative detection method for unmanned planetary vehicles, which can be adapted to the complex and changeable planetary environment through self-learning methods. First, the detection environment is rasterized; then, the convolutional neural network is used to process the detection area information and the lidar information to extract the environmental features; the reinforcement learning method is used to learn the strategy of detecting the environment. The reward function was designed in detail so that the planetary vehicles can quickly detect a given area without collision. At the same time, the detection information can be shared among planetary vehicles to accelerate the learning process. Finally, simulation experiments were carried out in Gazebo. The experimental results show that the planetary vehicles can detect the given area quickly and effectively.