Aerial-Terrestrial-Aquatic Robots for Search and Rescue in an ATA Extreme Environment

今年，我们完成了项目的主要部分，重点是ATA地图。为了估计空中-水上过渡过程中的点云，我们开发了一种新的多盒内插方法。我们提出的方法建立在研究空中和水上点云噪声特性的基础上。我们还开发了一种图像处理技术来预测是否应该融合点云。这使得机器人可以避免在过渡期间发生碰撞。我们使用热像仪绘制环境地图，通过开发基于机器学习的全景感知生成性对抗网络来帮助机器人在黑暗的水下区域进行热像到彩色图像的转换。

This year, we completed a major part of our project, focusing on ATA mapping. To estimate point clouds during aerial-aquatic transitions, we developed a new multi-box interpolation method. We based our proposed method on studying the noise characteristics of aerial and aquatic point clouds. We also developed an image processing technique to predict whether we should fuse point clouds or not. This allows the robot to avoid collisions during transitions. We mapped the environment using a thermal camera, which helps the robot to see in dark underwater areas by developed a machine learning-based panoptic perception-based generative adversarial network for thermal-to-color image translation.