Development of A Raindrop-aware Environment Detection Algorithm for 3D-LiDAR Based Outdoor Mobile Robot Navigation.

Stable perception of its environment is essential to navigating an autonomous mobile robot safely. Most autonomous mobile robots employ a LiDAR (light detection and ranging) to sense its surrounding. The LiDAR can stably and accurately identify the surrounding environment in a normal situation. However, it has a problem in the case of rain because it would detect raindrops as objects. When an autonomous mobile robot uses LiDAR to detect obstacles during rains, the laser emitted from LiDAR may hit raindrops, resulting in false-positive reading. In this paper, we employ a 3D-LiDAR and present a new method for automatic detection and tracking of raindrops. We consider the threedimensional raindrop behavior due to gravity to reconstruct a rain-free surrounding environment. We conducted experiments with actual and simulated raindrops and validated the effectiveness of the method.