An Autonomous System for Multimodal Mapping of Sea Ice Fields From Ships

This article presents an instrumentation system that maps sea ice fields from ships using a light detection and ranging (LiDAR) sensor and an optical camera. The output is a multimodal point cloud containing 3-D coordinates, LiDAR intensity values, and RGB colors from the camera. The system is designed to be completely independent of the ship it is mounted on, making it a scalable solution for in situ sea ice analysis with minimal installation effort. To achieve this, a Global Navigation Satellite System (GNSS)-aided inertial navigation system is developed to estimate the system’s trajectory during transit. The navigation system uses a factor graph approach to perform full-trajectory smoothing and integrates a novel method to include motion constraints from the ship’s dynamics. Specifically, we model how orientation changes are reflected as position displacements caused by a lever arm between the ship’s center of rotation and the instrumentation system. The navigation system is evaluated against ship navigation data, and results indicate that the system pose estimates align well with the ship estimates. The resulting ice maps are validated by comparing rendered point cloud images with photographs captured by the system camera. Using this technique, the presented examples illustrate how the generated point clouds provide insights into properties, such as ice deformation, surface reflectivity, and color.