Development of a measurement system for gas-autonomous surface vehicle to map marine obstacles using stereo depth and LiDAR cameras

In this work, we developed a measurement system for the Gas-Autonomous Surface Vehicle (G-ASV), a successor to the micro-Autonomous Surface Vehicle (μ-ASV), to facilitate conducting ocean research. We included a Real-Time Kinematic Global Positioning System (RTK GPS) and compared the results with a regular GPS. We also incorporated stereo depth and Light Detection and Ranging (LiDAR) Cameras with Artificial Intelligence (AI) to perform object detection and mapping. The purpose was to improve the position accuracy of the ASV, conduct surveillance, and map objects such as ships within its surrounding environment while maintaining low traffic on the G-ASV’s LTE network. We proposed two methods to reduce the size of the data produced by the depth and LiDAR cameras. The first method compresses the depth image generated by the depth camera into an RGB image transmitted over the LTE network. The second method consists of converting the depth image generated by the LiDAR camera into eight equally spaced rays along the horizontal field of view of the LiDAR camera. We then mounted the system onto the G-ASV and conducted a field experiment at Tokyo Bay Marina. We significantly increased position accuracy using RTK-GPS compared to a regular GPS. Furthermore, the depth compression and recovery process results are within the margin that is considered acceptable. Therefore, we performed obstacle detection with the G-ASV using RTK, Depth Camera, and LiDAR Camera in this study and successfully mapped the obstacles on a map.