TURTLMap: Real-time Localization and Dense Mapping of Low-texture Underwater Environments with a Low-cost Unmanned Underwater Vehicle

Abstract

Significant work has been done on advancing localization and mapping in underwater environments. Still, state-of-the-art methods are challenged by low-texture environments, which is common for underwater settings. This makes it difficult to use existing methods in diverse, real-world scenes. In this paper, we present TURTLMap, a novel solution that focuses on textureless underwater environments through a real-time localization and mapping method. We show that this method is low-cost, and capable of tracking the robot accurately, while constructing a dense map of a low-textured environment in real-time. We evaluate the proposed method using real-world data collected in an indoor water tank with a motion capture system and ground truth reference map. Qualitative and quantitative results validate the proposed system achieves accurate and robust localization and precise dense mapping, even when subject to wave conditions. The project page for TURTLMap is https://umfieldrobotics.github.io/TURTLMap.

Figures (8)

• Figure 1: Overview of TURTLMap. We propose a real-time localization and dense mapping solution for low-texture underwater environments with a low-cost underwater robot.
• Figure 2: An overview of TURTLMap for real-time localization and dense mapping. We develop the system using ROS ros. The localization stack is formulated as a pose graph optimization framework with DVL, IMU and barometer measurements. The estimated pose is published as a transform message using ROS tf tool. Voxblox oleynikova2017voxblox subscribes to the pose and the depth point cloud from the ZED mini camera to construct a dense map in real-time.
• Figure 3: A comparison of the stereo depth point cloud of a rock platform placed on the tank bottom with the default calibration (left) and the underwater calibration (right). The bottom surface of the tank is flat.
• Figure 4: An overview of the robot platform and the customized sensor enclosure. The Blue Robotics Bar30 barometer is located in the electronics enclosure of the BlueROV2. The Waterlinked DVL A50 is placed on the payload skid with its transducers facing downward. The ZED mini stereo camera is placed inside the customized sensor enclosure with an NVIDIA Jetson Orin Nano embedded computer. We also place a LORD MicroStrain 3DM-GX5-25 IMU inside the sensor enclosure. Additional sensors (e.g., Blue Robotics Ping360 scanning sonar, Blue Robotics Ping1D sonar) are mounted on the robot, but they are not used in this work.
• Figure 5: The towing tank at the University of Iowa. We show different regions for localization evaluation and mapping evaluation. The region where the MoCap system has valid tracking is relatively small. We exclude the artificial beach for map evaluation due to the lack of its reference structure.